AWS re:Invent 2020 - Keynote with Andy Jassy

Aug 16, 2023

AWS re:Invent 2020 - Keynote with Andy Jassy

Join Us Now at, AWS re:Invent, our Free 3-Week Virtual Conference - https://bit.ly/346Ut2q

Andy Jassy, CEO of Amazon Web Services, delivers his AWS re:Invent 2020 keynote, featuring the latest news and announcements, including the launches of Habana Gaudi-based Amazon EC2 instances, AWS Trainium, Amazon ECS Anywhere, Amazon EKS Anywhere, Lambda Container Support, AWS Proton, gp3 volumes for Amazon EBS, io2 Block Express for Amazon EBS, Amazon Aurora Serverless v2, Babelfish for Amazon Aurora PostgreSQL, AWS Glue Elastic Views, Amazon SageMaker Data Wrangler, Amazon SageMaker Feature Store, Amazon SageMaker Pipelines, Amazon DevOps Guru, Amazon QuickSight Q, Amazon Connect Wisdom, Amazon Connect Customer Profiles, Real-Time Contact Lens for Amazon Connect, Amazon Connect Tasks, Amazon Connect Voice ID, Amazon Monitron, Amazon Lookout for Equipment, AWS Panorama Appliance, AWS Panorama SDK, AWS Outposts in two new smaller sizes, and 12 additional AWS Local Zones. Guest speakers include Lori Beer, of JPMorgan Chase, Blake Scholl, of Boom, and David Gitlin, of Carrier.

Launch Announcements:
0:00 Andy Jassy Keynote
38:49 Habana Gaudi-based Amazon EC2 instances
39:37 AWS Trainium
43:53 Amazon ECS Anywhere
44:40 Amazon EKS Anywhere
48:59 Lambda Container Support
51:14 AWS Proton
1:00:10 gp3 volumes for EBS
1:03:01 io2 Block Express for EBS
1:08:54 Amazon Aurora Serverless v2
1:11:43 Babelfish for Amazon Aurora PostgreSQL
1:19:01 AWS Glue Elastic Views
1:40:43 Amazon SageMaker Data Wrangler
1:43:51 Amazon SageMaker Feature Store
1:45:34 Amazon SageMaker Pipelines
1:51:55 Amazon DevOps Guru
1:57:12 Amazon QuickSight Q
2:05:16 Amazon Connect Wisdom
2:06:51 Amazon Connect Customer Profiles
2:10:02 Real-Time Contact Lens for Amazon Connect
2:12:50 Amazon Connect Tasks
2:14:00 Amazon Connect Voice ID
2:28:52 Amazon Monitron
2:30:11 Amazon Lookout for Equipment
2:31:49 AWS Panorama Appliance
2:33:30 AWS Panorama SDK
2:42:53 AWS Outposts in two new smaller sizes
2:44:38 12 additional AWS Local Zones

Guest Speakers:
20:40 Lori Beer, of JPMorgan Chase
29:24 Jerry Hunter, of Snap Inc
55:06 Don MacAskill, of SmugMug
1:22:05 Blake Scholl, of Boom
1:31:34 Marianna Tessel, of Intuit
2:00:04 Paul Cheesbrough, of Fox Corporation
2:20:04 David Gitlin, of Carrier
2:35:03 Zach Blitz, of Riot Games

Subscribe:
More AWS videos http://bit.ly/2O3zS75
More AWS events videos http://bit.ly/316g9t4

#AWS

Content

0.483 -> [music playing]

58.273 -> Please welcome the CEO of Amazon Web Services, Andy Jassy.

62.944 -> [applause]

66.91 -> Thank you, and welcome to the ninth annual AWS

70.11 -> re:Invent and the first virtual one.

72.36 -> Now, I think we all wish we were together in Las Vegas right now

76.13 -> but we’ve got the next best thing, we’re live,

78.24 -> so I’m really excited to be with you.

79.76 -> I am going to share a few words about re:Invent,

81.56 -> it’s a little bit different this year.

83.44 -> So one thing that’s similar is we have a lot of people congregated.

87.01 -> We have well over 400,000 people

90.12 -> who have already registered for re:Invent.

91.72 -> We will be at 500,000 by the end of the week,

94.41 -> and this year is going to be different than prior years.

97.36 -> re:Invent is going to be over three weeks instead of one week,

100.69 -> and then we have five keynotes, so, the one you have today.

103.83 -> Thursday there will be a partner keynote.

105.93 -> Next Tuesday will be our first ever machine learning keynote with Swami.

109.96 -> Then, next Thursday, you will have Peter DeSantis

112.11 -> telling you more about our infrastructure and then,

114.57 -> Werner will do the anchor keynote the third week.

117.25 -> So we’ve got a lot in store for you.

122.81 -> re:Invent is first and foremost an education

125.37 -> and learning conference and this year will be no different.

128.09 -> We have well over 500 technical sessions

130.86 -> and while I think we would all like to be together,

133.25 -> I think one nice part about it being virtual

135.42 -> is that you don’t have that session collision that you sometimes have,

138.68 -> where there are two sessions you want to go to at the same time

140.95 -> or you have to travel. They will all be online.

143.58 -> Some will be scheduled. Most of them will be on demand.

146.68 -> So it will be much easier for you to see everything you want.

149.14 -> A lot for you.

151.33 -> So I am going to give a quick update.

153.18 -> But before I do I just want to acknowledge

155.74 -> what an unbelievably crazy last year

158.7 -> this has been, particularly these last nine months.

160.65 -> COVID has been so difficult for so many people.

163.6 -> We have had so many people pass away.

165.96 -> We’ve had so many people lose their jobs.

168.08 -> So many businesses are struggling. It’s just been really difficult

171.49 -> and I think if you live in the United States

173.94 -> it’s also pretty hard to not be struck

177.16 -> by the murders of Ahmaud Arbery, Breonna Taylor and George Floyd

181.55 -> and realize the sobering thought that we have so far

184.93 -> to go in this country in how we treat black people.

187.7 -> I think the reality is for the last several hundred years

190.94 -> the way we treated black people in this country is disgraceful

194.04 -> and something that has to change.

196.3 -> We’re working on it at Amazon. I know a lot of companies are

198.84 -> but I think the important thing for us all to realize

201.72 -> is that this does not get solved in a few months.

203.83 -> It can’t be something that we work on for a few months

206.93 -> and then turn our attention elsewhere.

208.25 -> It’s going to take several years of us working together

211.01 -> but we need to do it.

214.58 -> So a quick AWS update.

216.96 -> The business is at a $46 billion revenue run rate business

220.36 -> in the last announcement that we made growing 29% year-over-year.

224.68 -> I think it’s also important to talk about that year-over-year right

227.44 -> because I think a lot of people sometimes get confused

229.83 -> and they try to compare different companies’ year-over-year growth rate.

233.53 -> The year-over-year growth rate percentage

235.64 -> only matters as it relates to the base revenue.

238.58 -> So you can have a higher year-over-year growth rate

241.77 -> but be growing at a much lower absolute rate

243.93 -> if you have a much lower base of revenue.

246.37 -> If you look at just AWS as an example to grow to a $46 billion dollar

251.42 -> revenue run rate with 29% year-over-year growth,

254.11 -> it meant we had to grow an incremental $10 billion

257.65 -> in the last 12 months to get there.

259.31 -> That is much larger than you will see elsewhere in the cloud.

262.52 -> I think another data point to give you an idea of how fast AWS is growing,

266.972 -> it took us a 123 months, a little over 10 years

270.97 -> to grow to a $10 billion business.

273.48 -> Then it took us only 13 months to go from $10 to $20 – er, 23 months -

277.81 -> to go from $10 to $20 billion, 13 months to go from $20 to $30 billion

282.4 -> and then 12 months to go from $30 to $40 billion.

285.43 -> So the rate of growth in AWS continues to accelerate.

290.25 -> I thought this was also an interesting slide

292.73 -> which is: This shows you the top enterprise

295.92 -> IT companies out there ten years ago based on revenue.

300.65 -> And you can see on this list

301.9 -> there’s some companies that don’t really exist anymore in this forum

304.54 -> but you will also see

306.17 -> that AWS is nowhere to be seen in this list.

308.81 -> And then if you fast forward just ten years in 2020

313.33 -> you can see that AWS is now the fifth largest enterprise

317.35 -> IT company in the world.

319.02 -> Ahead of companies like SAP and Oracle and of course,

322.51 -> that growth is significantly driven by the growth of cloud computing

326.853 -> and the infrastructure technology space.

329.2 -> And I think that you can see that

330.9 -> with what’s happened during the pandemic.

333.38 -> And in the short term in the first nine months

337.47 -> or ten months of this thing virtually every company in the world,

340.98 -> including Amazon, has tried to save money any way that they can.

345.38 -> But what we have seen and this happens a lot of times

348.72 -> when you have a period of discontinuity like a pandemic

351.64 -> is that companies take a step back

353.78 -> and they rethink what they are doing and what they want to stop doing.

356.7 -> And one of the things that we’ve seen is that enterprises

359.8 -> that we’ve been talking with for many years

362.11 -> about moving to the cloud where there’s a lot of discussion

364.52 -> and dipping the toes in the water, but not real movement,

368.21 -> so many of those enterprises have gone

370.3 -> from talking to having a real plan.

372.84 -> And that I think is going to be one of the biggest changes you'll see.

375.524 -> See, I think when you look back on the history of the cloud

377.96 -> it will turn out

379.08 -> that the pandemic accelerated cloud adoption by several years.

385.93 -> It's really hard to build a business that sustains

389.82 -> for a long period of time - really, really hard.

393.14 -> And this is a metric that I think is interesting.

395.13 -> If you look at the Fortune 500 just 50 years ago in 1970

400.35 -> you can see that only 83 companies or 17% of them

404.02 -> are still in the Fortune 500.

405.7 -> If you look at just 20 years ago the 2000 Fortune 500,

409.91 -> only half of them are still in that list.

413.1 -> It is really hard to build a business

416.22 -> that lasts successfully for many years

418.47 -> and to do it you’re going to have to reinvent yourself

421.32 -> and often you are going to have to reinvent yourself multiple times over.

425.22 -> And so, in the last nine months I’ve thought a lot about reinvention

430.31 -> and what it takes to do reinvention well.

432.71 -> And typically what you see is the desperate kind of reinvention.

437.96 -> You see companies that are on the verge of falling apart

440.57 -> or going bankrupt deciding they have to reinvent themselves.

444.16 -> And when you wait to that point it’s a crapshoot

447.39 -> whether you’re going to be successful or not.

449.26 -> It’s a little bit like borrowing money.

450.89 -> Everyone will tell you that you don’t want to have to be borrowing money

453.55 -> when the business is in bad shape

454.99 -> because you may not get the rates you want

456.93 -> or you may not get money at all.

459.1 -> You want to be reinventing when you’re healthy.

461.63 -> You want to be reinventing all the time.

464.42 -> And so we thought about what are the keys to reinvention

468.02 -> and some of it is building the right reinvention culture

471.43 -> and some of it is knowing what technology is available to you

474.23 -> and jumping on it to make that reinvention happen.

476.9 -> So I thought I would share with you today

478.48 -> what we see as some of the keys to building that reinvention culture

482.08 -> and then some of the things that we see being reinvented as we speak.

487.56 -> So what does it take to reinvent?

490.15 -> And I am going to list eight keys that I think are important

493.24 -> if you want to build the right reinvention culture.

496.07 -> And the first is that you have to have the leadership will

500.111 -> to invent and reinvent, and those terms sometimes

503.72 -> they are a little different, they are also similar.

506.01 -> If you think about it, people often say invention

508.97 -> is inventing a new product or service from whole cloth

511.907 -> and reinvention is reimagining an existing concept.

515.07 -> But if you’re going to reinvent and reimagine

517.41 -> there’s a load of invention in there.

519.89 -> Just look at what Airbnb has done in the hospitality space

523.54 -> or what Peloton has done in the exercise bicycle space

526.85 -> or look at what Stripe has done in the payment space.

529.5 -> These are huge amounts of invention

532.28 -> that has gone into reimagining these spaces.

535.17 -> And so if you’re going to be a leader that’s going to reinvent

539.27 -> you have got to be maniacal and relentless and tenacious

544.68 -> about getting to the truth.

546.08 -> You have to know what competitors are doing in your space.

549.88 -> You have to know what your customers think about your product

553.16 -> and where you sit relatively speaking.

555.17 -> You have to know what’s working and what’s not working.

557.94 -> And you will always have a lot of people inside the company

561.49 -> who will try and obfuscate that data from you.

563.93 -> Sometimes they think they are doing you a favor

566.21 -> and sometimes it’s for self-preservation reasons

568.66 -> but it’s hard to get at that data and you have to be relentless about it.

574 -> You have to challenge people.

575.16 -> Often people who know a lot more about a subject than you do

578.2 -> but you have got to get to the truth.

579.96 -> And then when you realize that there’s something you have

582.62 -> to reinvent and change you have to have the courage

585.68 -> to pick the company up and force them to change and move.

589.28 -> And part of that is sometimes acknowledging

591.71 -> that you can’t fight gravity.

593.81 -> If you step back and have conviction that something is going to change

597.74 -> because it’s a better experience for customers, it is going to change.

601.7 -> Whether you want it to or not,

603.33 -> whether it’s convenient for you or not, it is going to change

606.33 -> and there are a lot of examples of this.

607.9 -> I think if you look at what Reed Hastings and Netflix

611.382 -> did several years ago

612.54 -> where they cannibalized their own DVD rental business

614.95 -> because they saw where it was headed with streaming,

617.17 -> I think that turned out to be a pretty good decision for them.

620.38 -> I think if you look at Amazon in the late 90s

624.61 -> we had this owned inventory retail business

627.12 -> which meant we bought all this product from publishers

631.78 -> and from distributors. We stored them in our warehouses

635 -> and then we shipped them to customers.

636.61 -> And what we started seeing was these companies like eBay

639.83 -> and Half.com that were actually offering third-party sellers’ products

644.3 -> and they were shipping the products to customers.

646.09 -> And we had this huge animated debate inside the company

649.42 -> on whether or not we should support that.

651.61 -> And the reasons that we were concerned about it

653.79 -> were we just didn’t believe

655.07 -> anybody was going to take care of customers the same way that we did.

658.35 -> And then also the whole culture was set up

660.775 -> to be this owned inventory business.

662.42 -> People worried, well if we worked with third-party sellers,

665.13 -> how would our publishers or distributors feel?

667.41 -> So it was a very hard decision but ultimately, we decided

670.9 -> to build a marketplace and offer third-party sellers.

673.74 -> And we did it because we know that you cannot fight gravity.

678.24 -> It was better for customers. It provided them better selection

681.64 -> and it gave them more assortment on price.

684.15 -> Now that also turned out to be a good decision for us

686.58 -> because we sell more than a half of our retail products

688.66 -> through third-party sellers.

690.48 -> But you have got to realize that if something’s going to happen

693.49 -> it is going to happen regardless of whether you want it to or not.

695.91 -> You’re much better off cannibalizing yourself

699.49 -> than having someone do it to you and chasing it.

702.75 -> The third thing you have got to make sure of

704.4 -> is that you have talent that’s hungry to invent.

706.93 -> Now this seems fairly obvious.

708.3 -> Everybody says I have talent that wants to invent.

710.63 -> But it’s not always true.

712.1 -> A lot of people who have been at the company

713.92 -> for a long time are very comfortable doing things

717.01 -> the way they have been doing them for a long time.

719.22 -> Have you ever noticed it’s often when you have new blood in the company

722.49 -> that they are leading the transformation?

724.51 -> And that’s not because existing people can’t lead reinvention.

728.2 -> It's just that you’re asking them to reinvent something they built.

731.99 -> It’s hard to rip up something you spent a lot of time

734.28 -> and energy and dedication doing.

736.15 -> And it means you have got to learn new skills

738.13 -> and it means that you have got to actually be curious

740.8 -> about getting trained on other technologies.

743.24 -> Sometimes that’s true sometimes that not true…

745.99 -> I will tell you a quick story.

748.02 -> There’s a CIO in a pharma life sciences company

751.63 -> that I’ve known a long time.

752.71 -> I have a huge amount of respect for him.

754.9 -> And I went to see him a few years ago

757.78 -> and I was talking to him for about 30 minutes

760.17 -> about why I thought they should be using the cloud more meaningfully.

762.6 -> They were barely using the cloud at that point.

764.96 -> And he listened to me and when I was done he said,

767.23 -> “Look, I agree with everything you just said, Andy,

769.95 -> I agree that we could be inventing at a much faster clip

773.36 -> but that will be the job of the next CIO.

775.38 -> It will not happen on my watch.”

777.21 -> And that’s what happened. It took him a year or two.

780.54 -> He retired, they hired a new CIO, that CIO said: “What are we doing?”

784.83 -> And they’ve significantly moved to AWS in the cloud.

787.54 -> But they lost two, three, four years of inventing

792.24 -> on behalf of their customers

793.64 -> and you have to make sure that you have got builders

796.618 -> who are curious about learning

798.21 -> who are excited about leaning forward and inventing

800.78 -> and reinventing their customer experience.

802.54 -> Now, you want builders and talent that’s hungry to invent

807.2 -> but you want to make sure you guard against the opposite

809.92 -> which is that you have people who actually solve problems.

813.98 -> That you want people to solve problems for customers

816.95 -> as opposed to solving problems because they like the technology

820.42 -> and they think it’s cool. And you see this a fair bit.

823.43 -> You know if you look in the enterprise technology space

827.06 -> there are some providers who are competitor focused.

831.41 -> They look at what their competitors are doing

833.21 -> and they try to fast follow and one up them.

835.74 -> We have a competitor like that across the lake from us here in Washington.

839.45 -> Then you have a number of other providers who are product focused

844.16 -> and they say look, it’s great that you have an idea

846.66 -> on a product Mr and Mrs Customer but leave that to the experts.

850.26 -> And that’s the group that you have got to be careful about

853.47 -> because they often are building things that they think are cool

857.48 -> as opposed to what really solves the problems for customers.

860.55 -> At AWS, we’re customer focused.

863.4 -> What we build is driven by what you tell us matters to you.

866.77 -> And even if you can’t articulate a feature we’ll try to read

869.84 -> between the lines, understand what you’re trying to build,

872.29 -> and invent on your behalf.

874.51 -> And so if you think about over the years of AWS,

878.03 -> we have built a lot of technology that we believe is pretty cool,

881.28 -> you know pretty ground-breaking stuff, S3 and EC2

884.66 -> and RDS and Aurora and SageMaker and Redshift.

887.76 -> I mean a whole host of technology but we never built it

891.52 -> because we thought it was cool.

893.61 -> We built it because we knew it would enable you

896.41 -> to build new experiences and change your business.

898.59 -> You’ve got to make sure that your scarce resource of engineers

902.5 -> are working on problems that really matter to your customers.

906.84 -> The fifth thing is speed.

909.43 -> Speed disproportionately matters at every stage of your business

914.17 -> and in every sized company.

916.23 -> And I think that a number of leaders at enterprises

920.57 -> have resigned themselves that they have to move slowly.

923.52 -> It’s just the nature of how big they are.

925.68 -> It’s the nature of their culture.

927.86 -> They have engineering teams that tell them

929.76 -> “Hey look, this is too risky. This is too big a lift.”

933.5 -> Sometimes you control those teams into trying something

936.09 -> and the first sign of a problem

937.18 -> they throw up their arms and say “See.”

939.94 -> Speed is not preordained. Speed is a choice.

945.65 -> You can make this choice and you’ve got to set up a culture

950.95 -> that has urgency and that actually wants to experiment

954.84 -> because you can’t flip a switch and suddenly get speed.

958.17 -> It doesn’t work like that. You’ve got to build muscle to get speed.

961.15 -> You’ve got to be doing it all the time.

963.6 -> And there is going to be time…

964.97 -> Frankly I think that time is happening right now.

967.23 -> It happens a lot more frequently than most companies realize.

969.89 -> But there is going to be seminal moments

971.82 -> where if you don’t have the ability

973.13 -> to have speed you will not be able to reinvent when you need to.

976.77 -> Now one of the enemies of speed is complexity

981.15 -> and you have to make sure

982.95 -> that you don’t over complexify what you’re doing.

986.16 -> When companies decide to make transformations

988.56 -> and big shifts a huge plethora of companies descend on them

993.37 -> and providers descend on them

995.49 -> and tell them all the ways that you have got to use their products:

998.42 -> “You need to use us for this even if you’re using these people

1001.06 -> for these three things, use for these two.”

1003.01 -> This company says: “Use us for this.”

1004.86 -> They don’t deal with the complexity that you have to deal with

1008.18 -> in managing all those different technologies and capabilities.

1012.57 -> The reality is for companies that are making big transformations

1016.98 -> and shifts, it is much easier to be successful

1020.62 -> if you predominantly choose a partner

1022.85 -> and you learn how to do it and you get momentum

1025.85 -> and you get success and you get real results for the company.

1029.41 -> Then later on if you want to layer on complexity

1032.1 -> and more providers, you should go for it.

1034.3 -> But it’s not a great way to start to a reinvention

1036.89 -> to have too much complexity upfront.

1039.58 -> And then one of the ways to help you avoid some of that complexity

1043.08 -> is making sure that you use the platform

1046 -> that has the most capabilities and the broadest set of tools.

1049.96 -> Now I have played a round of golf

1051.76 -> with somebody who used a 5-iron for every shot in the round.

1055.88 -> It’s doable. It was not pretty and it was not very effective either.

1061.41 -> I don’t recommend it.

1062.64 -> And when you think about the cloud, because all the services you only pay

1068.1 -> for as you consume them - you don’t pay for it upfront -

1070.71 -> Why would you possibly go with a platform

1073.23 -> that has a fraction of the functionality of a leader?

1076.2 -> If you go with the platform that has the most capabilities

1079.2 -> and gives you the right tools for the job you need to do,

1082.29 -> it not only makes it easier for you

1084.48 -> to migrate all your existing applications,

1086.48 -> but also to enable your builders to build anything they can imagine.

1090.47 -> And there’s nobody who is close to the capabilities

1093.12 -> across the cloud infrastructure technology platform as AWS.

1096.89 -> And you can see that whether you are talking about compute

1101.07 -> or storage or database or analytics or machine learning

1105.56 -> or the edge or IoT or robotics, in every one of these categories

1110.74 -> you get a lot more functionality in AWS than anywhere else.

1115.72 -> The eighth key is something that really wraps all of this together,

1119.86 -> which is that the leadership team has to build aggressive top-down goals

1125.79 -> that force the organization to move faster

1128.01 -> than it organically otherwise would.

1129.99 -> And there are lots of examples of this.

1131.47 -> I would like to talk about GE, where about ten years ago their CIO

1135.62 -> decided that they were going to move 50 applications to AWS in 30 days.

1139.98 -> And her entire team told her what a terrible idea this was

1142.99 -> and she listened to them and she said,

1144.27 -> “We’re doing it anyway.” They got to about 42 in 30 days.

1148.8 -> But in the process they figured out their security and governance model.

1152.58 -> They figured out how to operate in the cloud

1154.18 -> and they had success which built momentum

1156.98 -> and the ideas came flowing in such that she could set

1160.09 -> that second big top-down goal to move 9000 applications to AWS

1165.17 -> in a few years.

1166.137 -> Capital One did the same thing

1167.977 -> where they set this big audacious goal top-down

1170.987 -> goal: They were going to reinvent their consumer

1172.727 -> digital banking platform on AWS. And then that was on their way

1177.027 -> to moving everything to the cloud in AWS.

1179.467 -> Setting an aggressive top-down goal forces the organization to understand

1184.567 -> that they are not going to be able

1185.877 -> to dip their toe in the water for a number of years.

1188.207 -> That you mean business and you’re going to make this change

1191.007 -> and setting up the right mechanisms to inspect

1193.747 -> whether you are getting the right progress,

1195.707 -> and if not, getting the issues on the table

1197.867 -> so you can solve them is really important.

1200.287 -> Now, you will notice that most of these keys, all of these keys really,

1204.867 -> except for maybe one, I mentioned are not technical.

1209.317 -> They are really about leadership.

1211.487 -> And so you’ve got to make sure that you embrace these types of keys

1216.337 -> to build a reinvention culture like you can.

1218.677 -> It’s very doable but you have to embrace them.

1222.067 -> My first speaker today has embraced these keys

1225.627 -> and it’s really remarkable to see how she has led her company

1229.487 -> to start reinventing on top of AWS.

1231.667 -> It’s my privilege to welcome to the stage the CIO of JP Morgan Chase,

1235.797 -> Lori Beer.

1237.79 -> [applause]

1241.937 -> Thanks Andy, it’s great to be here. JPMorgan Chase serves customers

1246.757 -> and clients from around the world from individuals

1250.257 -> and their local communities to corporations and governments.

1254.307 -> We have an over 200-year history built on trust.

1258.247 -> A foundation that allows us to build simple and intuitive experiences.

1264.217 -> Those solutions which are increasingly technologically driven

1267.847 -> represent the complete spectrum of financial services.

1272.177 -> From lending to banking, markets to advisory,

1275.697 -> and everything in between.

1278.027 -> And everything we do we do at tremendous scale.

1282.537 -> We have $28 trillion in assets under custody

1287.087 -> and process $6 trillion of payments

1289.677 -> daily through our Consumer and Community Bank.

1292.967 -> We have a relationship with 50% of US households

1297.117 -> and we serve 54 million active digital customers.

1302.227 -> We have long realized that while our relationships and financial expertise

1306.597 -> are paramount it is technology that continues to help us differentiate.

1312.767 -> With a 200-year history we’ve been dealing with technological change

1316.747 -> since the time of Thomas Edison literally.

1320.347 -> The firm financed Edison’s efforts to invent the first light bulb

1324.367 -> and our former Wall Street Headquarters

1326.797 -> was the first office in Manhattan to draw on Edison Electricity.

1331.387 -> We were the first bank to offer ATMs

1334.127 -> and our industry embraced early enterprise

1336.847 -> computing to process transactions.

1340.037 -> We were quick to offer online banking services

1342.717 -> during the rise of the internet and our industry

1345.577 -> is increasingly driven by mobile computing today.

1349.777 -> Simply put, we’ve been reinventing

1352.267 -> the financial services industry for decades.

1355.647 -> But it’s different today. The pace of change has accelerated.

1360.247 -> Innovation is more rapid and cloud platforms

1363.707 -> are disrupting business models daily.

1367.357 -> So we decided to completely rethink

1369.677 -> our environment to embrace a true modernization effort

1374.587 -> across 250 thousand employees, 35,000 developers,

1380.237 -> 6,000 applications and 450 petabytes of data.

1386.427 -> We asked ourselves how do we leverage our history of innovation

1389.997 -> and institutional know-how in order to evolve our business

1394.317 -> for today’s technology revolution?

1397.127 -> We set the direction at the top of the house

1399.397 -> with business and technology leadership

1401.917 -> agreeing on a key vision. We will have the best tech talent.

1407.107 -> We will own our destiny in a hybrid cloud world.

1410.177 -> And we will work with leading technology companies like AWS

1414.407 -> to deliver unique innovation.

1417.467 -> We’re in the midst of a truly one-of-a-kind transformation.

1422.007 -> Leveraging AWS and modern engineering practices

1426.247 -> like refactoring our applications to be cloud native,

1430.847 -> leveraging more advanced AI and analytics

1434.077 -> than we ever have before, and doing all of this securely.

1439.647 -> AWS is helping us along this journey.

1442.737 -> We’ve benefitted from their breadth of services and capabilities.

1447.377 -> Understanding of the enterprise and willingness to innovate

1451.847 -> and collaborate on key strategic initiatives.

1456.907 -> We work together on a unique approach

1459.837 -> holding actual hackathons with AWS and JPMorgan

1463.647 -> Chase engineers to decompose applications and migrate workloads.

1469.797 -> This process helped us uncover problems,

1473.267 -> gain institutional knowledge and enhance our collaboration.

1477.977 -> We’ve developed repeatable blueprints

1481.067 -> helping developers architect modern applications

1484.347 -> in safe repeatable environments

1487.307 -> using the depth of AWS services.

1491.607 -> We are a systemically important institution

1494.987 -> serving clients through some of the most turbulent events in history.

1500.447 -> Working with AWS allows us to scale massive volumes like Amazon EMR

1506.477 -> for trading analytics or AWS Lambda and Amazon Elastic Kubernetes

1512.207 -> Service for risk calculations so we can innovate

1515.967 -> to stay ahead of our competitors.

1519.197 -> Transforming yourself takes a lot of work

1522.237 -> but we are creating paved ways for developers

1524.967 -> and accelerating adoption.

1527.047 -> Our developer community is now innovating at scale using AWS

1533.097 -> and we continue to migrate critical workloads

1536.147 -> that can take advantage of the unique capabilities of the platform.

1541.257 -> How can we transform our business through AI and better analytics?

1545.547 -> As I mentioned earlier we have customer relationships

1548.537 -> with half of all US households.

1551.567 -> The scale of our reach provides an opportunity to use analytics

1556.287 -> to better service them.

1558.657 -> To do that we need a modern AI platform

1561.817 -> that was secure and one that enabled rapid experimentation.

1567.287 -> That’s why we built our firm-wide AI platform OmniAI on AWS

1573.277 -> using Amazon SageMaker for machine learning.

1577.247 -> SageMaker has helped us create a platform

1579.507 -> to rapidly test and train machine learning algorithms

1583.177 -> enabling us to run more experiments with more data.

1587.837 -> These are real use cases providing real value.

1592.437 -> As I speak, our data scientists are leveraging our OmniAI platform

1597.027 -> to develop new capabilities.

1599.797 -> For example, incorporating natural language

1602.597 -> processing to make client interactions more personalized.

1607.127 -> Testing advanced machine learning models

1609.357 -> to have a more comprehensive view of risk.

1612.397 -> And performing real-time coaching

1614.407 -> and recommendations for call center agents

1617.427 -> so they can better serve customers.

1620.807 -> We use SageMaker across the model development lifecycle

1624.887 -> from data labeling to model selection

1628.027 -> to experimentation and model serving.

1631.937 -> Our success with model development on AWS

1635.057 -> has influenced our data management strategy.

1639.027 -> We’re now investing in cloud data

1640.767 -> warehousing technology with Amazon Redshift

1643.837 -> to more effectively scale our analytic capabilities

1646.677 -> in a modern environment.

1649.477 -> The combination of a scalable AI platform

1652.507 -> and AWS’s Elastic Compute environments will help us

1655.987 -> accelerate our efforts to infuse analytics in everything we do.

1662.317 -> I will close with what I started with.

1664.767 -> We are a business of trust.

1667.567 -> Everything we do is for the benefit of our customers and clients

1671.847 -> and we deliver new innovations in a safe and secure manner.

1676.757 -> This is why we devote significant resources

1679.537 -> and collaborate with AWS to protect and continuously

1683.387 -> improve the security of our systems.

1687.097 -> Our adoption of cloud technology is already paying off.

1690.597 -> We’re more agile.

1692.067 -> We’re more secure and we’re more efficient.

1695.467 -> And through this journey we will work with AWS

1698.787 -> to reinvent ourselves and build financial services of the future.

1704.327 -> Thank you.

1706.278 -> [applause]

1711.947 -> Thank you, Lori.

1713.297 -> It is really an honor for us to work with JPMorgan Chase.

1716.707 -> The partnership has come a long way

1719.447 -> and we’ve just gotten started with respect to what we can do together.

1722.237 -> So thank you very much.

1724.137 -> So I mentioned earlier that when we think about the key

1728.397 -> to reinventing it’s a combination of building

1731.107 -> the right reinvention culture

1733.267 -> and then also having a set of technologies

1736.677 -> that you know are available to you that you jump on to use to reinvent.

1740.397 -> So I thought I would spend some time

1742.417 -> talking with you now about some of the areas

1745.007 -> that we believe are actively being reinvented

1747.837 -> that will allow you to do this reinvention.

1750.057 -> And at the beginning of each of these sections

1752.007 -> I am going to ask the customer to say a few words

1754.047 -> about what they think about this particular area

1757.147 -> and we’re going to start with the SVP of Engineering

1759.387 -> at Snap, Jerry Hunter, who will talk about Compute.

1762.647 -> [music]

1770.457 -> All right we’re going to try and shoot this on Snap.

1773.287 -> I remember when we first started doing Compute -

1777.407 -> that was the biggest thing to hit tech.

1780.257 -> It disrupted so many parts of the industry

1783.477 -> and I really enjoyed being a part of it.

1785.767 -> Joining Snap where the company is cloud native

1788.927 -> and taking advantage of every innovation

1790.987 -> including things like Graviton2 and Serverless.

1794.887 -> Every time a new innovation comes out, we are one of the first to adopt it

1797.717 -> because it lowers cost and improves performance.

1800.207 -> You might think that there isn’t much left to reinvent

1802.677 -> when it comes to Compute but the innovations just keep coming.

1807.043 -> [applause]

1810.677 -> Thanks, Jerry, we really appreciate the partnership with Snap

1814.827 -> and we appreciate you continuing to move to AWS.

1817.137 -> And Jerry knows what he is talking about as it relates to Compute.

1820.787 -> He spent a lot of time at Sun in the early years

1823.267 -> and then he was at AWS where he ran infrastructure

1826.777 -> which are our data centers and our network and our hardware

1829.697 -> before going to Snap to run their engineering.

1832.237 -> So he’s kind of seen the rise and change in Compute.

1834.907 -> He’s absolutely right that Compute is continuing

1839.737 -> to be reinvented as we speak.

1842.437 -> There’s three major modes of Compute that we see.

1846.487 -> Instances, which is the traditional way

1848.697 -> that people have run Compute - particularly

1850.657 -> when they want to get all the resources

1852.337 -> on a box for their application.

1854.777 -> And then smaller units of Compute like containers,

1858.087 -> where people build on these smaller microservices

1860.397 -> because it lets them move faster and be more portable.

1862.637 -> And then event-driven serverless computing,

1865.257 -> when they don’t want to worry about servers or clusters at all.

1868.177 -> And these three modes of Compute are here to stay.

1870.407 -> They are going to be here for a long period of time.

1872.167 -> Let’s start with instances. If you look at Amazon EC2,

1876.227 -> which is our service that vends instances to you

1879.427 -> we not only have the broadest array of instances

1882.087 -> but we also have the most powerful instances within those families.

1886.077 -> These are things like the fastest networking instances

1889.267 -> where we have 100 gigabits per second in all our recent generations

1892.387 -> and up to 400 gigabytes per second in our P4d’s.

1895.997 -> We have the largest high memory instances at 24 terabytes for SAP

1901.697 -> use cases in our high memory instances.

1904.337 -> We have the largest local storage instances with our D3en instances

1908.577 -> which are launched today with up to 336 terabytes.

1912.877 -> We have the most powerful machine

1914.327 -> learning training instances with our P4d’s.

1917.177 -> We have the most powerful machine learning inference instances

1920.267 -> with our Inf1 instances.

1922.377 -> We have the best price performance graphics

1925.067 -> instances with our G4ad instances which are coming in a week.

1929.927 -> We’re the only ones who give you a macOS instance type

1933.827 -> which we just launched last night with our macOS EC2 instances

1938.307 -> that lets Apple’s millions of developers

1940.427 -> now leverage the cloud much more easily.

1943.047 -> We’re the only provider that gives you the ability

1945.057 -> to run instances with Intel, AMD, and ARM chips.

1948.257 -> It’s a very different set of capabilities than anybody else has.

1953.327 -> And we’re also iterating and innovating in a much faster clip.

1957.787 -> And people often ask us - they say well,

1960.687 -> how are you innovating at such a rapid clip right now?

1963.687 -> And there are really two reasons.

1965.387 -> The first is what we’ve done with something called Nitro.

1970.197 -> We spend five years rebuilding our virtualization layer

1975.097 -> and our Compute platform, and we launched in 2017.

1977.987 -> What we did with Nitro was we took the virtualization of the security,

1982.157 -> Networking, and storage off of the main server chip

1985.097 -> and we put that on our own Nitro chips.

1988.187 -> And what that did for customers was it meant

1991.017 -> that you got all of the CPU to run your instances

1994.527 -> so you get performance that’s indistinguishable from bare metal

1998.317 -> but at a lower price.

1999.777 -> And it also meant you got a stronger security posture

2003.037 -> because it used to be that when you have to troubleshoot VM’s

2006.137 -> you had to worry about whether somebody would do something

2008.077 -> to the main server with your instances.

2010.147 -> But because the security now is on that separate Nitro chip

2013.467 -> you don’t have to worry about that.

2015.097 -> What it did for us was, because we broke out a bunch of those pieces

2019.337 -> into separable cards or chips,

2021.647 -> it meant we didn’t have to make changes

2023.717 -> in making all these different things change and evolve in lockstep,

2027.457 -> which allows us to innovate at a much faster clip.

2031.377 -> Now what that means for you is that you get instances now

2035.907 -> instead of every two to three years - you get innovative brand now

2039.857 -> change-the-game types of instances in months.

2042.457 -> That’s a big difference.

2044.167 -> The second thing that has happened that has allowed us

2047.267 -> to innovate at such a rapid clip is what we have done with chips.

2051.677 -> And we have a deep relationship with both Intel and AMD

2055.667 -> and we will for the foreseeable future.

2058.207 -> However, we realized a few years ago

2060.937 -> that if we wanted to continue to push the envelope on price performance -

2064.867 -> which you have asked us to do -

2067.057 -> we knew we were going to have to develop some of our own chips.

2069.357 -> And so we acquired a company called Annapurna who were sophisticated

2073.657 -> and very experienced chip designers and builders -

2077.197 -> and we put them to work.

2078.327 -> And we tried to pick chips that will allow you to get the most done -

2083.627 -> have the most impact for you business.

2085.257 -> And so what we started with was we started with generalized Compute

2090.027 -> and we had the team build a chip on top of ARM

2093.557 -> that we called Graviton.

2095.197 -> And so Graviton first manifested itself in these A1 instances we had

2099.347 -> and they were really for scale-out workloads,

2101.277 -> things like Web tier workloads and things of that sort.

2104.407 -> And people loved them.

2105.617 -> They used them a lot quicker than we ever imagined and they said:

2109.107 -> “Gosh if you would make the chip more powerful so we could use it

2113.247 -> for more of our workloads that would really change the game for us.”

2115.987 -> And so that’s what the team did with their second version

2119.137 -> of the Graviton chip, which we call Graviton2,

2122.377 -> which gives customers 40% better price performance

2126.517 -> than the most recent generations

2128.017 -> of the x86 processors from other providers.

2130.867 -> And those manifest themselves on C6g instances and M6g

2135.067 -> instances and R6g instances and T4g instances,

2139.637 -> 40% better price performance, that is a big deal.

2144.237 -> Think about what you can do for your business if you have 40%

2148.047 -> better price performance on your Compute.

2150.317 -> And customers have loved this.

2152.167 -> We have a large number of customers who are already using it

2155.117 -> from Nielsen to Netflix to Snap and Lyft and NextRoll,

2159.457 -> and we are announcing today a brand new Graviton2 instance

2165.487 -> which is going to be the C6gn instance,

2169.037 -> which is our Compute heavy and networking heavy instance

2171.637 -> with a 100 gigabits per second

2173.527 -> that will be coming in the next week or two.

2175.417 -> We are not close to being done investing and inventing with Graviton.

2180.907 -> People are seeing a big difference. Let me give you a couple of examples.

2184.387 -> Here's a company called Honeycomb.io and this is a blog that they wrote.

2187.477 -> And Honeycomb.io is a company

2188.967 -> that offers a comprehensive debugging tool for development teams

2192.327 -> and they moved over to the M6g instances.

2195.487 -> And you can see they are saving 40% on price performance

2198.917 -> for their Compute

2200.177 -> and using a lot fewer instances than they were using before.

2203.437 -> Or if you look at NextRoll, which is a leader in digital advertising,

2206.847 -> they are saving 50% on price performance.

2210.437 -> This is big deal and so Graviton is saving people a lot of money.

2215.577 -> Very excited about it.

2216.747 -> We have a lot more coming in this space.

2219.107 -> Then we asked ourselves what else can we apply our chip team

2222.247 -> to that would solve problems

2224.217 -> that we know are big growth areas for customers?

2227.167 -> And what we looked at was we took machine learning.

2230.647 -> That’s the other area that’s growing unbelievably quickly.

2234.837 -> And what you often see with machine learning

2237.637 -> is that people talk about the training

2241.187 -> because we’re still in the relatively early stages of machine learning so,

2244.447 -> so many people are just getting their models trained.

2247.297 -> And there’s a lot of effort on machine learning training.

2250.757 -> But if you have models that work at scale,

2254.617 -> you know, if you take for instance Alexa, where it’s a big old machine

2258.957 -> learning model that we have to train periodically.

2261.537 -> But we’re spitting out predictions and inferences

2264.727 -> by the millions every hour.

2266.847 -> And so it turns out that 90% of your cost is not in the training.

2270.757 -> It’s in the inference of the predictions.

2272.767 -> And nobody was focused to try to help customers save money

2275.397 -> and be efficient there.

2276.627 -> So we built an inference-focused chip which we announced last year

2281.407 -> at re:Invent called Inferentia.

2283.537 -> And if you look at the number of customers using Inferentia

2287.547 -> that also was really swelling. But it's interesting to see -

2291.127 -> if you want to take a big use case just look at Alexa.

2294.087 -> Alexa today has moved 80% of their predictions to Inferentia.

2298.677 -> They are getting 30% better cost and 25% better latency

2302.887 -> than they got from their prior chip. That’s a big deal.

2305.957 -> Now while we focused on inference

2308.747 -> we haven’t forgotten about machine learning training.

2311.597 -> And we continue to iterate there as well.

2313.507 -> Of course we have these P4 instances which are the most powerful machine

2316.677 -> learning training instances in the cloud,

2318.687 -> but people understandably still want to find ways

2322.487 -> to be cost effective when they are training models.

2324.877 -> So today I have two announcements to share with you

2327.587 -> that aim to help there.

2329.207 -> The first is that we will offer next year, in the first half

2333.807 -> of the year, Habana Gaudi-based Amazon EC2 instances.

2338.948 -> [applause]

2344.687 -> And so this is a partnership between AWS and Intel

2347.737 -> and it will use Gaudi accelerators that will provide 40%

2351.907 -> better price performance

2353.177 -> than the best performing GPU instances we have today.

2356.797 -> It will work with all the main machine-learning frameworks PyTorch

2361.687 -> as well as TensorFlow,

2363.367 -> and it will be available in the first half of the year.

2365.867 -> Now like with generalized Compute we know that if we want

2371.137 -> to keep pushing that price performance envelope on machine

2374.517 -> learning training we’re going to have to invest in our own chips as well.

2377.887 -> And so I am excited to announce as well AWS

2383.167 -> Trainium, which is our machine learning chip that’s custom

2387.537 -> designed by AWS to deliver the most cost-effective training in the cloud.

2392.877 -> [applause]

2398.807 -> So Trainium will be even more cost effective than the Habana chip

2403.027 -> I mentioned. It will support all the major frameworks -

2405.617 -> TensorFlow and PyTorch and MXNet. You will get to use the same Neuron

2410.957 -> SDK that our Inferentia customers use.

2413.437 -> So if you use Inferentia for inference

2415.537 -> it will be easy to also get going on our machine

2418.637 -> learning chip - on Trainium. It will be available

2421.397 -> both as an EC2 instance as well as in SageMaker

2424.107 -> and that’s coming the second half of 2021.

2428.187 -> So when you look at the unmatched array of instances

2432.127 -> that you have in AWS

2433.257 -> coupled with the relentless innovation in chips,

2437.037 -> and in the virtualization layer, and the Compute platform,

2440.337 -> you’re now getting reinvented instances every few months

2444.567 -> instead of every few years which is a big deal.

2447.527 -> Now one of the things that’s really interesting that we’ve seen

2451.657 -> in the last several years

2453.557 -> is that people are moving to smaller and smaller units of Compute.

2458.047 -> And what I mean - there really are containers and serverless.

2461.967 -> And so I will start with containers. People really love containers.

2466.287 -> It allows them to build on these smaller chunks of Compute

2468.617 -> which lets them move faster as well as be more portable.

2472.557 -> And if you look at the growth in containers in computing

2475.907 -> it’s pretty astounding.

2477.557 -> And the vast majority of it continues to run in the cloud on top of AWS.

2481.617 -> Of the containers that run in the cloud, about two thirds run on AWS -

2485.547 -> and that’s because while most other providers have one containers

2491.507 -> Offering - typically a managed Kubernetes offering -

2494.697 -> AWS has three. If it turns out that what you value most

2499.767 -> is using the open source Kubernetes framework,

2502.707 -> then we have our Elastic Kubernetes Service, EKS.

2506.697 -> If it turns out what you value most is having the deepest integration

2511.557 -> with the rest of the AWS platform,

2513.917 -> you use our Elastic Container Service, or ECS, which we can do

2517.177 -> because since we control it we can make sure

2519.067 -> that everything launches integrated with ECS right from the get-go.

2522.557 -> If what you value most in containers is running containers

2525.977 -> without having to worry about servers or clusters,

2529.167 -> then you use AWS Fargate, which is our serverless container

2532.457 -> offering which nobody else has anything like.

2535.647 -> And so one of the things that was interesting

2537.737 -> when we launched containers and we launched these offerings -

2541.347 -> we wondered once we had a managed Kubernetes service

2545.997 -> would people use the other offerings?

2547.657 -> Or do Kubernetes have so much resonance

2549.617 -> that people would only use that?

2550.817 -> And what we found is that all three of these container offerings

2555.157 -> continue to grow like a weed. Unbelievably fast.

2558.987 -> If you look at ECS we have over 100,000 active customers using it.

2563.477 -> We have billions of Compute hours on EKS run every week on AWS.

2567.857 -> And if you look, most of the net-new container customers in AWS

2572.157 -> start with Fargate because it’s so easy to get going.

2575.487 -> And we actually have a lot of customers who use two

2579.167 -> or even three of these container offerings

2581.737 -> because different teams have different preferences

2585.087 -> and have different use cases.

2586.977 -> You want the right tool for the right job.

2589.737 -> You don’t want one tool to rule the world,

2591.967 -> because there are lots of different teams and preferences and use cases.

2596.367 -> So if you look at the problems

2603.557 -> that people who use containers are trying to solve today

2607.337 -> they say it’s wonderful that you have these three offerings.

2610.807 -> It gives me so much more choice.

2612.297 -> I can use different offerings for different use cases.

2614.637 -> However, I still have a lot of my containers

2619.177 -> that I need to run on-premises

2620.727 -> as I am making this transition to the cloud.

2623.927 -> And so people really wanted to have the same management

2627.257 -> and deployment mechanisms that they have in AWS

2630.367 -> also on-premises - and customers have asked us to work on this.

2633.787 -> And so I am excited to announce two new things to you.

2636.487 -> The first is the announcement of ECS Anywhere,

2641.22 -> which lets you run ECS in your own data center.

2645.002 -> [applause]

2651.336 -> So ECS Anywhere allows you to have all the same AWS-style APIs

2657.43 -> and cluster configuration management pieces

2660.42 -> on-premises that you have in the cloud.

2662.887 -> So it makes it easy as you’re making this transition, to be able to run -

2666.997 -> if you’re running ECS on AWS you can run it on-premises as well.

2670.487 -> It works with all of your on-premises infrastructure.

2673.327 -> So not surprisingly when ECS customers hear that we’ve got

2678.487 -> ECS Anywhere they say, “Well, what about Kubernetes?”

2681.737 -> So I am also excited to announce Amazon EKS Anywhere,

2685.959 -> which lets you run EKS in your own data center.

2689.227 -> [applause]

2695.527 -> So again, just like with ECS

2698.357 -> Anywhere, EKS Anywhere lets you run EKS in your data

2703.217 -> centers on-premises alongside with what you’re doing in AWS,

2707.097 -> it works again with all of your on-premises infrastructure.

2710.477 -> And it’s interesting when we’ve talked privately to customers

2713.537 -> about EKS Anywhere, they have been very excited

2716.56 -> and they’ve said, “Well, I know that both

2719.099 -> ECS Anywhere and EKS Anywhere are coming in 2021,

2722.897 -> but I want to actually get started getting ready for EKS Anywhere."

2726.337 -> And so what we’re also announcing today

2728.937 -> is that we’re going to open source

2730.847 -> the EKS Kubernetes distribution to you

2734.027 -> so that you can start using that on-premises.

2736.027 -> It will be exactly the same as what we do with EKS.

2739.117 -> We will make all the same patches and updates

2740.997 -> so you can actually be starting to transition

2743.43 -> as you get ready for EKS Anywhere.

2747.597 -> So I think our container customers are going to be excited by ECS

2752.407 -> and EKS Anywhere.

2754.257 -> But what we’re also starting to see in addition to huge amounts

2758.737 -> of containers growth and adoption is that more and more customers

2763.837 -> are using event driven serverless computing.

2766.787 -> And we pioneered this concept a few years ago with Lambda.

2771.037 -> And the problem that we were trying to solve

2773.837 -> was that we had some customers who said:

2776.027 -> “Look, I have certain workloads when something happens,

2779.477 -> it triggers needing to spin up some Compute

2782.097 -> and it forces to spin up EC2 instances

2784.637 -> and multiple availability zones for fault tolerance,

2787.217 -> and then they only really run these jobs for a few hundred milliseconds,

2790.657 -> but I have to leave them all up all the time

2792.517 -> because I don’t know when the jobs are going to come in.”

2794.717 -> And they wanted us to solve that problem

2797.257 -> and that’s why we built Lambda.

2799.667 -> Lambda lets you set a trigger with a few lines of code.

2802.877 -> It spins up Compute. We spin it up for you.

2806.007 -> We spin it up in a fault tolerant way,

2808.127 -> and then we spin it back down when you are no longer running jobs

2810.917 -> and then we bill you in increments of 100 milliseconds.

2814.447 -> By the way, I am going to announce right now

2816.407 -> that we are changing the increments in which we’re billing

2819.247 -> from 100 milliseconds to 1 millisecond,

2821.937 -> which means for a number of these workloads customers

2823.947 -> will be able to save up to 70%.

2827.017 -> Customers have really loved this event-driven computing model.

2830.797 -> They say, “Look I love this. I don’t want to actually just do it

2833.437 -> for these use cases where I spin up a little bit of Compute.

2836.177 -> I actually would like to have my architecture

2838.347 -> for applications run that way.”

2841.237 -> And so we started actually building other services

2844.937 -> that allowed you to run serverless end to end.

2847.137 -> These are services like API Gateway and EventBridge and Step Functions.

2852.777 -> And then we added triggers in a lot of our AWS services

2857.477 -> so that you could actually trigger serverless actions

2860.547 -> from those services. We have them now in 140 AWS services,

2864.797 -> which is seven times more than you will find anywhere else.

2867.477 -> And just to give you a sense of the type of momentum

2870.687 -> that you are seeing with serverless right now, if you look inside Amazon

2875.357 -> and you look at all the new applications that were built in 2020,

2880.017 -> half of them are using Lambda as their Compute engine.

2883.187 -> That’s incredible growth.

2884.297 -> Hundreds of thousands of customers now are using Lambda.

2887.647 -> And so Lambda is growing really quickly.

2890.657 -> Serverless is growing really quickly. Containers are growing really quickly,

2894.597 -> and not surprisingly a lot of customers

2897.417 -> are using both containers and serverless together.

2901.227 -> And they’ve said, “I really wish that you would make it easier for us

2904.867 -> to run these two smaller units of Compute together

2907.297 -> than you make it today.”

2908.517 -> And they asked for a couple of different things.

2909.977 -> The first thing they said was “Look, our usage and adoption

2913.447 -> in containers is a little bit earlier than it is for serverless” -

2918.107 -> in part because it looked a little bit more like instances

2920.057 -> and the tools were set up a little bit better

2921.697 -> than they were for serverless early on.

2923.707 -> But we’re growing like gangbusters on serverless and Lambda.

2927.157 -> And so they said, “Look, we’ve invested so much time

2930.657 -> and energy into these container images

2932.817 -> that we deploy from, why can’t you just make it

2935.537 -> so we can deploy Lambda functions from these container images?”

2938.807 -> So I am excited to announce today the launch of Lambda Container Support,

2943.517 -> which lets you build Lambda-based applications

2945.877 -> using existing container development workflows.

2949.102 -> [applause]

2953.807 -> So now you can package code dependencies

2957.017 -> as any Docker container image or any Open Container

2961.017 -> Initiative compatible container image,

2963.207 -> or really any third-party based container image,

2966.267 -> or something that AWS has maintained as a base container image.

2969.297 -> It totally changes your ability to deploy Lambda functions

2973.837 -> along with the tools that you have invested on containers.

2976.717 -> So I think customers are going to find this very handy.

2979.717 -> Another big challenge they asked us to try to help solve

2983.347 -> is this issue of trying to manage the deployment

2987.377 -> of the smaller units of Compute where you end up in these situations

2991.447 -> where you have all of these microservices that have to be

2995.117 -> deployed together that comprise an application.

2997.997 -> And it’s actually quite difficult to do

2999.977 -> if you think about it - it’s different than instances.

3002.447 -> With instances, typically you build it as a single block of code.

3007.237 -> You have code templates you use like CloudFormation

3009.897 -> to provision the infrastructure as code, or services like CodePipeline

3013.567 -> that do the CI/CD for you or you do monitoring with CloudWatch.

3019.337 -> Once it’s set up it doesn’t change that much. And code is

3022.727 -> usually maintained

3023.977 -> as kind of a single release, so it stays pretty coordinated.

3026.837 -> And there are tools today that make this pretty straightforward.

3029.727 -> But if you look at containers and serverless,

3032.927 -> these apps are assembled from a number of much smaller parts

3036.237 -> that together comprise an application.

3039.417 -> It’s actually hard, you know,

3040.537 -> if you look at each of these microservices

3042.357 -> they have their own code templates

3043.937 -> and they have their own CI/CD pipelines.

3046.717 -> They have their own monitoring and most are maintained by separate teams

3050.617 -> and it means that there’s all these changes happening

3053.047 -> all the time from all these different teams.

3055.707 -> And it’s quite difficult to coordinate these

3057.877 -> and keep them consistent - and it impacts all sorts of things

3061.147 -> including quality and security.

3062.567 -> And so there really isn’t anything out there that helps customers

3067.157 -> manage this deployment challenge in a pervasive way.

3071.117 -> And so this is something that our team has thought a lot about.

3074.087 -> And I am excited to announce the launch of AWS Proton,

3077.527 -> which is the first fully managed deployment service

3080.267 -> for container and serverless applications.

3083.266 -> [applause]

3088.857 -> So this is a game changer for managing

3091.667 -> the deployment of microservices. Here’s how it works.

3094.707 -> A central platform team or anybody central to an application

3098.307 -> will build a stack.

3099.807 -> And a stack is really a file that includes templates that use code

3104.177 -> to define and configure AWS services used in a microservice,

3107.537 -> including identity, and including monitoring.

3110.547 -> It also includes a CI/CD pipeline

3113.127 -> template that defines the compilation of the code

3115.707 -> and the testing and the deployment process.

3118.157 -> And it also includes a Proton schema that indicates parameters

3122.037 -> for the developers that they can add, things like memory allocation,

3124.947 -> Or a Docker file, or something like that.

3126.617 -> Basically, everything that’s needed to deploy a microservice

3130.147 -> except the actual application code.

3132.397 -> Then the platform - the central platform team -

3135.837 -> will publish the stack to the Proton console -

3139.277 -> pretty often they’re going to publish lots of stacks

3141.347 -> because there’s so many different use cases for microservices.

3144.427 -> And then when a developer is ready to deploy their code

3148.307 -> they will pick the template that best suits their use case, plug

3151.727 -> in the parameters they want, and hit “Deploy”.

3154.517 -> And Proton will do all the rest.

3156.257 -> It provisions the AWS services specified in the stack

3159.747 -> using the parameters provided.

3161.997 -> It pushes the code through the CI/CD pipeline that compiles and tests

3166.177 -> and deploys the code to AWS services.

3168.747 -> and then sets up all the monitoring and alarms.

3171.767 -> Proton also lists in the console

3174.837 -> all the downstream dependencies in a stack.

3177.777 -> So if the central engineering team

3180.237 -> makes some kind of change to that stack,

3183.847 -> they know all the downstream microservices teams

3187.627 -> that need to make those changes

3189.127 -> and can alert them and can track whether or not they made it.

3191.287 -> This is a game changer with regards to deploying containers

3196.207 -> and serverless apps. We’re very excited about this

3198.457 -> and I think our customers are going to be as well.

3202.277 -> So think about ten years ago, if you flash back ten years ago,

3208.167 -> the CPU and GPU providers effectively didn’t have any competition

3212.887 -> and it meant that you were able to get new instances

3215.127 -> every two to three years or so. Look at today.

3218.507 -> Look at what Graviton is doing with 40%

3221.147 -> better price performance on your most recent generations of x86 processors.

3227.157 -> It totally changes what you can get done Compute-wise.

3230.367 -> Or look at what’s happening with Inferentia and some of the machine

3233.527 -> learning training chips that I talked about earlier.

3236.187 -> You are going to get new instances that allow you to reinvent

3240.037 -> your business every few months now instead of every few years.

3243.717 -> Ten years ago, we weren’t really talking very much about containers

3247.617 -> or serverless at all. Look how fast those are growing.

3252.987 -> And if you think about it, if you think about these container offerings

3257.027 -> and how many you have available on AWS

3259.287 -> and the ability now to be able to manage them the same way both in AWS

3264.497 -> as well as on-premises,

3265.737 -> and you think about what’s happening with serverless

3268.257 -> and you think about the tools now that are allowing you

3270.527 -> to use these new smaller units of Compute

3273.107 -> and to be able to deploy much more easily,

3275.757 -> there is an incredible amount of reinvention happening in Compute.

3280.517 -> We are not close to being done reinventing in Compute.

3284.067 -> So you can expect a lot more to come.

3286.097 -> Now Compute is obviously being reinvented,

3290.197 -> but so is data and data stores in a very big way.

3294.527 -> And to share some thoughts about what he is seeing in this space,

3297.457 -> and he has been there since the very start of the cloud,

3300.767 -> it’s my pleasure to welcome the CEO and Founder of SmugMug, Don MacAskill.

3306.101 -> [music]

3310.897 -> The launch of Amazon S3 in 2006 was a seminal moment for data storage.

3315.917 -> This invention totally changed our trajectory at SmugMug.

3319.817 -> Today we use almost everything AWS has to offer.

3323.857 -> They’ve continued to add more and more data stores and analytics tools

3328.217 -> that have let us really pick the best tools

3330.877 -> for the best job to serve our customers.

3333.537 -> Tools like Athena, Redshift, Elasticsearch, Kinesis, and DynamoDB,

3338.927 -> which allows us to scale and serve photos

3342.037 -> at billions of requests per day. The cloud has totally reinvented

3346.557 -> how we store, secure, analyze, and share data at a scale

3351.867 -> that we couldn’t have even imagined 14 years ago.

3356.547 -> [applause]

3361.947 -> Thank you, Don.

3363.347 -> SmugMug was the very first big AWS customer

3368.267 -> and I remember in the early days of AWS -

3370.867 -> this would have been March 2006 - shortly after we launched S3.

3376.077 -> Basically, our only sales-person at the time, Rudy Valdez,

3379.367 -> wrote us an email and he said, “I just got a call from SmugMug

3383.477 -> and they are going to put 6 terabytes of data in S3.”

3386.367 -> And we said, “Wait, terabytes or gigabytes?”

3388.597 -> And he said, “No, terabytes.” We said, “With a T or with a G?”

3391.247 -> He said, “No, no, with a T.” We just couldn’t believe it.

3395.077 -> But SmugMug and Don have seen the rise of the cloud

3400.777 -> and the change in data and data stores.

3402.777 -> And in fact have been a big piece of informing what we’ve built.

3406.997 -> They’ve given us great feedback and we really appreciate the partnership.

3410.587 -> And Don is correct in explaining that data and data stores

3415.097 -> is radically being reinvented as we speak.

3418.277 -> And if you think about it today

3421.237 -> with the way that the cloud has made storage so much less expensive,

3424.997 -> and then Compute to do something with that storage so much less expensive,

3429.447 -> it’s astounding how much data is being created and stored today.

3434.367 -> Just a couple of data points: Analysts say that in every hour today

3439.817 -> we’re creating more data

3441.447 -> than what we did in an entire year 20 years ago.

3444.867 -> Or they predict that in the next three years

3448.427 -> there will be more data created than in the prior 30 years combined.

3453.707 -> This is an incredible amount of data growth and the old tools

3457.737 -> and the old data stores that existed in the last 20-30 years

3461.807 -> are not going to cut it to be able to handle this.

3464.657 -> Every single type of data store is being reinvented

3468.857 -> and will be reinvented multiple times over.

3470.667 -> Let me give you a few examples. Let’s start with block storage.

3473.747 -> So if you look at block storage,

3475.707 -> it’s a foundational and pervasive type of storage used in computing.

3481.007 -> Unlike object storage which has metadata that governs the access

3486.097 -> and the classification of it.

3489.007 -> Block storage has its storage or its data

3493.197 -> split into evenly sized blocks

3495.437 -> that just have an address and nothing else.

3497.177 -> And since it doesn’t have that metadata it means

3499.937 -> that the reads and the writes and access to them are much faster.

3504.097 -> And it’s why people use block storage with virtually every EC2 use case.

3510.647 -> It’s also why that probably the very most animated debate

3515.137 -> that we had as an AWS team before we launched AWS

3519.067 -> was: Could we launch EC2 with just direct-attached storage

3523.707 -> that was ephemeral?

3525.037 -> Or could we not launch EC2 until we had high performance block storage?

3529.447 -> And it was a very animated debate, lots of opinions

3532.877 -> and we ultimately kind of figured out

3535.017 -> that it was going to take us two more years to build

3537.477 -> to a high-performance block store and we didn’t wait to give you EC2.

3541.427 -> So we launched EC2 with just that direct-attached storage.

3544.387 -> But we hustled like heck to get to building a block store

3549.047 -> and we launched our Amazon Elastic Block Store, or EBS, in 2008.

3553.737 -> It was the first high performance scalable block store in the cloud

3557.877 -> and it gave you an easy ability to provision what storage you needed

3561.187 -> and what IOPS you needed and what throughput you needed.

3564.257 -> And then you could adjust as you saw fit.

3567.527 -> In 2014, we built the current version of our general-purpose volume.

3573.557 -> It was called GP2 which is what the vast majority of EBS

3577.367 -> workloads run on top of. Every imaginable workload you can imagine.

3581.817 -> The feedback that we’ve gotten the last year or two from customers

3585.147 -> is that, “We love GP2. But if we had a wish list

3588.987 -> there’s a couple of things that we would like from you.”

3591.307 -> We’d like one, of course, we’d like the cost per gigabyte to be less

3595.647 -> and then we want to be able to sometimes scale

3599.047 -> throughput or scale IOPS

3601.297 -> without also having to scale the storage with it,

3603.967 -> which is what GP2 asks you to do.

3606.407 -> So the team has been working on this and I’m excited to announce today

3610.197 -> the new version of our general purpose volumes, gp3,

3614.257 -> which allows you to have 20% better cost per gigabyte

3619.267 -> as well as be able to provision IOPS

3621.577 -> and throughput separately from storage.

3624.406 -> [applause]

3630.737 -> So the baseline performance of these new gp3 volumes is 3,000 IOPS

3635.387 -> and 125 megabytes per second, but you can burst that and scale

3638.997 -> that up to a peak of 1,000 megabytes per second,

3641.897 -> which is four times that of GP2.

3644.497 -> And you’ll see that customers will be able to run many

3647.807 -> more of their demanding workloads on gp3s and they even work for GP2.

3652.557 -> Yet, there are certain types of use cases

3656.047 -> where you need a lot more IOPS than you get with gp3s.

3660.067 -> Cluster databases are a good example of this.

3662.807 -> And that’s why we built our Provisioned IOPS

3665.647 -> or io2 volumes for EBS

3668.287 -> which give you four times the amount of IOPS of a gp3 volume.

3671.987 -> And you can, of course if you start with gp3s,

3674.057 -> and it turns out you think you need io2s,

3675.977 -> you can use our elastic volumes feature

3678.007 -> to seamlessly move that volume.

3681.407 -> But it’s also true, while people run their most demanding

3685.567 -> IOPS applications on top of io2s,

3689.237 -> that there are still workloads that are even more demanding.

3692.517 -> Some of the most demanding Oracle databases

3694.617 -> are SAP HANA databases and offerings

3697.887 -> where people need in the neighborhood of 200,000 IOPS

3703.087 -> or 3,000 to 4,000 megabytes per second of throughput.

3707.097 -> And it’s true that you can stripe together

3709.557 -> a number of these io2 volumes

3711.107 -> to give you more collective IOPS of throughput

3713.817 -> but the more that you have to stripe together,

3715.747 -> the harder it is to manage and to keep consistent

3720.057 -> and to get the performance you want. And so what customers have said is,

3722.677 -> “Look, what you’re forcing me to do for these most demanding workloads,

3728.217 -> is you’re forcing me to use these storage area networks,

3730.967 -> or SANs, which really they’re a bunch of clustered discs

3735.517 -> with networking attached to them.

3737.757 -> You’re forcing me to run these SANs on-premises

3740.297 -> and I don’t want to have to run these SANs because they’re expensive.”

3744.377 -> If you get a good deal on a SAN it’s about $100,000,

3747.697 -> but then when you factor in support and maintenance,

3750.037 -> and maintaining it across multiple data centers,

3752.607 -> you get to $200,000 pretty quickly

3754.387 -> and if it turns out you exceed the capacity of a SAN,

3757.247 -> you have to buy an increment

3759.057 -> of another couple hundred thousand dollars.

3761.357 -> So people don’t like the cost, but it’s also hard to manage.

3765.197 -> You have to update and maintain the software,

3767.637 -> you have to do the same thing with the hardware,

3769.347 -> you have to manage it in data centers.

3770.947 -> You have to make sure it’s fault-tolerant.

3772.977 -> The customers have said, “Look, you’ve left me with no option

3777.267 -> because there is no SAN in the cloud.” Until now.

3781.177 -> I’m excited to announce io2 Block Express,

3784.377 -> which is the first SAN built for the cloud.

3787.461 -> [applause]

3793.897 -> So Block Express volumes give you up to 256,000

3797.597 -> IOPS, 4,000 megabytes per second, 64 terabytes of storage capacity.

3803.117 -> That’s 4X the dimensions of the io2s on every single one of those.

3808.637 -> That is massive for a single volume.

3811.647 -> Nobody has anything close to that in the cloud

3814.337 -> and what it means is that you now get the performance of SANs in the cloud,

3820.067 -> but without the headaches around cost and management.

3822.607 -> You just spin up a Block Express volume,

3826.207 -> we manage it and we’ll back it up and maintain it for you.

3829.267 -> We’ll replicate it across an AZ.

3831.917 -> You can use the EBS snapshot capability to auto lifecycle,

3835.907 -> a policy to back it up to S3, and if you need more capacity,

3840.687 -> you just spin up another Block Express volume

3843.347 -> at a much lower cost

3844.697 -> than trying to do it at $200,000 a clip.

3848.537 -> So, we will add additional SAN features in 2021,

3853.847 -> things like multi-attach and IO fencing

3856.867 -> and make elastic volumes work with it.

3858.917 -> And this is pretty exciting.

3860.187 -> There was a lot of very complicated, sophisticated,

3863.557 -> innovative engineering to give you Block Express,

3866.987 -> but it’s a huge gamechanger for your most demanding applications

3871.347 -> that you want to run in EBS. So block stores are being reinvented.

3875.737 -> What about databases?

3877.327 -> Databases as you all know are right in the middle of every application

3881.147 -> and they’re hard to manage.

3882.227 -> You have to set them up, you have to tune them,

3884.077 -> you have to patch them, you have to make sure

3885.667 -> you get the right fault tolerance and performance.

3887.987 -> And it’s why companies have so many database professionals

3891.817 -> that they have to hire.

3893.447 -> It’s also why we built our relational database service,

3896.717 -> which is our managed relational database service

3899.247 -> which we launched about ten years ago and has been wildly popular.

3903.047 -> But if you look at, despite the growth of things like RDS,

3908.297 -> it’s still true that the overwhelming majority of relational databases

3914.187 -> live on-premises and they live largely

3917.917 -> with these old-guard commercial grade database

3921.987 -> providers named Oracle and Microsoft with SQL Server.

3925.987 -> And this is an unhappy place for customers.

3928.657 -> We’ve talked about this for several years.

3930.197 -> This is why you’re seeing so much movement.

3931.857 -> But it’s unhappy because those offerings are expensive,

3935.757 -> they’re proprietary, they have high amounts of lock-in

3939.477 -> and those companies have punitive licensing terms

3942.197 -> where they’re willing to audit you and if they find any discrepancy,

3946.537 -> they extract more money out of you.

3948.657 -> And these companies also have no qualms

3951.917 -> about changing the licensing terms mid-stream on you.

3955.577 -> If you just look at what Microsoft did with SQL Server

3958.017 -> in the last year or two.

3959.657 -> They basically changed the terms so you can’t use your SQL Server

3962.627 -> licenses anywhere but Microsoft’s cloud.

3965.207 -> Is that good for customers? Hell no. Is that good for Microsoft?

3969.547 -> I think they think so. I think it’s short-term thinking

3972.417 -> because our experience over the fullness of time

3974.417 -> is that customers flee companies the first chance they get

3977.707 -> when they feel like they’re being abused.

3980.037 -> And this is something that customers are fed up with

3983.727 -> and they’re sick of - and it’s why they’re moving as fast

3986.647 -> as they can to these open engines like MySQL and Postgres.

3990.647 -> But to get the type of performance

3993.407 -> you get in a commercial grade database

3995.087 -> in these open engines, you can do it, but it’s hard work.

3998.407 -> We’ve done a lot of it at Amazon and our customers

4000.387 -> asked us to fix that problem for them and that’s why we built Aurora.

4005.747 -> And Aurora has 100% compatible versions with MySQL and Postgres.

4011.787 -> It has several times better performance

4014.507 -> than those community grade versions.

4016.687 -> It has at least the same fault tolerance and durability

4020.067 -> and availability

4021.167 -> as the commercial grade databases, but at one tenth of the cost.

4026.277 -> This is why customers have been flocking to Aurora

4029.447 -> as quickly as they have.

4030.707 -> It’s the fastest growing service in the history of AWS.

4033.337 -> It has been since its launch,

4035.357 -> and you see that we have over 100,000 customers now using Aurora.

4041.147 -> These are companies like Airbnb and AstraZeneca and BP,

4044.367 -> and Capital One, and Fannie Mae, and Petco, and Verizon, and Volkswagen.

4050.387 -> One of the great things though about having a service

4053.727 -> that people love so much and is growing so fast

4056.577 -> is that you get a lot of feedback

4059.837 -> on what else people would love you to build and that is fuel for us.

4063.157 -> Please keep it coming. That is how we choose what to build.

4066.997 -> And a lot of our Aurora customers said, “Look, we love Aurora.

4071.127 -> If we had a big ask it would be that we’d like to be able to run

4074.807 -> Aurora taking advantage of that serverless architecture

4077.887 -> where we didn’t have to think about servers and clusters.”

4080.807 -> And so that’s why we built Aurora Serverless,

4084.677 -> which is really an auto-configuration for Aurora

4087.737 -> which allows you to set up an Aurora database

4091.747 -> and then when you need to scale up because of capacity,

4093.887 -> we’ll scale it up in 5 to 50 seconds,

4096.897 -> usually doubling the capacity each time you need more capacity.

4100.167 -> And we have thousands of customers who’ve been using Aurora Serverless

4103.987 -> primarily for dev and test workloads.

4105.997 -> And they said, “Look, we want to run… We love it.

4108.697 -> We want to run it for production workloads,

4111.047 -> but we need some things from Aurora,

4113.357 -> what we’d normally get from Aurora if we’re going to do that.”

4115.947 -> And they said, “We need Multi-AZ. We need Read Replicas

4120.187 -> and then when we need to scale up capacity-wise,

4122.927 -> we need it to happen instantaneously. We can’t wait 5 to 50 seconds,

4126.347 -> and we only want to scale up in the precise increment

4129.727 -> that we need to scale up.”

4131.207 -> So the team went away and started working on that this year.

4134.327 -> I’m excited to announce the launch of Amazon Aurora Serverless v2,

4139.087 -> which allows you to scale to hundreds of thousands of transactions

4142.107 -> in a fraction of a second.

4144.225 -> [applause]

4150.717 -> And so Aurora Serverless v2 totally changes the game for you

4156.627 -> with serverless as it relates to Aurora.

4159.797 -> You can scale up as big as you need to instantaneously.

4163.657 -> It only scales you up in the precise increments that you need to.

4167.407 -> So if you’re using Aurora Serverless v2,

4169.817 -> you can save up to 90% versus provisioning Aurora for the peak.

4174.547 -> It adds in a lot of the Aurora capabilities people want - and Multi-AZ

4178.427 -> and Global Database and Read Replicas and Backtrack and Parallel Query -

4183.447 -> and it really makes Aurora Serverless v2 ideal

4186.437 -> for virtually every Aurora workload.

4189.237 -> MySQL is available for you now with Aurora Serverless v2

4193.077 -> and Postgres will be available in the first half of 2021.

4196.277 -> Now I ask you, how many of these old-guard commercial grade

4200.907 -> database companies would build something like Serverless v2

4205.307 -> that’s clearly going to take a meaningful amount of revenue

4208.347 -> away from their core database offering?

4210.727 -> I wouldn’t hold your breath. I think the answer is none of them.

4213.577 -> They’re just not built that way.

4215.797 -> But we have a different way of thinking about our business,

4218.747 -> which is that we’re trying to build a set of relationships

4222.727 -> and a business that outlasts all of us.

4225.017 -> And the best way we know of doing

4226.837 -> that is listening to what customers care about.

4230.047 -> And if we can help you build more, more efficiently, more effectively,

4234.087 -> change your experience for the long-term,

4236.527 -> even if it means short-term pain for us or less revenue for us,

4240.477 -> we’re willing to do it because we’re in this with you for the long haul.

4243.307 -> I think that’s one of the reasons why customers trust AWS in general

4247.457 -> and have trusted us in the database space over the last number of years.

4251.187 -> If you look at it, just in the last few years,

4253.657 -> we’ve had more than 350,000 databases migrate to AWS

4258.697 -> using our Database Migration Service

4260.777 -> and the pattern that usually follows, they say,

4262.507 -> “Look, we used the Database Migration Service to move our database data.

4267.057 -> We used your Schema Conversion Tool to convert the schema,

4270.667 -> but there’s a third area that’s making this harder

4274.207 -> than we wish it were, that we want your help with,

4276.107 -> and that’s trying to figure out what to do with the application code

4279.927 -> that is tied to that proprietary database.

4283.887 -> And so customers have asked us,

4285.987 -> “Can you do something to make this easier for us

4288.657 -> because we want to move these workloads to Aurora?”

4291.237 -> And especially with the way they’ve watched Microsoft

4294.007 -> get more punitive and more constrained and more aggressive

4297.097 -> with their licensing, they want help.

4299.637 -> So the team has been working on this for a little bit more than a year

4302.647 -> and I’m excited to announce today

4304.817 -> the launch of Babelfish for Amazon Aurora Postgres,

4308.377 -> which lets you run SQL Server applications on Aurora

4311.747 -> Postgres with little to no code changes.

4315.266 -> [applause]

4321.477 -> So Babelfish is a new translation capability that lets you run

4325.747 -> SQL Server applications on Aurora Postgres.

4328.337 -> And what Babelfish does is it understands

4331.447 -> Microsoft SQL Server’s

4332.697 -> T-SQL dialect and it creates a tabular

4335.567 -> Data Stream or a TDS endpoint for your app to connect to,

4338.617 -> and it understands Microsoft’s proprietary schemas.

4341.817 -> And what it means for you is that now you can use the Database Migration

4345.917 -> Service to move your database data, the Schema Conversion

4348.657 -> Tool to move your schema or convert your schema, and then you can use

4352.957 -> Babelfish to update your application configuration

4356.447 -> to point to Aurora Postgres instead of SQL Server

4359.427 -> and you get to shed those expensive and constraining SQL Server licenses.

4364.027 -> Because Aurora now is able to understand both T-SQL and

4368.247 -> Postgres, you can write application functionality in Postgres

4371.447 -> to run side by side with your legacy SQL Server code.

4374.937 -> And so customers that we’ve spoken to privately about this,

4379.207 -> to say they’re excited would be one of the more large understatements

4383.327 -> I could make. Very excited. And they were so excited about this

4387.637 -> that as we were preparing for re:Invent and discussing it,

4391.397 -> we realized that this was probably bigger than just Aurora Postgres.

4395.497 -> That people really wanted the freedom

4397.627 -> to move away from these proprietary databases

4400.277 -> and to Postgres which is where most people move from these.

4403.497 -> And so we decided that we’re going to open source

4406.057 -> Babelfish. And so I’m excited to announce Babelfish for Postgres,

4410.947 -> which is an open source project.

4413.553 -> [applause]

4419.687 -> So Babelfish for Postgres will use the permissive Apache 2.0 license.

4424.857 -> It means that you can modify or tweak or distribute

4427.527 -> in whatever fashion you see fit.

4430.747 -> All the work and planning is going to be done on

4432.757 -> GitHub so you have transparency of what’s happening with the project.

4436.137 -> So this is a huge enabler for customers

4438.897 -> to move away from these frustrated, old-guard proprietary

4442.797 -> databases to the open engines of Postgres.

4446.367 -> You can sign up for Babelfish for Aurora Postgres today

4450.537 -> and then you’ll be able to sign up for the open source project in 2021.

4455.777 -> So block stores are being reinvented,

4459.747 -> relational databases are being reinvented.

4463.517 -> Heck, actually all of databases are being reinvented.

4466.137 -> If you think about it,

4467.267 -> we’ve been talking about this for the last few years.

4471.657 -> That world that you lived in for 20 or 30 years

4475.257 -> where you used a relational database for every single workload,

4479.287 -> that time has come and gone.

4483.547 -> If you have data in volumes that are gigabytes and sometimes terabytes,

4490.067 -> you can get away with using a relational database for everything.

4493.557 -> It’s not ideal, but you can get away with it.

4495.067 -> But in a world where you’re now dealing with terabytes of data

4498.757 -> and petabytes of data, and sometimes exabytes of data,

4502.047 -> a relational database doesn’t make sense.

4504.487 -> It’s too expensive, it’s too complicated,

4507.237 -> and it doesn’t perform as well as purpose-built databases

4510.707 -> that do a particular workload or use case extremely well.

4515.877 -> And that’s what we’ve been working on the last several years.

4518.217 -> We have built seven of these purpose-built databases,

4521.157 -> more than you’ll find anywhere else by a fair amount,

4523.767 -> that allow you to have the right tool for the right job.

4526.427 -> So, I’ll give you some examples. If you’re a company like Lyft

4530.227 -> and you have millions of geolocation and driver combinations,

4534.757 -> you don’t want a relational database.

4536.667 -> It’s too complex, it’s too expensive, it’s not performing.

4539.527 -> You want a high throughput, low latency, key-value store

4544.057 -> like DynamoDB, or if you’re a company like Peloton,

4547.637 -> you want to show your dashboards in microseconds,

4551.097 -> you want an in-memory database like ElastiCache.

4554.897 -> Or if you’re a company like Nike and you’re trying to connect

4557.367 -> all these different graphs of information

4560.537 -> that have relationships attached to them,

4562.297 -> you want a graph database like Neptune.

4564.467 -> Or if you’re doing work at the edge with IoT

4568.367 -> where the data’s coming in a timestream format,

4570.727 -> you want a time series database called Timestream, which we have.

4575.457 -> If you want to run managed Mongo, if you want to run managed Cassandra,

4578.957 -> you want those databases

4580.957 -> that allow you to have the right tool for the right job.

4584.167 -> I think you’re seeing the same exact thing happening

4587.077 -> with purpose-built analytics stores.

4589.517 -> So if it turns out that you want to do

4591.497 -> querying directly on your data lake, or in S3, you use Athena.

4597.127 -> If it turns out that you want to actually process vast

4599.977 -> amounts of unstructured data across dynamically scalable clusters

4604.057 -> using popular distributor frameworks like Spark or Hadoop or Presto,

4608.697 -> you use EMR.

4610.207 -> If you want to do large-scale analytics on log data

4613.407 -> for operations you use our Elasticsearch service.

4616.387 -> If you want to do real-time processing and streaming data,

4618.827 -> you use Kinesis, and if you have structured data

4622.197 -> where you need super-fast querying results,

4624.767 -> you want something like a data warehouse;

4626.377 -> You want Redshift, which was the first data warehouse built for the cloud -

4630.637 -> continues to be the largest data warehouse in the cloud -

4633.627 -> and they continue to innovate at a rapid rate.

4635.477 -> Last year at re:Invent you saw the RA3 instances,

4638.697 -> which separates storage from the compute, which people have loved.

4642.617 -> We’re just around the corner from the general availability of AQUA

4646.637 -> which moves the compute to the storage

4648.157 -> and will give you 10x better query capabilities

4650.437 -> in terms of speed than anywhere else.

4652.667 -> You want these purpose-built databases

4655.937 -> and these purpose-built analytics stores.

4658.017 -> And we see customer flocking to them.

4660.487 -> Now, it’s actually brought up a really interesting challenge

4664.197 -> and question

4665.277 -> for customers which is: Most companies either have a data lake

4670.237 -> or will build a data lake to take all that data from disparate silos

4674.197 -> and move it together,

4675.527 -> so you have one place where you can do your analytics

4678.037 -> and your machine learning from. And most of those are built on S3.

4681.537 -> We have tens of thousands of data lakes built in S3,

4684.317 -> more than you’ll find anywhere else because of the security

4687.107 -> and reliability and governance and compliance capabilities,

4690.277 -> and the broad features and the flexible

4692.947 -> and cost-effective performance of S3.

4696.427 -> So,you can have these data lakes

4699.427 -> that you centralize all your data,

4701.337 -> so you can run analytics and machine learning from,

4704.067 -> but as we just talked about, you’re increasingly seeing customers

4707.397 -> using more and more of these purpose-built data stores.

4710.907 -> And so customers say, “Well, I want my data

4714.567 -> to be able to move back and forth between these different stores

4717.577 -> because it’s very useful to take some of these views that I have

4720.927 -> and use them in other spots.”

4722.757 -> And while we have capabilities in many of these services

4725.727 -> that let people move them back and forth,

4727.947 -> it’s really not easy enough for people to do,

4730.437 -> such that they do it in a pervasive way.

4732.407 -> So customers really,

4733.907 -> really want more freedom for their data to move around.

4738.317 -> Our team’s worked on this for the better part of a year,

4740.997 -> and I’m excited to announce today the launch of AWS Glue Elastic Views,

4747.507 -> which lets you easily build materialized views

4749.917 -> that automatically can find and replicate data

4751.957 -> across multiple data stores.

4754.368 -> [applause]

4760.877 -> So, Glue is AWS’s ETL service, and you can write SQL

4766.087 -> or you can use our Glue studio visual tool

4771.447 -> to do extracting and loading and orchestration.

4774.647 -> And we recently a few weeks ago launched something called DataBrew

4778.077 -> to make it easy to clean and normalize data with no coding.

4781.597 -> But Elastic Views is different.

4783.607 -> What it lets you do is it lets you write a little of SQL

4786.767 -> to create a virtual table or a materialized view of the data

4790.797 -> that you want to copy and move and combine from source data store

4795.347 -> to a target data store.

4798.057 -> Well in the old days you’d have to figure out a way to write code

4802.447 -> and make sure you can get all the DynamoDB attributes

4804.797 -> and figure out how to move that, and move it in sync,

4807.517 -> keep it up-to-date as the attributes changed

4809.627 -> and new ones came in or things got adjusted in any way.

4813.897 -> All of that muck is taken away by Elastic Views.

4817.737 -> And so what Elastic Views does is it allows you

4820.597 -> to set up a materialized view to copy that data

4824.427 -> and move that data from one of those source data stores

4828.397 -> to a target data store and then it manages

4831.297 -> all of the dependencies of those steps as they move.

4835.147 -> And if something changes in the source data store,

4839.697 -> Elastic Views takes that and automatically changes it

4842.417 -> in the target store in seconds. If it turns out, for whatever reason,

4846.287 -> the data structure changes from one of the data stores,

4851.137 -> Elastic Views will alert the person that created the materialized view

4854.317 -> and allow them to adjust that.

4856.137 -> This is a huge game changer in being able to move data.

4860.187 -> It takes a lot of the work that people had to do –

4863.447 -> and frankly just found it so much work that they rarely did.

4867.317 -> You know, when you have the ability to move data

4872.737 -> and to have purpose-built data stores and have a data lake,

4875.857 -> but also move that data easily from data store to data store,

4881.177 -> there is a lot of power in giving that freedom of movement of data

4884.927 -> and this is going to be a big game changer for customers,

4886.897 -> having Elastic Views. We’re very excited to give it to you.

4890.957 -> A company that I think is doing one of the most revolutionary

4895.597 -> and transformational things around today

4899.147 -> and that’s also using data and computing in very innovative ways –

4904.987 -> is what’s happening at a company called Boom,

4907.477 -> which is trying to build the first supersonic airplane in 60 years.

4911.457 -> And to share with you how they’re doing that

4913.707 -> and what they’re doing with data and compute on top of AWS,

4916.887 -> it’s my pleasure to welcome the founder and CEO of Boom,

4919.837 -> Blake Scholl.

4921.887 -> [applause]

4925.539 -> [music playing]

4946.325 -> [music]

4966.407 -> So what does a revolution in high-performance computing

4969.647 -> have to do with a revolution in high-performance aircraft?

4973.057 -> What does cloud computing have to do

4975.287 -> with how we fly through actual clouds?

4978.277 -> And why now is a software engineer building

4981.167 -> the next commercial aircraft company?

4983.997 -> Air travel is integral to our modern lives,

4987.757 -> yet aviation is a domain starving for step-change innovation.

4993.137 -> The last big new thing was the jet airliner

4996.257 -> invented more than six decades ago.

4999.477 -> Tokyo to Seattle has been a nine-hour flight for going on 60 years.

5006.067 -> At Boom, we are guided by one fundamental mission –

5009.777 -> to make the world dramatically more accessible.

5013.647 -> And by the end of the decade, millions of everyday travelers

5017.517 -> will enjoy the benefits of supersonic flight

5020.117 -> aboard Overture, an airliner twice as fast as any flying today.

5025.817 -> That means that Tokyo will be just four-and-a-half hours

5028.817 -> from Seattle, and London just three-and-a-half from New York.

5033.347 -> Speed unlocks new possibilities for human relationships

5038.357 -> and business connections.

5040.197 -> That’s why major aerospace players such as Rolls Royce, Japan Airlines,

5045.607 -> and the United States Airforce are among Boom’s partners.

5049.657 -> All of this is possible thanks to advances

5052.257 -> in computing that enable a startup company

5055.647 -> to spark a revolution in speed.

5059.427 -> Twenty years ago, I hit two life milestones as a new graduate.

5064.397 -> I started my first job here at Amazon as a software engineer,

5068.747 -> and I started taking flying lessons

5070.747 -> just a few miles down the street at Boeing Field.

5074.347 -> Around the same time, Amazon was building the fundamental web services

5078.517 -> that would later become AWS,

5081.097 -> charting a new future of computing across all industries.

5085.527 -> Little did I know that these things would intersect

5088.827 -> so powerfully in my future.

5091.397 -> I never could have predicted that the innovations in cloud

5094.337 -> computing happening at Amazon and AWS

5097.297 -> would fundamentally change how we all fly.

5100.807 -> Well fast forward to today and Boom is designing

5103.787 -> and building the world’s fastest and most sustainable airliner,

5108.657 -> and we’ve just announced that we’re going all in with AWS.

5113.317 -> Why?

5114.387 -> Well it turns out that high performance

5115.977 -> computing is key to this new era of airplane design.

5120.437 -> And AWS is the leading cloud provider,

5123.357 -> allowing us to leverage a wide range of capabilities and services.

5128.057 -> Plus, Amazon’s relentless focus on the customer

5131.287 -> means that some of the best minds in cloud

5133.617 -> computing are helping Boom innovate and get to market faster.

5138.957 -> AWS levels the playing field in aerospace,

5142.467 -> allowing a startup company to develop

5145.127 -> what previously only big companies or governments could do.

5150.447 -> So this is how airplanes were designed before the age of computing.

5154.347 -> Engineers worked with drafting paper and slide rules.

5157.967 -> They built scale models to test in wind tunnels

5160.747 -> leading to a process of iteration that was slow and costly.

5165.537 -> Today, to design faster airplanes,

5168.197 -> we need the fastest computers. And computational methods leveraging AWS

5173.187 -> save us literally years of schedule and millions of dollars.

5177.397 -> Moreover, because we can now test many designs quickly

5181.667 -> and inexpensively, we can deliver a better airplane.

5186.697 -> In October, we rolled out XB-1, history’s first

5190.467 -> independently developed supersonic jet.

5193.527 -> To design XB-1, we leveraged EC2 to stand up HPC clusters

5198.927 -> often with more than 500 cores - hundreds of possible airplane

5203.657 -> designs flew through virtual wind tunnel

5206.117 -> tests encompassing thousands of flight scenarios.

5210.187 -> Because AWS allowed us to run many hundreds of these simulations

5214.347 -> concurrently,

5215.477 -> we achieved a sixfold increase in team product productivity.

5219.877 -> Simply put, without AWS today

5223.257 -> we would be looking at a sketch of a future airplane concept,

5226.797 -> not an assembled jet because years of design work would still remain.

5232.367 -> Since airplanes are amongst the most complex machines

5235.677 -> ever created by humanity, Boom will generate petabytes of data

5240.347 -> as we design and develop our Overture airliner.

5243.957 -> Already we are transferring 525 terabytes of XB-1 design

5249.397 -> and test data to AWS.

5252.047 -> Because they let us put compute next to data,

5255.587 -> we can run models across our dataset gaining actionable insights.

5260.267 -> For example, we’re using machine learning to calibrate simulations

5264.157 -> to wind tunnel results, accelerating model convergence

5267.727 -> and allowing us to deliver a more optimized aircraft.

5271.087 -> All in all, we have used 53 million core hours in AWS

5275.657 -> to design and test XB-1.

5278.627 -> And in the same manner we expect to use over 100 million core hours

5283.187 -> as we finalize the design of our Overture airliner.

5286.687 -> Because our pursuit of speed is about making earth more accessible,

5291.987 -> we’re taking great care to build an environmentally

5295.527 -> and socially responsible supersonic jet,

5298.417 -> and I am proud that Overture will be 100% carbon neutral from day

5304.447 -> one thanks to its use of alternative fuels,

5307.977 -> which means that supersonic flight

5309.847 -> is going to be more affordable than ever before.

5314.387 -> Well, great revolutions and fundamental technologies

5318.687 -> enable benefits that are difficult to predict or even imagine.

5323.717 -> Think of how many industries AWS

5326.357 -> has already transformed, and today supersonic flight

5330.707 -> is one of those surprising benefits of computing.

5334.437 -> Just as AWS is reinventing computing, at Boom we are reinventing travel.

5340.357 -> So what further breakthroughs will be sparked

5342.827 -> by a revolution on how we fly?

5345.457 -> At Boom our vision is one of accessibility,

5348.677 -> of new possibilities unlocked in the world around us.

5352.847 -> By the end of the decade your flights will be cut in half.

5357.257 -> So, what will you do when Australia is as accessible as Hawaii is today?

5362.287 -> And your flight is completely carbon neutral?

5365.077 -> How might you transform an industry?

5367.477 -> What new people could you come to call friends?

5370.637 -> But Overture is merely the first step

5373.517 -> towards our vision of a supersonic future.

5376.177 -> Because I dream of a day where you can go anywhere in the world

5379.877 -> in four hours for just $100.

5382.567 -> Where the fastest flight is also the most affordable.

5386.817 -> Boom is making supersonic flight mainstream and AWS

5391.117 -> is helping us deliver on that promise.

5393.857 -> I am so excited to see what you will invent

5397.217 -> when more of the world is within your reach.

5399.857 -> Thank you.

5400.74 -> [applause]

5405.727 -> Thank you, Blake. I’ll tell you what I'm going to do

5408.387 -> when Australia is as accessible as Hawaii –

5411.587 -> I’m going to go to Australia more. It’s incredible what Boom is doing.

5417.657 -> We often talk at Amazon about thinking big.

5420.917 -> That is thinking big. It’s very impressive, it’s very exciting.

5424.607 -> It could really change life for all of us

5427.217 -> and we’re honored to be partnering with Boom.

5430.507 -> You know, I remember a time about 25 years ago or so,

5434.017 -> people thought that search was boring,

5436.347 -> and there was really nothing left to invent.

5439.007 -> And I remember a time maybe 15 years ago

5441.587 -> where people felt like technology and infrastructure was boring,

5445.367 -> and there wasn't much left to invent and those turned out to be wrong.

5449.657 -> And I think oftentimes people think of data stores

5453.617 -> and databases as being boring and what is there left to invent.

5458.297 -> And I hope you can see the answer’s a lot.

5461.327 -> You see that block stores are being reinvented.

5464.187 -> You see that relational databases are being reinvented.

5467.237 -> You see that purpose-built data stores are being reinvented.

5470.367 -> And then the movement of that data between those stores that frees up

5474.557 -> the power of what you can build with Elastic Views…

5478.147 -> huge amounts of reinvention happening with data stores.

5481.467 -> Now hand-in-hand with data is what has been happening

5484.477 -> with machine learning and the adoption of it.

5486.657 -> And to start us off on this section I’m going to transition it over

5490.047 -> to the CTO of Intuit, Marianna Tessel.

5493.922 -> [music playing]

5498.487 -> Intuit is an early adopter of AI,

5500.467 -> starting the journey over a decade ago.

5502.747 -> We see tremendous opportunity in applying AI

5505.407 -> to revolutionize our business and to benefit our customers.

5508.657 -> And given the potential, this is just the beginning.

5511.707 -> Working together with AWS, we developed a robust ML platform

5516.497 -> that empowers our engineers to incorporate AI into our products.

5520.947 -> We use AWS tools for model development, training,

5523.917 -> and hosting, and integrate our own capabilities

5526.377 -> for orchestration and feature engineering.

5528.827 -> This has been game changing.

5530.377 -> Together with AWS we made great strides

5532.847 -> in driving AI/ML innovation with speed,

5536.367 -> helping us deliver smarter products

5538.287 -> faster to more than 50 million consumers,

5541.557 -> small businesses, and self-employed customers around the world.

5546.031 -> [applause]

5550.647 -> Thank you, Marianna.

5551.947 -> It is really interesting when we talked about earlier

5556.447 -> about being able to build the reinvention culture.

5559.897 -> Intuit has done that. And you can see it,

5562.347 -> not only with how they have been more sophisticated

5565.847 -> and more leaning forward with respect to using the cloud

5568.277 -> for their technology infrastructure,

5570.247 -> but also with respect to what they are doing with machine

5573.127 -> learning where they’re well ahead of most companies.

5575.877 -> So people have been talking about machine learning for over 20 years.

5579.607 -> In the cloud, with cost structure around compute

5582.987 -> and then the amazing amount of capacity you have,

5585.357 -> has made machine learning much more practical.

5588.317 -> And while more and more companies -

5590.517 -> amazing progress - have started using machine learning,

5594.347 -> make no mistake about it,

5595.737 -> it is very early in the history of machine learning

5600.187 -> and almost everything is continually being reinvented.

5603.997 -> There’s loads of examples.

5605.187 -> I’ll just give you one simple one here.

5607.317 -> So this is the top frameworks that have been used in new machine

5611.887 -> learning scientific publications in the last five years,

5614.487 -> which is often a leading indicator of what people use.

5617.357 -> And you heard us talk a few years ago

5619.907 -> when TensorFlow was so dominant in the frameworks

5622.327 -> used that the one constant we see in machine learning is change.

5627.077 -> And you can see that.

5628.317 -> Look at how PyTorch has caught TensorFlow over the last few years,

5632.597 -> because it’s much easier to use and to get started with.

5635.617 -> And when we talk to machine learning practitioners,

5639.857 -> 90% of them tell us they use more than one framework

5642.767 -> and 60% use more than two.

5645.577 -> And so clearly you can see that machine learning

5650.477 -> is in the very early stages.

5652.267 -> Frameworks was one of the most stable things a few years ago,

5655.567 -> but you can see it’s changing very substantially.

5657.867 -> Now we have a lot more machine learning capability

5660.797 -> than anybody else by a fair bit,

5662.487 -> and it’s one of the reasons why last year

5664.457 -> in my keynote the section on machine learning was 75 minutes.

5668.917 -> I’m not going to do that this year. For the first time, we have broken out

5672.047 -> a machine learning keynote that Swami’s going to do next Tuesday.

5676.087 -> I’m going to leave a lot of the goodies for him.

5677.777 -> But I have a few that I’m going to share

5679.727 -> and I’m going to really frame them in the context

5682.357 -> of the asks we get from our machine learning customers.

5685.877 -> The first ask we get is they say,

5688.037 -> “Look, we want to have the right tools for expert machine

5691.687 -> learning practitioners,”

5692.917 -> and that typically involves chips and frameworks

5697.247 -> and this is the group of people

5698.547 -> by the way who operate at that bottom layer of the machine learning stack.

5703.497 -> They’re comfortable building and training

5705.097 -> and tuning and deploying machine learning models of scale.

5708.317 -> And we talked earlier about all the areas

5710.657 -> we’re trying to help our expert machine

5712.777 -> learning practitioners with chips around inference

5715.667 -> and machine learning training,

5717.777 -> but I also think it’s important to think about the frameworks.

5721.697 -> We took an approach a few years ago

5723.767 -> where we said we’re going to support every single one of the major machine

5728.737 -> learning frameworks and we built… that was a unique approach.

5731.787 -> Everybody else was just focused on TensorFlow.

5734.497 -> And what we did was we built separable teams.

5737.397 -> We have one team that’s just focused on how to optimize TensorFlow in AWS,

5742.007 -> one team that’s just optimized on how to run PyTorch in AWS,

5746.397 -> and one that’s just focused on MXNet in AWS.

5749.457 -> And that’s why you get the best performance

5751.677 -> in all those frameworks on AWS.

5753.907 -> I usually show you benchmarks at this point.

5755.837 -> I’m going to leave that to Swami for next week.

5758.287 -> But we believe that we’re not done seeing new frameworks pop up

5764.247 -> that you’re going to care about.

5765.747 -> And the commitment you have from us is that we will continue

5771.767 -> to support every single one of those major frameworks

5774.257 -> because we know that expert machine learning practitioners

5777.937 -> want the flexibility to build however they see fit.

5782.887 -> The second ask we get is,

5784.597 -> “Look, the reality is there aren’t that many expert machine

5788.117 -> learning practitioners in the cloud

5790.317 -> and there aren’t that many in the world,

5792.117 -> and those that exist tend to live at the big technology companies.”

5795.467 -> So, if we want machine learning to be as expansive

5799.727 -> as we all believe it should be,

5802.007 -> you’ve got to make it easier for everyday developers

5805.197 -> and data scientists to be able to use machine learning.

5807.157 -> And that’s why we built a few years ago

5810.237 -> SageMaker, which sits at that middle layer of the machine learning stack.

5814.287 -> And SageMaker was a step-level change in the ease

5818.207 -> with which you can build, train,

5819.877 -> tune, and deploy a machine learning model.

5822.167 -> We have tens of thousands of customers

5824.707 -> who are standardizing on top of SageMaker and using it.

5828.547 -> And these are companies like 3M and ADP and Cerner

5833.187 -> and Intuit and GE and Snap and the NFL

5838.127 -> and T-Mobile and Vanguard, just a broad group.

5841.257 -> And one of the things that people like most about SageMaker

5846.027 -> is they see how quickly SageMaker’s continuing to iterate.

5849.867 -> This is the second year in a row that the SageMaker team has added

5853.187 -> over 50 new features in the last 12 months.

5855.807 -> That’s like one a week. That’s really unusual.

5858.637 -> And we haven’t stopped.

5860.047 -> If you think about last year at re:Invent,

5862.967 -> I announced in my keynote the launch of SageMaker Studio

5866.037 -> which was the first Integrated Development

5867.897 -> Environment, or IDE, for machine learning.

5870.357 -> And we gave you Notebooks,

5872.297 -> which were easy to create in one click and share. And Debugger,

5875.787 -> which made it easy to debug your models.

5877.737 -> And Experiments, which saved all the artifacts of your experiments

5881.147 -> so you could kind of figure out what happened and share them.

5883.597 -> And Model Monitor, so you could tell if you had model drift, and Autopilot

5887.887 -> which looked at automatically what was in a CSV file

5890.657 -> and created a new model for you,

5892.497 -> machine learning model automatically with transparency

5894.767 -> and how it was created, so you could pick it up

5896.907 -> when you were done seeing it, if you wanted to evolve it yourself.

5899.827 -> That was a huge amount of new innovation last year

5903.957 -> and customers have loved SageMaker Studio.

5907.257 -> And they’re using it in a really expansive way.

5910.487 -> And what we often do,

5912.597 -> again we have services that have this amount of traction.

5916.507 -> We get a lot of gratuitous feedback from customers which we love,

5921.227 -> but we also constantly ask customers,

5923.857 -> “What else can we build for you that would make your life easier?”

5927.907 -> And the topic that seems to come up first and foremost

5931.937 -> almost every time is,

5934.097 -> “How can you make doing data preparation for machine

5938.027 -> learning much easier?”

5939.307 -> And data preparation is hard. If you think about it,

5942.747 -> to build a model you’ve got to collect all this data

5945.167 -> from different sources that come in different formats.

5947.937 -> None of it’s normalized which you need for the models.

5950.307 -> And when you’re building these models,

5952.067 -> you need to look at all these different dimensions,

5954.007 -> what they call machine learning features.

5957.477 -> I’ll take something like a real estate app.

5959.557 -> If you want to predict the prices of real estate,

5963.527 -> you need to look at features when you build a model -

5965.537 -> like how many bedrooms

5967.217 -> or how many bathrooms or how many square feet

5969.437 -> is the house or what are other houses on that street selling for?

5973.127 -> Does it have a Starbucks within five miles?

5975.977 -> These are all features that you need that arrive in different formats

5979.687 -> that the model cannot understand,

5982.037 -> where you have to convert these features into the right format

5985.687 -> that the model can understand.

5987.027 -> This is what is called feature engineering.

5989.727 -> And then there are also, by the way, times

5992.297 -> where you want to take two different features

5994.537 -> and combine them, or two or more different features and combine them.

5997.867 -> Take something like, in a real estate example,

6000.377 -> you want a house pricing index where you’re combining different features

6004.187 -> that make the model more efficient.

6006.217 -> Converting these models, converting these features,

6010.527 -> combining features themselves –

6012.317 -> that type of feature engineering is really hard

6015.147 -> and it takes a lot of time. You’ve got to write queries and code

6017.657 -> to get all that data from various sources.

6020.097 -> You have to convert the data to the format the algorithms can use.

6023.897 -> You sometimes want to actually combine the features.

6026.907 -> You want to prototype these features to see

6028.927 -> if your feature engineering worked

6031.427 -> or not before you actually apply those transformations everywhere.

6034.387 -> And then you’ve got to apply the transformations

6036.837 -> and make sure you don't have missing data or outliers.

6039.067 -> It's just a lot of work.

6041.177 -> And people said there must be an easier way,

6044.227 -> which is why I’m excited to announce today

6045.887 -> the launch of Amazon SageMaker Data Wrangler,

6048.857 -> which is the fastest way to prepare data for machine learning.

6052.75 -> [applause]

6059.027 -> So Data Wrangler is a total game changer

6062.257 -> with respect to speed of doing machine learning data preparation.

6066.107 -> The way it works is you point Data Wrangler

6068.447 -> at the appropriate AWS data store or third-party data store

6072.387 -> and then Data Wrangler has over 300 built-in conversions

6076.517 -> and transformations that, through machine

6078.987 -> Learning, will automatically recognize the type of data coming in

6082.497 -> and suggest the right transformation for you to make that you can apply.

6086.037 -> You can of course do your own thing as well.

6088.797 -> It also makes it much easier in the Data Wrangler console

6092.437 -> to combine features to build those composite features I was mentioning.

6096.637 -> You can preview very easily these transformations in SageMaker Studio,

6101.377 -> and then if you like what you see, you simply apply that transformation

6105.787 -> to the whole dataset and Data Wrangler

6108.527 -> manages all the infrastructure and all that work under the covers.

6111.947 -> It’s a total game changer in the time

6114.867 -> that it takes to do data preparation for machine learning.

6118.667 -> Now, because you spend so much time on the data prep

6121.987 -> as well as on the feature engineering,

6124.017 -> not surprisingly you want to use these features

6126.997 -> in lots of other models,

6128.547 -> so you need a place to store these features

6130.897 -> and make them easily accessible.

6132.227 -> And this turns out to be a hard problem.

6134.417 -> Sometimes you might step back and say,

6135.927 -> “Well why is this a hard problem? Why don't you just store them in S3?”

6138.697 -> The problem is, you may be able to do that if you wanted to store

6143.457 -> as an object maybe a simple set of features

6147.517 -> that are mapped to one model,

6149.057 -> but features are hardly ever mapped to just one model.

6152.427 -> They’re usually mapped to lots of models because they are highly useful,

6155.377 -> and you did all the work to get the features in a state

6157.807 -> where the model can understand them.

6159.657 -> And then sometimes you’ve got subsets in that set of features

6163.267 -> that want to be their own sets of features.

6166.057 -> And then you’ve got multiple people who want to access those features

6169.707 -> and the different sets with multiple models,

6171.537 -> and pretty quickly it becomes complicated

6174.607 -> and hard to keep track of,

6175.837 -> which is why people often try to keep track of this

6178.697 -> in an email or spreadsheets or sometimes build a clunky UI

6182.257 -> that takes time to really work very well.

6185.587 -> You also need to… these same features need to be used to train models

6191.647 -> and then also to make predictions.

6194.457 -> And they’re really different use cases.

6196.497 -> When you’re training a model,

6198.497 -> you’ll use all the data in a particular feature

6201.957 -> to be able to get the best model possible to make predictions.

6205.637 -> But when you're making predictions, you often will take

6208.337 -> just maybe the last five data points in that particular feature,

6212.037 -> but they have to be stored the same way

6213.457 -> and they’ve got to be accessible.

6215.327 -> When you’re actually using these features for inference

6218.557 -> and predictions, you need really low latency

6220.587 -> because you're trying to make predictions

6222.157 -> in close to real time in your applications.

6224.007 -> So it’s really hard to do all these things in a generalized store.

6229.397 -> And so that’s why we have built and I’m excited to announce today

6233.147 -> SageMaker Feature Store, which is a new repository

6236.117 -> that makes it easier to store, update, and share machine learning features.

6240.453 -> [applause]

6246.187 -> And so the Feature Store,

6248.047 -> it is a purpose-built feature store that’s accessible in SageMaker Studio

6253.067 -> and it makes it much simpler to name, organize, find,

6256.837 -> and share features with teams

6259.277 -> because we’ve built a purpose-built store for features.

6263.227 -> It turns out it makes it really easy for features

6267.617 -> to actually be accessed either for training or for inference.

6271.347 -> Even though they’re different use cases,

6272.887 -> we’ve built a store that makes that simple.

6274.857 -> And because Feature Store is located in SageMaker

6279.247 -> close to where your machine learning models are running,

6282.057 -> you get really low latency on your inference and prediction.

6285.207 -> So this is another big deal for people

6287.637 -> that are trying to build machine learning models.

6290.357 -> Now I think you can tell from understanding what happens in machine

6294.247 -> learning or looking at data preparation or Feature Store,

6297.857 -> that machine learning has a lot of things that have to happen

6302.027 -> sequentially or sometimes in parallel,

6304.427 -> but really lend themselves well to orchestration and to automation.

6309.337 -> And this is true in normal code

6312.537 -> as well when you’re building applications.

6314.107 -> It’s why they’ve built these CI/CD pipelines.

6317.917 -> But in machine learning there is no CI/CD.

6321.657 -> None of them exist pervasively. People have tried to build their own.

6325.957 -> They’ve tried to do it homegrown. It hasn’t been that scalable,

6329.927 -> it hasn’t worked the way they wanted it to.

6332.177 -> And customers want an easier way to do this.

6334.727 -> And so I’m excited to announce today

6336.907 -> the launch of Amazon SageMaker Pipelines,

6339.267 -> which is the first purpose-built easy-to-use CI/CD

6342.627 -> service for machine learning.

6345.024 -> [applause]

6351.477 -> And so with Pipelines you can quickly create machine

6355.177 -> learning workflows with our Python SDK.

6358.447 -> And then you can automate a bunch of the steps

6361.237 -> from a number of the things

6363.067 -> you have to do on data preparation in Data Wrangler -

6365.547 -> to moving the data from Data Wrangler to the Feature Store

6368.187 -> to some of the activities you want to take once in the Feature Store,

6372.187 -> to training, to tuning, to hosting your models.

6375.517 -> You can build all of these things in a workflow

6378.887 -> that happens automatically.

6380.877 -> And then Pipelines manages all of those dependencies

6384.147 -> between each of the steps and orchestrates the flow

6386.967 -> so you can have any number of these automated steps

6390.557 -> in a Pipeline workflow.

6392.187 -> We also give you Pipelines with preconfigured templates

6396.117 -> for building and deploying

6398.147 -> so that you don't have to create these from scratch.

6400.427 -> You can either use those verbatim, you can use those as a base

6403.827 -> and then customize how you see fit,

6405.547 -> or you can actually build them from scratch.

6407.097 -> You can do all of those things in Pipelines.

6409.327 -> And then Pipelines automatically keeps track of all of the artifacts

6413.067 -> and then keeps an audit trail of all the changes

6415.647 -> so it's easy to troubleshoot if you need to do that.

6418.777 -> So SageMaker has completely changed the game

6422.807 -> for everyday developers and data scientists in being able to build,

6427.347 -> train, tune, and deploy machine learning models.

6430.497 -> And people have flocked to SageMaker,

6434.467 -> not only because there's nothing else like it,

6437.207 -> but also because of the relentless iteration and innovation

6441.607 -> that you continue to see us applying into SageMaker.

6444.137 -> It’s not just the launch of the service.

6446.147 -> You can look at what we did with SageMaker Studio last year.

6448.997 -> You can look at what we’ve done this year,

6450.957 -> at least what you’ve heard so far – there’ll be more.

6453.437 -> And what we’ve done around data preparation and Feature Stores

6456.367 -> and the first CI/CD in machine learning.

6459.237 -> We’re not close to being done innovating here.

6463.217 -> The third ask that we often get is customers say,

6468.107 -> “I would like to be able to use your machine learning,

6471.007 -> but I don't want to have to be responsible

6472.907 -> for creating the model myself. I want to send you data.

6477.127 -> I want to use models that you have built and trained on data

6480.827 -> and get the answers back via APIs.”

6483.477 -> And that’s what we think of as this top layer

6486.287 -> of the machine learning stack. We’ve a lot of services in this area.

6490.067 -> For people that want to look at an object and say,

6491.927 -> “What’s in this?” or a video and say, “What’s in this?” we have an object -

6496.497 -> a video recognition service - called Rekognition.

6499.197 -> We have services that allow you to go from text to speech,

6503.447 -> to transcribe audio to text, to translate that transcribed audio,

6509.917 -> to do natural language processing on all that translated,

6512.867 -> transcribed audio so you don't have to read it all

6514.907 -> and know what’s happening. It lets you do OCR, but also OCR++

6519.697 -> and being able to pull out data from tables and formats

6522.307 -> that don’t usually come out in OCR.

6525.117 -> We give you the ability to build chatbots with Lex.

6529.017 -> We give you the ability to do internal enterprise

6531.047 -> search with Kendra, which we launched last year.

6533.167 -> There was a bunch of Amazon capabilities that you asked us

6537.147 -> to expose as machine learning services which we have.

6540.517 -> So we have a personalization machine learning service

6542.867 -> and a forecasting service and a fraud detection service,

6546.607 -> and a code inspection service called CodeGuru.

6550.617 -> And it’s interesting.

6552.827 -> As you start using these models more and more,

6556.397 -> there's that point in every company where you think about,

6560.737 -> “How much can I trust these predictions

6563.387 -> and these answers to use in my applications?”

6566.767 -> And there are certain use cases where the ramifications are low enough

6570.737 -> that you can take more chances.

6572.047 -> Oftentimes in applications that are doing translation or transcription,

6576.467 -> if you see good results early on,

6578.487 -> you’re going to really trust it as your major input.

6580.867 -> But there are other use cases where the ramifications of getting it wrong

6585.717 -> are significant enough where you’re going to wait a little bit,

6588.887 -> and these are services like facial recognition

6592.197 -> or forecasting your inventory or the quality of your code.

6596.597 -> You’re going to look and make sure that you like the predictions,

6600.177 -> that they’re actually high quality, and you’re going to use those inputs

6605.377 -> really as one input in several inputs of a decision.

6608.517 -> So it’s really machine aided in that case.

6610.747 -> And what you’re going to wait for

6612.067 -> is you’re going to wait for the predictions

6613.957 -> to become more and more commonplace that you say,

6616.827 -> “Yes, that’s the answer,” and you want to see that

6619.597 -> that provider is going to continue to invest in that service

6623.387 -> so you know it’s going to be robust and you can really rely on it

6626.567 -> and cut the cord from what you were doing before.

6628.227 -> Let’s look at an example of this. Let’s look at CodeGuru.

6632.597 -> CodeGuru is a service we launched last year,

6635.327 -> which is a machine learning service

6636.637 -> that lets you automate code reviews to tell

6638.837 -> if you have any kind of code that you’ve written

6641.487 -> that we think is going to lead to a problem.

6643.387 -> And it also allows you to identify your most expensive line of code.

6647.347 -> By the way, we have over 120,000 apps at Amazon

6652.327 -> that are using that Profiler part

6654.107 -> to find the most expensive line of code,

6656.397 -> which not only is helping them find operational bottlenecks,

6659.547 -> but also saving them a significant amount of money.

6663.277 -> And so customers who are trying to assess,

6665.967 -> “Can I use… can I rely on CodeGuru as a major

6669.737 -> input in how I think about my code?”

6672.797 -> they say, “Well what’s your commitment?

6676.397 -> I understand you’ve got a profiler or you have a code reviewer,

6678.487 -> but how about more programming languages?”

6680.447 -> So the team launched Python and a few other languages.

6682.717 -> Then they said, “Well how about something around security?”

6685.877 -> So the team launched a security detector feature

6688.417 -> that lets you know whether or not you have code

6690.527 -> that would lead to some kind of security issue.

6693.317 -> And then people said, “Well, that’s great.

6696.127 -> You have so much information about the way that my application operates

6701.217 -> and the way that all kinds of applications operate on AWS.

6705.427 -> Why can’t you build a service using machine learning

6709.027 -> that allows us to predict

6711.217 -> when there are going to be operational problems?”

6713.067 -> And so we went away and thought about that.

6714.967 -> And I’m excited to launch a new service today

6717.677 -> called Amazon DevOps Guru,

6719.467 -> which is a new service that uses machine learning

6722.187 -> to identify operational issues long before they impact customers.

6726.727 -> [applause]

6732.587 -> And so DevOps Guru makes it much easier for developers

6737.237 -> to anticipate operational issues before it bites them.

6740.657 -> And so using machine learning informed

6743.357 -> by years of Amazon and AWS operational experience in code,

6748.197 -> what we do in the service is we identify potential issues

6750.987 -> with missing or misconfigured alarms or things that are…

6755.107 -> resources that are approaching resource limits or big changes

6758.777 -> that could cause outages or under-provision capacity

6762.427 -> or overutilization of databases or memory leaks.

6767.987 -> When we find one of those that we think could lead to a problem,

6771.477 -> we will notify you either by an SNS notification

6775.067 -> or a CloudWatch event or third-party tools like Slack.

6778.297 -> And we’ll also recommend the remediation.

6781.297 -> And so when you look at these top-level services

6785.027 -> and you find them super useful, as you go through this journey

6788.917 -> that you will go through as a company in determining

6792.817 -> when you can trust them as either the major

6796.217 -> or a major input in your applications,

6799.037 -> you’re always going to want to get

6800.497 -> a lot of experience running them at scale

6802.457 -> and see the predictions being more and more accurate,

6805.167 -> and then also making sure that the provider that you’re betting on

6808.657 -> for that service is continuing to invest in that service

6812.337 -> and make it easier and easier for you to use that as that input.

6815.277 -> And I think that’s what people are starting to see with CodeGuru.

6820.767 -> The fourth ask that we get is from customers who say the following:

6825.837 -> “It is awesome that you have so much machine

6828.407 -> learning capability, more than I could find elsewhere,

6830.517 -> but I would like to somehow benefit from machine learning

6834.707 -> without having to even know I’m doing machine learning.”

6838.137 -> Perhaps my favorite business school professor,

6841.407 -> and also perhaps the best innovation writer of our time,

6845.157 -> was a fellow named Clay Christensen - and Clay just passed away.

6848.417 -> But what Clay used to always talk about was he used to always say

6853.637 -> that customers hire products and services for a job.

6858.197 -> And in this case what customers are telling us is,

6861.227 -> “I don’t want to hire you for machine learning.

6864.017 -> I want to hire you for a particular job

6867.007 -> and if you use machine learning to get it done,

6869.317 -> great, that’s fine either way.

6871.277 -> But I want to hire you to get that job done.”

6874.187 -> And let me give you a good example of that.

6876.657 -> If you look at business intelligence, or BI, of course it started

6881.177 -> with these kinds of older forms of this, like Oracle OBIEE,

6885.127 -> and then really over the last few years

6886.687 -> Tableau has done a great job reinventing this space

6890.007 -> and building beautiful visualizations and dashboards.

6893.197 -> And we’re now at a stage where BI is being reinvented again.

6896.207 -> And what people really want - is they really want

6899.797 -> their BI service to be serverless,

6901.737 -> so they don’t have to manage all that infrastructure.

6904.537 -> They want the ability to be able to embed all their dashboards

6909.077 -> and visualizations everywhere easily, flexibly.

6912.997 -> And then increasingly, they want to be able to use all kinds

6917.717 -> of machine learning and natural language.

6919.587 -> Really what they want is natural language,

6921.497 -> whether machine learning fuels it or not,

6923.947 -> that allow them to understand the results more easily

6927.507 -> than having to do all the work to glean the insights themselves.

6929.957 -> And that’s why we built Amazon QuickSight back in 2016.

6934.837 -> QuickSight is serverless, it’s very scalable.

6939.127 -> It has, not just all the features you need in a BI service,

6943.587 -> but it also does embedding better than you’ll find elsewhere.

6946.767 -> And then we’ve started to invest in machine

6949.717 -> learning that will allow you to have experiences

6952.717 -> that get things done in a much more natural way.

6955.697 -> So for instance, we have this feature in QuickSight called Autonarratives

6960.137 -> and what it means is that when you actually do a query

6962.647 -> and you get the results,

6965.847 -> you don’t have to actually always do the work

6967.967 -> to figure out what the results mean.

6970.237 -> We will provide natural language what we think the key insight is.

6975.017 -> Now we do that through machine learning, but nobody who uses

6978.287 -> QuickSight has to know anything about machine learning.

6980.837 -> And customers really love that feature.

6983.497 -> And they’ve said, “I wish you could use that type of capability

6987.777 -> in other things that we have to do in our BI usage.”

6991.107 -> So for instance, I don’t like that our users have to know

6995.877 -> which databases

6997.267 -> or data stores to get the data from, or how the data’s structured

7000.197 -> or how to ask a question in a certain way or…

7003.747 -> I want them just to be able to type into a search bar

7007.937 -> a natural language question and get answers.

7010.347 -> And this is a hard problem.

7011.427 -> A lot of companies have tried to take a shot at solving this,

7014.587 -> usually with natural language query,

7016.177 -> and they just haven’t been able to get good results.

7018.717 -> And it’s really hard to build the right dataset

7020.817 -> because there’s millions of questions and pretty quickly you find

7024.767 -> you didn’t have all the questions. You’ve got to go back to IT

7027.217 -> and they have to build it when they get a chance.

7028.917 -> It just hasn’t worked out until now.

7031.637 -> So I’m excited to announce the launch of Amazon QuickSight Q,

7035.557 -> which allows you to ask Q any question in natural language

7038.677 -> and get answers in seconds.

7040.789 -> [applause]

7047.447 -> And so the way Q works is you just type in a question

7050.907 -> in natural language in Q in the search bar.

7054.507 -> You can ask a question like,

7056.057 -> “Give me the trailing 12 month sales of product X,” or,

7059.167 -> “Compare product X sales to product Y sales the last 12 weeks,”

7064.457 -> and then you get an answer in seconds.

7066.797 -> That’s all you have to do. You don’t have to know the tables,

7068.847 -> you don’t have to know the data stores.

7070.537 -> You don’t have to figure out how to ask the question in the right way.

7072.727 -> You just ask it the way you speak.

7075.197 -> And Q will do auto filling and spellchecking

7077.757 -> so that it makes it even easier to get those queries written.

7080.867 -> Q uses very sophisticated deep learning,

7085.137 -> natural language processing, schema understanding, and semantic

7088.797 -> parsing of SQL code generation. It is not simple.

7094.347 -> However, customers don’t need to know anything about that machine learning.

7099.807 -> For them, the experience is: I type in natural words

7104.807 -> the way I would speak and I get answers in seconds.

7108.017 -> By the way, we’ve trained those models over millions of datapoints

7111.187 -> and we’ve also trained them over all these different vertical domains,

7114.207 -> virtually every one you can imagine.

7116.137 -> This is going to completely change the BI experience.

7118.927 -> So again, customers didn’t hire us to do machine learning here.

7124.667 -> They hired us to make it easier to ask the questions that

7127.637 -> they want business intelligence on, to get answers quickly,

7130.687 -> whether we use machine learning or not -

7132.537 -> and that’s what we try to do here with Q.

7136.867 -> So it’s pretty amazing how fast machine learning

7140.057 -> is continuing to evolve.

7141.277 -> And although we have over 100,000 customers

7144.317 -> who are using our machine learning services,

7146.187 -> a lot more than you’ll find anywhere else,

7148.397 -> and a much broader array of machine

7150.387 -> learning services than you’ll find elsewhere,

7153.047 -> we understand that it is still very early days in machine learning

7157.817 -> and we have a lot to invent in all four of these areas

7161.307 -> that we get asked regularly from, from customers.

7164.537 -> Now we’ve talked about the reinvention of compute

7169.927 -> and the reinvention of data stores

7172.467 -> and the reinvention of machine learning,

7175.017 -> and I think that if you look more broadly,

7178.427 -> if you’re wondering are horizontal application areas

7182.977 -> being reinvented or are vertical industry segments being reinvented,

7188.267 -> the answer is absolutely yes. And to kick us off on this section

7192.737 -> and to share how they’re reinventing broadcasting,

7195.147 -> it’s my privilege to bring to the stage

7197.677 -> the CTO and President of Digital for Fox, Paul Cheesbrough.

7202.297 -> [music playing]

7205.917 -> Through our partnership with AWS,

7207.657 -> we’ve combined best-of-breed infrastructure

7209.937 -> with leading-edge media operations.

7212.497 -> Production teams who are more distributed than ever before

7215.977 -> will be able to deliver uncompressed video

7218.497 -> through and from the cloud with full redundancy,

7221.937 -> which is an industry first.

7223.527 -> We’ll be able to produce live events with less latency,

7226.537 -> increased reliability, and more efficiently than ever before.

7230.247 -> This collaboration has resulted

7231.807 -> in the development of entirely new services.

7234.457 -> Working with AWS, we’ve reinvented how we produce and distribute content

7239.117 -> to our consumers across all platforms and devices using the cloud.

7243.847 -> We’ve transformed our existing operation

7246.057 -> but, more importantly, we’ve laid the tracks for the future

7249.547 -> in a way that will help us innovate and adapt for many years to come.

7254.532 -> [applause]

7259.413 -> Thank you, Paul. We love working with Paul and his team.

7262.847 -> They are unafraid of leaning forward and inventing.

7265.777 -> They have built a real reinvention culture,

7268.587 -> and it’s remarkable to see the way

7270.767 -> that they are reinventing broadcasting as we speak.

7273.387 -> Very honored to be partnered with them.

7275.447 -> One of the questions that we get asked

7277.677 -> most from venture capital companies and private equity companies is,

7281.797 -> “Which areas do you think are going to be reinvented?

7284.567 -> Which application areas?

7285.747 -> Which vertical business segments? What’s going to get reinvented?”

7288.977 -> And the answer we give is often unsatisfying,

7291.217 -> because in our opinion the answer is, “All of them”.

7294.197 -> And that’s true in every single area. I think you’re seeing it happen today

7299.357 -> and those that you’re not seeing it happen

7300.997 -> in yet, you will see it in the next few years.

7303.427 -> And so let’s start with horizontal application areas.

7305.667 -> And let’s start with an example. Let’s look at call centers.

7309.247 -> So if you look at call center solutions,

7312.647 -> the last 20-30 years people have not liked these solutions very much.

7317.007 -> They’ve been hard to set up, they required expensive consultants

7320.917 -> and lots of hours,

7322.307 -> if you need to change anything you require a complicated code,

7326.247 -> they were hard to scale up and down, they are expensive,

7329.747 -> and they are missing the two most transformational technology

7335.557 -> advances of the last 15 years, which is cloud and machine learning.

7340.927 -> Now, we learned this first-hand at Amazon,

7343.417 -> where we had to build our own call center solution

7345.737 -> to handle our retail business,

7347.107 -> and we had a lot of AWS customers who said,

7349.557 -> “Why don’t you just generalize that and expose it to us as a service?”

7353.447 -> And that’s why we built Amazon Connect,

7356.177 -> which is our call center service, which we built in 2017.

7360.167 -> And people love this service.

7362.517 -> It’s one of the fastest growing services in the history of AWS,

7366.487 -> and the reason they love it is that it’s easy to set up,

7372.497 -> it takes minutes instead of outside consultants

7375.777 -> and lots of money, it’s easy to scale up or down agents,

7379.927 -> it’s much more cost effective,

7382.477 -> because you only pay for what you consume,

7385.487 -> and you are only paying for when your agents

7387.747 -> are interacting with customers.

7389.977 -> There is no infrastructure for you to deploy,

7392.867 -> and then it’s built right from the get-go on top of the cloud

7397.357 -> and with a lot of machine learning and AI embedded,

7400.207 -> so you can do things like chat bots

7402.007 -> and IVR and transcription of audio and sentiment analysis

7405.787 -> without having to worry about the machine learning at all.

7408.507 -> And Connect is growing really rapidly.

7410.587 -> You see thousands and thousands of customers are using it,

7414.237 -> companies like Capital One and Intuit, John Hancock,

7417.597 -> and GE and Petco and Barclays and State Farm.

7421.577 -> And what’s also interesting is,

7422.987 -> during the pandemic over 5,000 new companies,

7427.957 -> new Connect customers, have started using the service

7430.697 -> where they spun up call centers remotely to help them deal

7434.207 -> with the fact that all their customer service agents were now remote.

7438.327 -> And so this is another one of those services

7440.927 -> that’s growing unbelievably quickly,

7443.487 -> and people are excited about and giving us all sorts of input

7447.407 -> on what else they would like to see us solve with the service.

7450.977 -> And so I am going to share a few of their asks across three categories

7454.237 -> and five different feature ideas.

7455.687 -> And the first is really around how can we make it easier

7460.517 -> for our customer service agents

7462.277 -> to have the right information about products and customers

7465.757 -> right in the moment, in real-time, when they are dealing with customers?

7468.547 -> And the problem is an interesting challenge.

7471.857 -> When customers call in about something, a service or product,

7475.737 -> you have all this information that lives

7477.517 -> in all these different databases,

7479.027 -> some on-premises and first-party and some of which are third-party,

7483.247 -> which you have your agents try to access and toggle

7486.707 -> between different databases and find the right information,

7489.517 -> which is slow or they don’t do it all.

7491.457 -> The same thing happens when you have customers

7494.907 -> who interact with you across lots of different silos

7497.567 -> in your business, and different databases,

7499.807 -> either first-party or third-party. And so customers said,

7503.577 -> “Can you make it much easier for us to have the right product information

7507.507 -> and customer information for our agents

7509.867 -> to have that fast holistic service?”

7511.817 -> So I have two new features to announce here for Connect

7515.007 -> to help with this problem. The first is Amazon Connect Wisdom,

7518.757 -> which is a new capability that uses machine

7520.687 -> learning to deliver agents the product and services information

7524.327 -> they need to solve issues in real-time.

7527.366 -> [applause]

7533.747 -> So Wisdom has all these built-in connectors

7536.937 -> to relevant knowledge repositories,

7538.817 -> either your own or there’s some third-party ones.

7541.197 -> We’ll start with having connectors to Salesforce and ServiceNow,

7544.937 -> but there will be more coming.

7546.737 -> And then what Wisdom lets you do is that, as a call is happening,

7551.707 -> Wisdom is using machine learning to listen

7553.997 -> to that call transcription from Contact Lens

7557.497 -> and then detect issues that are happening

7560.127 -> and put the right information in front of agents.

7562.457 -> So, for instance, if you have a call and Wisdom hears

7566.587 -> ‘arrived broken’, it will search all the relevant data repositories

7571.387 -> for the information that you need around what to do

7574.487 -> when you have a product that arrived broken.

7576.177 -> So the agents have it right there in front of them automatically,

7578.827 -> right in their console. That’s a game changer.

7582.147 -> And if it turns out that you don’t get the right recommendation,

7585.227 -> or you want to get more information,

7587.837 -> agents can also just type in natural language questions

7591.427 -> into the Wisdom search bar

7593.157 -> which pulls all the information across all those databases.

7595.937 -> So this totally changes what information

7599.007 -> you have available to you, product and service-wise, for your agents.

7602.577 -> And then, for trying to figure out how to

7604.187 -> provide a more holistic customer experience,

7607.747 -> and pull together all the customer profile information,

7611.127 -> I am excited to share a second feature,

7612.927 -> which is the launch of Amazon Connect Customer Profiles,

7616.477 -> which gives agents a unified profile of each customer

7619.927 -> to provide a more personalized service during a call.

7623.851 -> [applause]

7629.047 -> So here’s how Customer Profiles works.

7631.567 -> You will point Customers Profiles at your internal databases,

7636.727 -> where again, a third-party database will launch Customers Profiles

7640.107 -> with your being able to access your information in Salesforce,

7643.987 -> in Zendesk, ServiceNow, and Marketo. And then, when the call comes,

7649.987 -> Customer Profiles connects that phone number or contact ID

7654.037 -> with a customer ID

7656.017 -> that’s used consistently across all those data stores.

7660.207 -> And so, Customer Profiles knows how to ingest and normalize

7663.477 -> that data across customer contact history,

7665.687 -> or e-commerce status, or order management,

7668.637 -> or marketing communications you’ve sent,

7670.387 -> or sales or CRM information.

7673.307 -> And they know how to display it in a concerted organized way

7677.067 -> so all that appears in front of an agent’s screen.

7680.187 -> And so, take a simple example.

7681.767 -> Let’s say you are a hotel company that’s using Connect,

7684.527 -> and let’s say you have a customer call

7686.037 -> in who’s unhappy about a stay they had the prior night.

7689.437 -> You might handle that call differently

7692.387 -> if you also knew that that same customer

7695.537 -> called you earlier in the week for their company to ask for a quote

7699.387 -> to do a five-day off-site at that hotel.

7701.837 -> You might have a different conversation.

7704.157 -> Today, agents don’t have access to that information.

7708.137 -> With Customer Profiles,

7709.537 -> all that information is there for agents to use at the same time,

7713.317 -> and changes how they can have holistic customer interactions.

7716.917 -> Very, very useful.

7718.307 -> The next category of asks we get around Connect is,

7724.197 -> “How can I make it easier that when I have a customer contact

7729.237 -> or call that’s going off the rails,

7731.987 -> how can I intercept that before the call is done

7735.427 -> and before I have a bad experience for the brand?”

7738.917 -> And you may remember that last year we launched a service

7741.387 -> in Connect called Contact Lens, which is a call analytics service.

7745.047 -> And what Contact Lens does is it stores all your calls in S3,

7750.057 -> and then it transcribes it to text,

7752.237 -> and then it indexes and tags that information so it

7754.607 -> makes it easier to search and find,

7757.067 -> and then it gives people, agents or managers,

7760.907 -> the ability to have full-text transcriptions of those calls,

7765.037 -> to be able to understand when there was a negative sentiment,

7769.377 -> or when there were long lapses in conversation

7771.617 -> or people were raising their voice or talking over each other.

7774.277 -> And so it totally changes your knowledge and information

7777.767 -> about how these contacts went.

7779.547 -> And customers have loved Contact Lens,

7782.017 -> and they have used it very pervasively.

7784.287 -> But a number of our customers

7785.597 -> and the supervisors of these customers have said,

7787.957 -> “Look, I wish I had the chance to know in real-time

7791.987 -> when a call was going awry,

7793.327 -> because I could either coach that agent in real-time,

7796.097 -> or have it move to me to avoid any brand harm.”

7799.737 -> And so the team has worked on this for the last year,

7802.197 -> and I am excited to announce the launch of Real-Time Contact Lens,

7806.427 -> which identifies issues in real time

7808.897 -> to impact customer interactions during the call itself.

7812.258 -> [applause]

7817.695 -> And so Real-Time Contact Lens uses more developed machine

7821.997 -> learning to do natural language processing

7824.067 -> on audio calls in real time,

7826.577 -> instead of waiting a few minutes after the call ends

7829.507 -> and the data and the speech is transcribed to text.

7833.347 -> They do it in real time. This is not easy to do,

7837.687 -> which is why it took the team a year to do it.

7839.627 -> And it’s a good example of not only being able to leverage

7842.777 -> AWS’s machine learning expertise, but again, being able to use machine

7846.947 -> learning where it’s working for you under the covers,

7849.437 -> where you don’t actually have to worry about it at all.

7851.827 -> The job you’re trying to get done

7853.737 -> is you want to be able to have an impact on calls in real time.

7857.197 -> And so what managers do who are using Real-Time Contact Lens,

7861.077 -> is they specify terms where they want to be alerted.

7864.297 -> So they may say, “If, in the transcript I hear

7867.597 -> ‘bad experience’ or ‘unhappy’ or ‘never using you again’…”

7872.527 -> When that criteria is met,

7874.667 -> it sends an alert in the Contact Lens dashboard to supervisors

7878.927 -> and they can either choose to coach agents real-time,

7882.327 -> or have the call transferred to them.

7884.677 -> And if the call is transferred to them,

7886.667 -> they also get the real-time transcription of that call

7891.177 -> right in front of them,

7892.277 -> so they don’t have to ask customers all the same questions,

7895.717 -> which is frustrating for those customers.

7898.067 -> So again, a really big deal for people

7901.877 -> who are doing customer service.

7903.107 -> If you can know when a call is going awry in real time,

7907.507 -> and help that customer have a better experience

7909.727 -> and not do harm to your brand, that’s a big deal.

7912.547 -> The third category that we get people asking us for help in is,

7916.597 -> “How can I further optimize my agent’s time in various areas?”

7922.317 -> And the two areas that they asked are, first, they said,

7926.397 -> “About half of my time for agents is spent on doing tasks,

7931.587 -> tasks outside of calls”.

7933.627 -> You might have to file an insurance claim for a customer.

7936.077 -> You may need to contact the customer for a status change.

7939.827 -> Half of their time is spent on these tasks.

7942.297 -> These tasks live everywhere.

7944.357 -> They are not organized in any one central place.

7947.147 -> It’s hard to prioritize them. They are kept on pieces of paper,

7950.377 -> that’s why oftentimes people lose them.

7952.147 -> This is a big problem.

7953.827 -> The other thing that people ask to optimize agent’s time is they say,

7958.027 -> “At the beginning of a lot of these calls, our agents have to spend time

7962.487 -> manually authenticating customers and asking all these manual questions,

7966.837 -> and it wastes a lot of time for agents,

7968.587 -> it wastes a lot of time for customers.

7970.077 -> Can you help with that?”

7971.167 -> And so we have two solutions here to help you with these problems.

7974.527 -> The first is the launch of Amazon Connect Tasks,

7978.177 -> which automates, tracks, and manages tasks for contact center agents.

7982.844 -> [applause]

7988.227 -> And so Tasks makes it much,

7991.807 -> much easier to be able to manage tasks.

7995.487 -> Now you have one central place where an agent to see all their tasks,

8000.567 -> managers and supervisors get that same view,

8003.367 -> managers can easily assign the tasks there,

8006.067 -> they can choose to prioritize

8007.607 -> and where they want what agent to get each task done.

8011.487 -> They can actually decide to assign agents’ tasks

8015.007 -> based on how busy they are

8016.487 -> and whether they are on the phone or not.

8018.387 -> And then agents have everything right in front of them.

8020.817 -> And the nice thing also is that both agents and supervisors

8024.557 -> can automate a number of elements of tasks

8027.437 -> to make it even faster for them to get these things done.

8030.047 -> So this is very useful.

8032.677 -> This is going to change the efficiency of agents

8034.967 -> when they are not on calls. People are very excited about this.

8037.727 -> And then, for the call authentication issue,

8040.287 -> I am excited to share with you the launch of Amazon Connect Voice ID,

8043.747 -> which is real-time caller authentication using machine

8047.117 -> learning powered voice analysis.

8049.437 -> [applause]

8054.947 -> And so the way that voice ID works is that if you’re using it,

8059.137 -> you’ll ask whether customers want to opt into it,

8062.527 -> so they can have a better chance of avoiding fraud

8065.307 -> and also so they can save time in their calls.

8067.807 -> They opt in, and they say a few sentences

8070.687 -> and we build effectively a voice print for each customer.

8074.427 -> And then, when a caller calls in,

8078.227 -> we will just allow them to start talking.

8080.517 -> They’ll start talking about what they’re calling in for

8082.397 -> and what their problem is, and within a few seconds Voice ID

8086.287 -> will return to you who they think that customer is,

8089.277 -> and with a confidence level of that prediction.

8091.587 -> Each company will set their own threshold,

8093.937 -> but if it meets that threshold, you just continue having that call,

8097.677 -> you know there is no chance of fraud,

8099.307 -> you don’t waste all that time on the manual questions.

8101.827 -> And if it’s under that threshold that the company sets,

8104.727 -> then either you can choose to do manual verification,

8107.487 -> or transfer that call to a fraud specialist, but again,

8110.757 -> it totally changes the efficiency and the productivity both for agents

8116.657 -> as well as for customers who are calling in.

8118.857 -> And so if you think about it, call centers

8122.657 -> and call center solutions existed in a certain way for a long time.

8127.477 -> And even though customers didn’t like them,

8130.367 -> what happened was they left this area ripe for reinvention.

8135.337 -> And when you have an opportunity to reimagine an area

8138.427 -> in an experience like a call center,

8141.027 -> and you find a way to make it easy to get started and easy to scale,

8144.967 -> and much more cost effective, and you build capabilities

8147.857 -> that put all the relevant information about products

8150.947 -> and customers in front of agents

8152.397 -> so they can handle those in real time,

8154.417 -> and when you allow agents to be more productive

8157.247 -> with their tasks outside of calls,

8159.077 -> and you allow them to actually start calls much quicker,

8162.097 -> and then you allow them to use machine learning under the covers

8165.017 -> to do things like have the ability to impact calls in real time

8169.007 -> before there is any harm done to a brand -

8171.167 -> when you do all those types of things,

8173.487 -> you can completely transform any horizontal application space

8177.737 -> as we have with Connect.

8179.347 -> Every single one of these horizontal application spaces

8184.647 -> has the ability to be reinvented

8187.067 -> and will be reinvented in the next number of years.

8191.077 -> How about vertical business segments? Which of these are being reinvented?

8195.427 -> And the answer is, all of them.

8197.387 -> And I am going to give you a few examples.

8199.447 -> If you look in the auto space.

8201.467 -> This may be the space that has the most elements

8203.537 -> of reinvention happening simultaneously,

8205.577 -> but you have the electrification of cars,

8208.447 -> which is partially driven by the environment and partially driven

8211.747 -> by companies that came up right from the get-go,

8214.507 -> electric only companies like what Rivian has done on top of AWS.

8218.547 -> You have companies who are working and building autonomous vehicles.

8222.107 -> We do a lot of work with BMW and Lyft on these,

8224.287 -> but virtually every company is working on this.

8227.447 -> The car gives you a chance to have all these new connected experiences

8231.317 -> that weren’t really thought of many years ago.

8233.917 -> We’ve been working with companies like Toyota on this,

8235.947 -> but even today you may or may not have seen the announcement

8239.297 -> we just made with BlackBerry,

8241.377 -> where they have built a new intelligent vehicle data platform

8244.577 -> that they call IVY, on top of AWS,

8248.157 -> where they basically took their QNX software

8250.707 -> that does sensor management

8252.067 -> and that runs on 175 million vehicles on the road,

8256.497 -> and then what they’ve done is they’ve helped automakers

8259.477 -> better access, process, analyze, and share data

8263.117 -> from those connected vehicle sensors to build brand new experiences

8267.047 -> for car owners or drivers or passengers.

8269.397 -> So auto is transforming in a rapid way.

8273.087 -> The same thing with healthcare.

8274.757 -> Healthcare companies will tell you that the Holy Grail for them

8278.317 -> is to be able to take all these disparate pieces of information

8281.227 -> that live in all these different formats

8282.817 -> so they have a 360-degree view of customers.

8285.541 -> And they are not there yet,

8287.397 -> although we’ll have more to say on this

8288.887 -> and to help with this in the next couple of weeks.

8291.367 -> But they’re not there yet. But they are making progress.

8294.507 -> And they’re using machine learning to solve problems

8297.347 -> that they couldn’t solve before.

8298.587 -> So a good example is look at what Moderna has done in the last year

8302.817 -> or so, the last nine months.

8304.487 -> They’ve built an entire digital manufacturing suite on top of AWS

8308.197 -> to sequence their most recent COVID-19 candidate

8311.467 -> that they just submitted and it has 94% effectiveness.

8314.917 -> And they did it on AWS in 42 days,

8318.427 -> instead of the typical 20 months it takes.

8321.067 -> Just think about that.

8322.287 -> Think about what it changes with regard

8324.937 -> to what’s possible in healthcare if you’re able to take actions

8328.257 -> that are so momentous like that much more quickly and much more easily.

8331.947 -> You see the same type of invention happening in media and entertainment.

8336.147 -> I think most people here know

8338.447 -> that the way that we have started streaming

8341.047 -> and allowing consumers to use video content

8345.217 -> has radically changed in the last five years,

8347.957 -> and if you look at all the major streaming services,

8351.187 -> Netflix and Disney+ and Hulu and Prime Video,

8354.517 -> they all use AWS to do that,

8356.567 -> but you are also seeing really audacious new inventions

8361.017 -> like what Paul was talking about with Fox and ViacomCBS

8364.007 -> are trying the same type of thing, which is completely reinventing

8368.327 -> how they do broadcast and the ease and speed

8370.807 -> with which they can spin up brand new channels for people.

8375.487 -> And then you see the same thing

8376.977 -> happening in industrial manufacturing,

8379.657 -> where they are using edge and the data

8381.197 -> to rethink all their design processes on the production line,

8385.737 -> thinking about how to change their supply chains,

8388.227 -> how to change their products.

8389.487 -> And a company that’s a great example of this,

8392.177 -> which is totally reinventing themselves on top of AWS,

8395.717 -> is my next guest.

8397.307 -> It’s a privilege to welcome the President and CEO of Carrier,

8400.917 -> Dave Gitlin.

8402.914 -> [applause]

8405.667 -> Thank you, Andy. I am honored to be with you today.

8409.547 -> Like Amazon, Carrier is a company founded on innovation.

8414.207 -> Willis Carrier invented modern air conditioning

8416.607 -> at the age of 25 back in 1902, and it changed the way we live.

8421.707 -> Innovation is in our DNA and it is embedded in what we call

8425.487 -> ‘the Carrier way’.

8427.307 -> With a laser focus on customers, Innovation, and agility,

8430.707 -> we introduce more than 100 new products a year.

8433.797 -> And that kind of relentless customer focus and quick execution

8437.797 -> is what Amazon calls Day One. We literally just had our Day One

8442.007 -> when we became a public company on April 3rd,

8445.017 -> following our spin from United Technologies.

8447.657 -> And it has been so energizing.

8450.477 -> We think of ourselves as a 100-year old start-up.

8453.657 -> And our focus is clear.

8455.427 -> To be the world leader in healthy, safe, and sustainable building

8459.277 -> and cold chain solutions.

8461.677 -> On the building side, COVID has shined a light

8464.497 -> on the criticality of health and safety,

8467.077 -> and we are leading the way to ensure healthy buildings

8469.547 -> and safe indoor air environments.

8472.217 -> But today I am going to focus on our other ecosystem,

8476.687 -> and that’s the cold chain and our transformative alliance with AWS.

8482.097 -> So let me start with a problem statement.

8485.127 -> The safe distribution of many foods and pharmaceuticals requires

8488.367 -> uninterrupted cooling from inception to destination.

8493.427 -> Maintaining precise cooling conditions

8495.657 -> is more difficult than it sounds,

8497.107 -> because foods and pharmaceuticals can travel thousands of miles

8501.137 -> through a complex refrigerated distribution network

8504.267 -> called the cold chain. So consider a banana.

8507.927 -> Bananas are the most consumed food on earth

8509.827 -> and are extremely temperature sensitive.

8512.707 -> Once harvested, their clock starts ticking.

8515.517 -> Within hours, bananas must be cooled

8517.447 -> and maintained between 56 degrees and 58 degrees Fahrenheit,

8520.927 -> with relative humidity between 90 and 95%.

8524.617 -> Even a one degree temperature variation

8527.287 -> during that refrigerated journey can cause damage

8530.107 -> which doesn’t show up until arrival at the supermarket.

8533.787 -> So consider the two to three-week journey of a banana

8536.217 -> from, say, Ecuador to the United States.

8538.987 -> It starts at the farm.

8540.867 -> There, bananas are packed into a non-cooled cardboard box.

8545.827 -> They are then put into a series of refrigerated environments

8549.557 -> starting with a cold room on the farm to a truck,

8552.627 -> to an ocean container, shipped over the ocean,

8556.267 -> then onto a trailer to a distribution center,

8559.447 -> onto another truck to a ripening facility,

8561.947 -> to yet another truck to the grocery store.

8565.147 -> Each of these hand-offs from one refrigerated environment to another,

8569.017 -> occurs at outside temperature, and there is no margin for error.

8572.397 -> And today, there is no central repository

8575.327 -> for the temperature data for the entire journey.

8579.007 -> So now it’s not surprising to learn

8580.847 -> that one third of the world’s food produced every day is never consumed,

8585.607 -> due to waste, loss, or damage. Imagine that.

8589.527 -> Food waste costs the global economy nearly $1 trillion.

8593.777 -> But here’s what’s even more tragic.

8596.117 -> One in nine people go to bed hungry every night.

8600.067 -> If we could eliminate all of the cold chain challenges,

8603.047 -> we’d preserve enough food to feed 950 million people.

8607.237 -> That’s more than all of the hungry people in the world today.

8611.497 -> Think about the impact on the planet.

8612.917 -> Consider all of the resources and resulting emissions

8615.657 -> that go into producing food that never gets consumed.

8619.417 -> If food waste were a country,

8621.467 -> it would be the third largest emitter of greenhouse gases.

8624.867 -> It’s tragic on a human level, an environmental level,

8627.557 -> and a financial level.

8629.427 -> And the same problem exists with pharmaceuticals.

8631.547 -> They must be kept cold within very strict standards

8633.917 -> from the time the active pharmaceutical ingredient

8636.297 -> is first manufactured,

8638.357 -> through a series of hand-offs spanning ocean,

8640.577 -> air, warehouses, trailers, vans, freezers,

8643.317 -> and then temperature-sensitive medications must then be administered

8646.867 -> within hours of hitting room temperature.

8649.557 -> Today, we lose $35 billion in biopharma each year

8654.557 -> from temperature-controlled logistics failures.

8657.597 -> It’s a challenging problem in normal times,

8660.177 -> but an unprecedented challenge with the COVID vaccine.

8664.367 -> So now let’s talk solutions. We at Carrier are a global leader

8669.147 -> in providing equipment that cools containers,

8671.547 -> trucks, trailers, temporary remote storage, and cabinets.

8675.797 -> We also provide sensors that track temperature,

8678.967 -> location, and cargo condition.

8681.797 -> AWS has the world’s most widely adopted cloud platform

8685.597 -> with the broadest and deepest set of services.

8688.837 -> Together we are going to transform how perishable food

8692.407 -> and pharmaceuticals are protected through a new offering

8695.407 -> that we’ve developed, called Lynx.

8697.787 -> You can think of Lynx as an end-to-end digital

8700.717 -> connective platform for the cold chain.

8703.667 -> We will use AWS’s IoT, Analytics, and ML services

8707.867 -> to help customers reduce food and medicine loss,

8711.057 -> optimize supply chain logistics,

8712.917 -> and enhance environmental sustainability.

8716.127 -> Where data was previously siloed,

8718.517 -> we’ll now be able to aggregate temperature data

8720.967 -> across the products’ entire journey onto a data lake.

8724.267 -> Say, from the time the banana is first cooled on the farm

8727.537 -> to when it arrives in the supermarket.

8730.167 -> And we can use AWS ML capabilities to provide faster,

8734.017 -> smarter, actionable intelligence to better manage the cold chain

8738.047 -> and prevent breakpoints. With AWS ML

8740.777 -> services, for example, we can learn that truck doors

8744.187 -> should be not be left open longer than half an hour

8746.437 -> when loading or unloading cargo.

8749.427 -> With AWS ML services

8750.927 -> we can determine if a container has to be pre-cooled to 56 degrees

8755.277 -> and then use position location to initiate the pre-cooling process.

8760.047 -> We can also use weather data to detect

8762.097 -> and avoid transportation delays,

8764.097 -> and then use updated arrival times to alert the supermarket to a delay,

8767.847 -> so they can then adjust their produce displays.

8771.477 -> The data can also be used for equipment health,

8774.567 -> to remotely diagnose a compressor issue

8776.697 -> while the container is at sea, or better yet, use AWS IoT

8780.777 -> and analytics to help anticipate when a unit will fail,

8785.147 -> so we can then perform preventative maintenance.

8787.357 -> And we can also optimize logistics by combining truck

8790.427 -> positioning data with traffic

8792.437 -> and weather to help a distribution company

8794.997 -> improve fuel consumption and fleet up-time.

8798.167 -> All of this, helping food and pharmaceuticals

8801.107 -> get to where they are needed when they are needed

8803.737 -> and in the conditions that they are needed.

8806.327 -> So exciting times for us here at Carrier.

8809.707 -> With AWS we will be addressing profound issues like hunger,

8813.737 -> climate change, and safe vaccine distribution,

8816.847 -> and we could not be more excited to be on this journey

8819.667 -> with our friends at AWS. Thank you.

8822.922 -> [applause]

8827.207 -> Thank you, Dave.

8828.467 -> It is really impressive what Carrier is doing,

8831.567 -> and really important, particularly at this time.

8834.837 -> And as we said earlier, the keys to reinventing

8838.417 -> is a combination of building the right reinvention culture

8842.327 -> and then knowing what technology is available to you

8844.767 -> to make that reinvention change and using it.

8847.217 -> And that’s exactly what Carrier is leveraging.

8849.307 -> It’s very important and very impressive.

8852.807 -> So manufacturing and industrial companies

8855.847 -> are a group of customers who have said to us,

8859.667 -> “We know that we could change our customer experience

8864.507 -> and how we operate our plants

8866.117 -> so significantly using machine learning,

8869.357 -> but we just don’t often have the equipment

8871.857 -> or the talent to make that happen,

8873.487 -> and I really wish AWS would help us.”

8876.557 -> And so let me give you a couple of examples.

8879.157 -> If you look at machine data,

8881.697 -> there are a lot of industrial companies

8884.707 -> who know that if they could do predictive maintenance better,

8888.047 -> they could save a lot of money and time.

8890.557 -> And what manufacturing companies will tell you

8892.977 -> is that it’s always much less expensive

8896.327 -> to fix something before it breaks,

8898.187 -> not to mention saving the money of the downtime in the plant.

8902.647 -> But it’s actually not that easy to tell.

8906.207 -> Most people at a plant will tell you that you can hear it

8909.397 -> or you can feel it through the vibration before you can see it.

8913.037 -> But a lot of companies either don’t have sensors,

8916.037 -> or they are not modern and powerful sensors,

8918.377 -> or not consistent and they don’t know

8920.687 -> how to take that data from the sensors

8922.547 -> and send it to the cloud,

8924.127 -> and they don’t know how to build machine learning models.

8926.307 -> And our manufacturing companies that we work with have said,

8929.517 -> “Could you just solve this?

8930.717 -> Can you build an end-to-end solution?”

8932.777 -> So I am excited to announce today the launch of Amazon Monitron,

8936.567 -> which is an end-to-end solution for equipment monitoring.

8939.827 -> [applause]

8945.227 -> And so Monitron gives customers sensors,

8949.147 -> a gateway device to send the data to AWS.

8952.097 -> We build custom machine learning models for you

8954.987 -> in a mobile app with a UI so you can tell what’s happening.

8958.287 -> And all our customers do is you mount the sensors to your equipment,

8961.767 -> you start sending the data through the gateway device,

8964.567 -> and then, as we take the data in, we build the machine learning model

8968.817 -> that looks at what ‘normal’ looks like for you on sound or vibration,

8973.297 -> and then as you continue to stream that data to us,

8976.457 -> we will use the model to show you where there are anomalies

8980.047 -> and send that back to you in the mobile app,

8982.387 -> so you can tell where you might need to do predictive maintenance.

8985.087 -> That’s a big deal.

8986.277 -> That makes it much, much easier for companies to do.

8988.827 -> Now, there are other companies who say,

8990.357 -> “Look, I have modern sensors that I’m fine with.

8993.377 -> I’m also okay taking the data from those sensors

8996.187 -> and sending it to AWS,

8997.617 -> but I don’t want to build the machine learning models.

8999.487 -> I just want to send you the data, use your models,

9001.867 -> have the predictions come back to me through the API.”

9005.137 -> That third layer of that machine learning stack.

9008.047 -> And so we have something for this group of customers

9010.017 -> as well to announce today,

9011.657 -> which is the launch of Amazon Lookout for Equipment,

9014.607 -> which shows anomaly detection for industrial machinery.

9018.765 -> [applause]

9023.977 -> And so with Lookout for Equipment you just send the data to S3,

9028.407 -> or we have a service called IoT SiteWise

9030.417 -> that lets you send your machine data

9031.817 -> in a structured way for your analytics.

9033.987 -> You send the data to AWS,

9036.777 -> we will assess sound, vibration, temperature,

9041.707 -> and we’ll again build a model of what normal looks like,

9044.687 -> and as we see anomalies we’ll send them to you via the API,

9047.777 -> so that you can do predictive maintenance.

9050.117 -> These are game changers for industrial companies

9053.057 -> that want to be doing predictive maintenance and saving money.

9055.957 -> And the second problem they asked to help with was computer vision.

9059.497 -> And if you think about it, there are a lot of split decisions

9062.347 -> that you’ve got to make in facilities,

9064.477 -> on your production lines,

9065.747 -> or even in interactions between people

9067.507 -> when you’re trying to be socially distanced,

9069.647 -> where you just don’t have the time

9071.607 -> to send that information to the cloud and get an answer back.

9074.287 -> You need to make that decision in real time.

9077.297 -> So what these industrial companies need,

9080.297 -> and often try to employ in some fashion, are cameras -

9082.837 -> these smart cameras that allow them to do streaming video.

9086.467 -> But the problem is, most of the smart cameras out there today

9090.247 -> are just not powerful enough to run sophisticated

9092.787 -> computer vision models at the edge. And most companies you talk to,

9097.327 -> they don’t want to rip out all their cameras that they have installed,

9099.817 -> but they know they need to give help to those cameras

9102.917 -> to be able to do computer vision in a sophisticated way.

9106.507 -> They asked us if we would try to help with that.

9108.837 -> And so I am excited to announce today

9110.287 -> the launch of the AWS Panorama Appliance,

9113.327 -> which is a new hardware appliance

9114.747 -> that allows organizations to add computer vision

9117.487 -> to existing on-premises smart cameras.

9120.547 -> [applause]

9125.667 -> So here’s how it works.

9127.227 -> You simply plug in the Panorama Appliance

9129.957 -> and connect it to your network,

9131.857 -> and Panorama starts to recognize and pick up video streams

9135.177 -> from your other cameras in the facility.

9137.777 -> The Panorama Appliance can accept streams of up

9141.007 -> to 20 concurrent streams and operate on those.

9143.687 -> If you need to have more concurrently,

9145.367 -> you can buy more Panorama Appliances.

9148.467 -> And then we have pre-built models inside Panorama

9152.327 -> that do computer vision for you and that we’ve optimized by industry.

9156.767 -> So we’ve got them in manufacturing,

9158.777 -> in construction, retail, and a host of others.

9161.677 -> You can, of course, choose not to use the pre-built models

9164.487 -> and build your own in SageMaker,

9166.367 -> and then just deploy those to Panorama.

9168.987 -> And then Panorama also integrates seamlessly

9171.897 -> with the rest of the AWS IoT and machine learning services,

9174.657 -> where, if you actually want to send that data,

9177.767 -> not for real-time actions,

9179.127 -> but to do large-scale analytics on what’s happening in the plant,

9182.277 -> you can send them to us through Panorama

9184.127 -> and you can use it in the rest of your AWS Regions.

9187.187 -> Now, this is pretty exciting,

9189.867 -> and people are pretty excited about the possibility

9192.027 -> of having real computer vision there, but they also have told us that,

9196.847 -> “Look, we’re going to buy the next generation of smart cameras,

9200.587 -> and those smart camera manufacturers have told us

9203.517 -> we want to actually embed something

9205.637 -> that allows us to run more powerful computer vision models in there,”

9209.447 -> so we are also providing a brand new AWS Panorama SDK

9214.807 -> which enables hardware vendors to build new cameras

9217.727 -> that run more meaningful computer vision models at the edge.

9221.247 -> [applause]

9227.297 -> And so what this will do is,

9229.537 -> if you are a company that runs cameras,

9234.617 -> and you are building the next generation of cameras,

9236.567 -> you will be able to use this SDK and the APIs associated with it.

9240.047 -> We’ve done all this work to optimize the parameters

9243.267 -> around memory and latency so you can actually, in the camera,

9246.617 -> fit more powerful models in a more constrained space,

9249.857 -> and it’s going to change what’s available for companies

9252.667 -> as they’re building smart cameras moving forward.

9255.927 -> If you’re an industrial company,

9258.237 -> and you’ve built a culture that’s able to reinvent,

9261.927 -> if you use these tools I just talked about,

9264.257 -> along with a host of others that AWS provides,

9267.097 -> you can totally reinvent what you are doing

9271.597 -> in your industrial manufacturing company,

9273.447 -> much like Carrier has.

9275.117 -> And the reality is that this is true and available to every single company

9280.587 -> in every vertical business segment. That reinvention is there for you.

9285.007 -> If you’re ready, we have a lot for you, with a lot more coming.

9290.237 -> So the final area of reinvention that I’m going to talk about today

9294.137 -> is really around hybrid infrastructure,

9296.367 -> and I am going to yield the floor

9297.557 -> to the Head of Infrastructure at Riot Games, Zach Blitz.

9302.336 -> [music playing]

9307.017 -> With Outposts, AWS gave us a unique solution

9309.827 -> to ensure a level playing field for our players,

9312.177 -> and streamlined our deployments using the same tools and APIs on-

9315.897 -> premises and in the cloud. We rolled out Valorant fast

9320.087 -> and we are continuing to reinvent

9321.977 -> how we design and deploy our games

9323.707 -> to provide our players with the best possible game experience.

9327.167 -> We shipped AWS Outposts to new colos quickly,

9330.237 -> enabling a rapid deployment of game servers,

9332.647 -> standardized on a single build, test, and production pipeline.

9336.067 -> And best of all, using AWS

9338.157 -> Outposts, we reduced latency for players by 10-20 milliseconds,

9341.877 -> minimizing peeker’s advantage

9343.627 -> and creating a level playing field for all players.

9347.234 -> [applause]

9351.147 -> Thank you, Zach. I appreciate it, Zach.

9353.677 -> It’s very impressive to continue to see the way

9357.057 -> that Riot is innovating for their players,

9359.157 -> and it’s also impressive to see how they’re using AWS in Regions

9364.487 -> as well as on-premises in a seamless and consistent way.

9370.847 -> When you think about the term ‘hybrid’

9372.637 -> and ‘hybrid infrastructure’, I think a lot of people believe

9376.497 -> that this term and these solutions are pretty set.

9381.027 -> But in our very strong opinion,

9383.347 -> both the definition of the term and the solutions themselves,

9388.337 -> are innovating and being reinvented really quickly.

9393.047 -> When people ask, “What’s hybrid?”, you know,

9396.117 -> typically, people view it as a mix of modes.

9398.757 -> And people define it early on as a combination of cloud

9402.427 -> alongside on-premises data centers.

9404.337 -> And one of the reasons it was defined that way

9406.947 -> is the people who popularized this term

9410.187 -> were on on-premises infrastructure product providers,

9413.357 -> and they wanted to ride along with the momentum of the cloud.

9416.747 -> So that’s how the definition started off,

9419.767 -> which was cloud and on-premises data centers.

9423.287 -> And it led to all this breathless debate

9425.867 -> about whether this was going to turn out to be a binary situation.

9428.737 -> Would you only use the cloud or only use on-premises?

9432.867 -> And we probably contributed a little bit to that confusion,

9435.597 -> because we stated our then very strong belief

9439.147 -> and now even stronger belief, that the vast majority of companies,

9443.927 -> in the fullness of time, will not have their own data centers.

9447.677 -> And those that do will have much smaller footprints.

9449.957 -> But we always thought that was going to happen

9452.327 -> in the fullness of time, not this year.

9454.877 -> And we knew that it would take several years

9457.257 -> and so we just spent a lot of our time

9459.237 -> in the early years of AWS on the cloud-specific pieces,

9463.497 -> and then building sensible bridges back to on-premises data centers.

9467.437 -> And these are things like Virtual Private Clouds,

9470.007 -> or VPCs, or Direct Connect, or Storage Gateway.

9472.887 -> But in the meantime, you had a number of companies

9476.397 -> who tried to jump on owning what hybrid was,

9479.347 -> and built all of these hugely-hyped capabilities

9483.977 -> that were supposed to be hybrid capabilities

9485.517 -> that never lived up to the hype, and really never got any traction.

9489.357 -> And so we were watching this happening,

9492.457 -> and we went back to first principles and we asked ourselves,

9495.087 -> “So, wait a second. What really is hybrid?”

9499.347 -> So it’s cloud and on-premises. What does on-premises mean?

9503.587 -> Is it just on-premises data centers?

9506.517 -> What about a restaurant? Is that on-premises?

9509.347 -> What about an agricultural field? Is that on-premises?

9513.227 -> If those are on-premises, they have very different requirements

9516.427 -> than on-premises data centers,

9517.757 -> and so we think of hybrid infrastructure

9521.497 -> as including the cloud, along with various other edge nodes,

9525.187 -> on-premises data centers being one of them.

9527.557 -> But there are several of them.

9528.927 -> And the way that customers have told us

9530.687 -> they want to consume our hybrid offering is with the same APIs,

9535.297 -> the same control plane, the same tools,

9537.837 -> and the same hardware that they are used to using in AWS Regions.

9540.977 -> Effectively, they want us to distribute AWS

9544.507 -> to these various edge nodes.

9546.687 -> So we reimagined for ourselves what hybrid was,

9550.157 -> and we started to build solutions

9552.367 -> that picked off the biggest use cases,

9554.257 -> but in a way that worked for customers both short and long-term.

9559.037 -> And so we started with, customers said,

9562.707 -> “Look, I want to be able to use the same tools

9566.377 -> I’ve used to manage my infrastructure on-premises

9568.797 -> for the last number of years. I want to use that in the cloud.

9571.607 -> I want to use it to manage my cloud deployment.”

9574.367 -> Most of the world, at this point, has virtualized on top of VMware.

9578.887 -> So we started working with Pat Gelsinger

9580.907 -> and the VMware team on how to build a new offering,

9583.947 -> which is called VMware Cloud on AWS,

9586.347 -> that allows customers to use those VMware tools

9589.097 -> they have been using for many years

9590.577 -> to manage their on-premises infrastructure,

9592.987 -> and manage their infrastructure on AWS.

9596.007 -> And this is a very unusual collaboration.

9600.247 -> There’s no other managed service

9601.647 -> that VMware runs with another cloud provider.

9604.197 -> There is none that have the functionality

9605.917 -> and capability of this VMware Cloud on AWS.

9608.887 -> It’s not just that both VMware and AWS

9611.257 -> have its engineering teams

9612.407 -> and its product teams closely tied at the hip,

9614.977 -> but also our field teams and our partner teams work together

9618.687 -> with customers. It’s a very unusual partnership.

9621.817 -> And it’s gaining a lot of momentum and a lot of steam.

9625.917 -> And you can see there are a lot of customers,

9627.837 -> whether you are talking about S&P Global or PennyMac,

9630.317 -> or Johnson & Johnson or Phillips or Palantir, Scottish Government,

9634.277 -> Lots of customers are using VMware Cloud on AWS.

9637.757 -> You also see the growth,

9639.137 -> almost double the amount of nodes year over year.

9642.827 -> IDC just had a report that showed over five years

9645.807 -> you get a 500% return on investment.

9648.617 -> There’s a lot of momentum in VMware Cloud on AWS

9651.857 -> and it’s really, really handy,

9654.687 -> as you’re moving from on-premises infrastructure to the cloud.

9657.807 -> Then a lot of our customers said,

9659.027 -> “Well, that is awesome, that is very useful,

9661.747 -> but what about when I need to keep workloads on-premises

9667.057 -> for the foreseeable future?

9668.217 -> Maybe they need to be close to a factory or something,

9670.717 -> that lives near my data centers, but I want to use

9672.907 -> AWS there right on-premises.” So there are a number of companies

9677.787 -> that have tried to solve this solution,

9679.347 -> none of which have gotten any traction.

9681.347 -> And we tried to go to school on what wasn’t working.

9684.137 -> And what customers hated about those solutions

9687.387 -> was that it didn’t have the same APIs or tools

9690.187 -> or control plane or hardware. They were totally different.

9693.017 -> There was too much work, and they weren’t doing it

9695.077 -> because it was too much friction.

9697.017 -> And so we changed how we thought about what people wanted there,

9701.547 -> and we changed our mindset to realize what they really wanted

9705.547 -> was they wanted us to distribute AWS to the on-premises locations and nodes.

9711.187 -> And so that’s why we built Outposts,

9713.627 -> which we announced two years ago and launched last year,

9716.997 -> which effectively rolls in racks of AWS services.

9721.307 -> You’ve got compute, storage, database and analytics,

9726.047 -> and soon machine learning. It’s fully managed.

9728.237 -> We deliver it, we install it, we’ll do all the maintenance.

9731.157 -> And it comes with the same APIs, the same control plane,

9734.687 -> the same tools, and the same hardware.

9736.957 -> People are very excited about Outposts.

9738.667 -> In a short amount of time with thousands of customers using them,

9741.817 -> these are customers like Philips and Volkswagen and Erikson and Cisco,

9745.667 -> Lockheed Martin, T-Systems, and Toyota.

9749.497 -> Just loads of companies who are using these services,

9753.967 -> and using Outposts and very excited about it.

9755.937 -> And so customers said, “While we love Outposts,

9760.197 -> and it really worked well on on-premises data centers,

9762.977 -> but they are big racks. What if I want to use Outposts

9768.167 -> in a much smaller space where I can’t afford to have racks?...

9771.577 -> I don’t have the space. I need servers.”

9773.847 -> And so I am excited to announce today two new formats of Outposts,

9779.897 -> smaller Outposts

9781.647 -> that let you run AWS infrastructure in locations with less space.

9786.406 -> [applause]

9792.017 -> And so instead of these big rack solutions,

9796.237 -> these are server solutions for Outposts.

9799.077 -> So the first size is what we call 1U size,

9802.077 -> which is 1¾ inches tall, the size of a pizza box.

9806.157 -> It’s a 40th of the size of the bigger rack

9808.937 -> Outposts that we launched a year ago.

9811.427 -> And then we have a 2U size which is 3½ inches tall,

9815.387 -> which is almost like two pizza boxes stacked, and these two smaller

9820.307 -> Outposts formats have the same functionality as

9823.317 -> Outposts, just for a smaller space. And now it means

9826.607 -> that restaurants or hospitals, or retail stores or factories can use

9832.447 -> Outposts to have AWS distributed to that edge.

9837.257 -> Then customers said, “Okay, that’s really cool. I can have

9840.767 -> Outposts on-premises data centers in a much smaller space.

9844.277 -> How about the use case where I need

9847.357 -> AWS distributed in major metropolitan areas

9851.017 -> where it may not be cost-effective either for you at AWS

9853.827 -> or me, the customer, to have a data center there,

9856.887 -> but I’d be willing to pay a little extra

9859.077 -> in exchange for my most demanding low-latency applications

9862.857 -> being able to reside in major metropolitan areas?”

9865.947 -> And that’s why we built and announced last year AWS Local Zones.

9870.317 -> We started with the first one in Los Angeles,

9872.467 -> aimed at the film makers and graphics renderers and gaming companies.

9878.077 -> And I am excited to announce today

9881.277 -> we have three new local zones that are launching today.

9884.687 -> In Boston, in Houston, and in Miami,

9888.547 -> and then 12 more in the United States in 2021, in Atlanta, Chicago,

9895.177 -> Dallas, Denver, Kansas City, Las Vegas, Minneapolis, New York,

9901.687 -> Philadelphia, Phoenix, Portland, and Seattle.

9907.217 -> [applause]

9913.277 -> I don’t know if you’re applauding

9914.457 -> because you’re excited about the Local Zones,

9916.577 -> or that I remembered all 12 of those. But we’re excited about it

9919.507 -> and we think it’s going to help you deploy AWS

9921.767 -> as metropolitan city edges.

9924.827 -> “What about when I actually need AWS distributed to the edge

9929.347 -> where there’s no connectivity, or where the terrain is so rugged

9933.857 -> that I need something that can be banged around a little bit?”

9936.717 -> This might be in an agricultural field or an oil field,

9939.497 -> or a military battlefield.

9941.917 -> And for this, we built our Snow Family of products.

9945.347 -> So, Snowball Edge, and Snowcone, which are different sizes,

9948.287 -> but both are hardware appliances

9950.867 -> that you can bring to that disconnected edge.

9953.317 -> It stores data that’s running off the different assets

9956.057 -> that you have with the edge. It has Compute on it,

9958.507 -> so you can do some analytics and processing on it,

9960.797 -> and then, if at any point you want to actually detach that appliance

9965.447 -> and send it back to us to have it adjusted in our data centers

9969.557 -> so that you can do large-scale analytics,

9971.907 -> you can do that as well. “How about distributing AWS

9977.927 -> if I want to build mobile applications

9980.337 -> and I want to take advantage of 5G?” And people are very excited about 5G,

9985.607 -> because the latency and the power of it

9987.827 -> give you a chance to build sub-10 millisecond latency applications

9992.387 -> that can do things like smart manufacturing

9994.407 -> or autonomous vehicles or various things in games.

9998.637 -> But any application that wants to do anything interesting

10001.987 -> is going to need compute, it’s going to need storage,

10004.437 -> it’s going to need infrastructure.

10005.747 -> And the problem is, for mobile applications

10008.337 -> to leverage that infrastructure which is typically AWS,

10012.447 -> they have to go from the device to the mobile network,

10015.837 -> to the local aggregation location,

10017.647 -> to the regional aggregation location, to the internet, to AWS and back.

10021.747 -> And that’s not 10 milliseconds, that’s seconds.

10024.407 -> And so customers wanted a way to change that,

10027.467 -> and that’s why we announced last year

10029.857 -> the launch of what we built, called AWS

10033.097 -> Wavelength, which extends AWS infrastructure to the 5G edge,

10037.477 -> so now you only have to go from the device

10041.267 -> to the natural 5G aggregation site, and AWS has racks of Outposts

10045.997 -> right there for you to do your infrastructure.

10048.577 -> We launched this last year with Verizon,

10051.337 -> who’s been an amazing partner.

10053.137 -> We already have eight US cities that we’ve launched,

10055.747 -> with more coming in the coming weeks. We are launching with KDDI in Tokyo,

10060.557 -> and SK Telecom in South Korea in the next few weeks,

10064.037 -> and then with Vodafone in London in early 2021.

10067.387 -> And one of the problems that we also thought about

10069.247 -> and saw that the customers asked about was they said,

10071.277 -> “Look, there are all these different telecom providers.

10074.227 -> They all have different semantics.

10076.097 -> I don’t want to have to learn all those.

10078.057 -> Can you build an abstraction

10079.417 -> so I’m just writing to a Wavelength zone,

10082.357 -> and then you do all the normalizing under the covers?”,

10085.987 -> which is what we’ve done.

10087.167 -> So there are a lot of ways of being able

10091.597 -> to bring the AWS experience to customers wherever they are.

10095.507 -> And so when you go back to asking “What is hybrid?”

10100.187 -> It’s not just cloud and on-premises data centers.

10104.097 -> It’s cloud along with various edge nodes,

10107.687 -> on-premises data centers being one of them, with several others.

10111.607 -> And we think most of this computing will end up in the cloud over time,

10116.447 -> which will be like the hub just given the cost and the agility

10119.097 -> and functionality and productivity advantages for builders.

10121.877 -> But several other workloads will reside where it makes most sense.

10126.047 -> You will have on-premises data centers

10128.417 -> when you’re doing this transition from on-premises to the cloud,

10131.047 -> where they need to be close to something

10133.187 -> that lives near an on-premises data center.

10135.787 -> You’ll have them in various smaller venues where you want them,

10139.027 -> in a restaurant or in a hospital, or in a factory.

10143.877 -> You’ll have them in major metropolitan areas

10146.947 -> where you have your most demanding low-latency workloads,

10150.607 -> where you are willing to pay a little bit extra

10152.057 -> to have that low latency.

10153.877 -> You’ll be able to have them in the disconnected edge,

10157.587 -> and you’ll be able to have AWS as well when you’re building

10160.517 -> 5G mobile applications that need to sit at a 5G mobile edge.

10164.557 -> And people will want this hybrid experience to be delivered by AWS

10169.267 -> by distributing AWS to these edge nodes with the same APIs,

10174.167 -> the same control plane, the same tools,

10176.857 -> and the same hardware they get in AWS Regions.

10179.617 -> That’s where we believe hybrid is heading,

10181.927 -> and how we’re trying to enable it.

10186.667 -> So I’m going to close with a song lyric,

10189.267 -> and because of COVID and the way we did this virtually,

10191.617 -> we didn’t have the band this year,

10192.857 -> but I’m going to use a song lyric to try and bring us home.

10196.667 -> And the lyric says, “I wish I could Google my ending.

10199.887 -> Someone give me reassurance, answers, anything will do.”

10203.347 -> And this is from one of my very favorite song writers,

10206.147 -> and what she was talking about

10208.397 -> was really the uncertainty of being a young adult,

10212.187 -> and what’s going to happen.

10213.677 -> But I think that same message is really applicable to companies,

10218.547 -> and what’s going to happen in the future.

10220.277 -> If you’re a missionary and you’re focused

10222.687 -> on building a lasting company versus making a quick buck

10225.477 -> and being a mercenary,

10226.967 -> you know how hard it is to build a sustainable business.

10231.027 -> So many things can derail you.

10234.017 -> New technology, new competitors, losing leaders,

10237.787 -> losing key contributors, regulation, pandemics.

10242.017 -> There are a lot of things. And like for a young person

10245.477 -> wondering what the future holds for them,

10247.597 -> the same is true for companies. It’s daunting.

10251.007 -> And I think the same counsel applies.

10253.677 -> You can’t control for every change and every development,

10257.147 -> but you can build the capacity to get to the truth,

10261.427 -> to make changes when they’re needed,

10263.407 -> to have people around you who want to help you change

10265.867 -> and can help you make the change, stay focused on what matters most,

10270.557 -> to move fast when speed is required, which is more often than you realize,

10274.847 -> and then to be aware of what’s available to you

10276.967 -> and what’s changing around you, so you can reinvent who you are

10280.857 -> and what your customer experience will be.

10282.997 -> That’s what you need for reinvention, and in my opinion,

10287.807 -> companies who aren’t already reinventing themselves

10290.187 -> in some meaningful way are unwinding, whether they realize it or not.

10294.627 -> The good news though is that invention

10297.177 -> and reinvention is very doable,

10300.157 -> if you’re intentional and focused on it.

10302.717 -> And we’ll be here every step of the way to help you do it.

10305.847 -> I want to thank you for listening. I hope everybody stays safe,

10309.107 -> and I hope you have a great few weeks of re:Invent.

10311.067 -> Thank you very much.

10313.718 -> [music playing]

10316.468 -> [applause - cheering]

Source: https://www.youtube.com/watch?v=xZ3k7Fd6_eU