WEBVTT - AWS CEO Matt Garman Talks OpenAI Tech After Microsoft Cedes Exclusivity

0:00:02.520 --> 0:00:07.080
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news.

0:00:07.400 --> 0:00:11.440
<v Speaker 2>You've built quick AI that works across all applications, but

0:00:11.480 --> 0:00:13.680
<v Speaker 2>I think that the main issue or point is that

0:00:13.720 --> 0:00:15.920
<v Speaker 2>it builds context over time.

0:00:16.600 --> 0:00:19.000
<v Speaker 3>What were you solving for with quick.

0:00:19.040 --> 0:00:22.599
<v Speaker 1>Yeah, quick quick as it's the way for companies to

0:00:22.640 --> 0:00:25.239
<v Speaker 1>get AI and an AI assistant in the hands of

0:00:25.440 --> 0:00:28.400
<v Speaker 1>every employee that they have across the company, and it

0:00:28.440 --> 0:00:32.240
<v Speaker 1>allows customers to leverage data across all of their enterprise.

0:00:32.320 --> 0:00:36.400
<v Speaker 1>But excitingly, today we're launching a desktop application and I

0:00:36.479 --> 0:00:38.080
<v Speaker 1>personally have been using it for the last couple of

0:00:38.120 --> 0:00:41.279
<v Speaker 1>weeks and it's the single best AI productivity enhancement I've

0:00:41.280 --> 0:00:43.640
<v Speaker 1>tried of anything. It's really incredible. It allows you to

0:00:43.680 --> 0:00:45.920
<v Speaker 1>work across your email, across your slack, across all the

0:00:45.920 --> 0:00:49.960
<v Speaker 1>information on your desktop, quickly pull information across various different settings,

0:00:50.000 --> 0:00:51.960
<v Speaker 1>and it learns with you as you go, so it

0:00:52.040 --> 0:00:54.320
<v Speaker 1>learns who your coworkers are, learns who you operate with,

0:00:54.520 --> 0:00:58.680
<v Speaker 1>learns who you write to. It's an incredible, incredible productivity enhancer,

0:00:59.320 --> 0:01:01.680
<v Speaker 1>and it's something that launching today and really excited about.

0:01:02.240 --> 0:01:08.399
<v Speaker 2>AWS is pushing deeper into applications in the agentic AI

0:01:08.440 --> 0:01:10.080
<v Speaker 2>context that is everything.

0:01:09.760 --> 0:01:13.960
<v Speaker 3>From healthcare, hiring, supply chain.

0:01:15.400 --> 0:01:18.080
<v Speaker 2>What was the need that you saw, because that a

0:01:18.120 --> 0:01:20.000
<v Speaker 2>lot of the customer of Base and aws will for

0:01:20.040 --> 0:01:20.920
<v Speaker 2>something quite similar.

0:01:21.200 --> 0:01:24.640
<v Speaker 1>Yeah, we think, and it's true many of our customers

0:01:24.680 --> 0:01:26.760
<v Speaker 1>are going to reinvent many of the applications that are

0:01:26.760 --> 0:01:29.280
<v Speaker 1>out there. But we see that with agents out there

0:01:29.280 --> 0:01:33.480
<v Speaker 1>today and AI, every application is being remade. And with

0:01:33.520 --> 0:01:37.880
<v Speaker 1>Amazon Quick we saw that opportunity to remake personal productivity

0:01:37.920 --> 0:01:41.800
<v Speaker 1>inside of an enterprise setting. And with connect we're doing

0:01:41.800 --> 0:01:45.280
<v Speaker 1>the same across many enterprise applications, whether it's with supply chain,

0:01:45.520 --> 0:01:48.760
<v Speaker 1>whether it's with contact centers, whether it's with healthcare. We

0:01:48.800 --> 0:01:52.200
<v Speaker 1>see the application opportunity to apply AI and agents to

0:01:52.240 --> 0:01:54.680
<v Speaker 1>a number of these applications. And these are areas where

0:01:54.720 --> 0:01:57.640
<v Speaker 1>Amazon has a lot of expertise and where we can

0:01:57.680 --> 0:02:01.080
<v Speaker 1>add that expertise in the customers use AI to completely

0:02:01.080 --> 0:02:04.800
<v Speaker 1>reimagine many of these traditional thirty forty fifty year old processes.

0:02:04.880 --> 0:02:07.560
<v Speaker 2>The companies it makes you think of names like Salesforce

0:02:07.600 --> 0:02:12.480
<v Speaker 2>Service now sap key customers. What kind of conversations did

0:02:12.520 --> 0:02:17.480
<v Speaker 2>you have with them about why the broad suite.

0:02:17.160 --> 0:02:18.760
<v Speaker 3>Through Amazon Connects is useful?

0:02:19.000 --> 0:02:21.359
<v Speaker 1>Yeah, and they may these the things that we're building

0:02:21.360 --> 0:02:23.960
<v Speaker 1>are actually quite complementary to many of those partner offerings,

0:02:23.960 --> 0:02:27.680
<v Speaker 1>and so I think if you think about actually Connect,

0:02:27.919 --> 0:02:30.200
<v Speaker 1>Connect and Salesforce have had a partnership for a really

0:02:30.240 --> 0:02:33.200
<v Speaker 1>long time where many customers use those together in order

0:02:33.240 --> 0:02:36.760
<v Speaker 1>to provide even better outcomes for their customers. And we

0:02:36.880 --> 0:02:39.200
<v Speaker 1>think that's going to continue to be the case today,

0:02:39.240 --> 0:02:44.880
<v Speaker 1>where things like Connect Decisions is really a helpful add

0:02:44.919 --> 0:02:47.840
<v Speaker 1>on the customers that are doing supply chain and it's

0:02:47.880 --> 0:02:50.840
<v Speaker 1>an agentic helper to help you make great supply chain decisions,

0:02:51.040 --> 0:02:53.919
<v Speaker 1>but use it together with your existing supply chain solution,

0:02:54.000 --> 0:02:56.520
<v Speaker 1>and so it's really complementary to many of those products

0:02:56.520 --> 0:02:59.280
<v Speaker 1>that our customers and our partners build on top of

0:02:59.280 --> 0:03:01.440
<v Speaker 1>AWS today.

0:03:01.520 --> 0:03:05.000
<v Speaker 2>What's the strategy and goal for AWS and pushing deeper

0:03:05.040 --> 0:03:08.280
<v Speaker 2>into the application layer. What do you think it will

0:03:08.360 --> 0:03:10.600
<v Speaker 2>change for you rather than than what you just outlined

0:03:10.600 --> 0:03:11.160
<v Speaker 2>with the customers.

0:03:11.240 --> 0:03:15.120
<v Speaker 1>We've always thought that AWS and we always have produced

0:03:15.160 --> 0:03:17.600
<v Speaker 1>offerings for customers at every layer of the stack. We

0:03:17.639 --> 0:03:22.040
<v Speaker 1>obviously excel at the infrastructure layer, where AWS Cloud is

0:03:22.080 --> 0:03:25.120
<v Speaker 1>by far the largest and the most established, and we've

0:03:25.160 --> 0:03:28.200
<v Speaker 1>also built services higher up the stack, things like databases

0:03:28.280 --> 0:03:33.320
<v Speaker 1>things like analytics services, things like AI services, and increasingly

0:03:33.360 --> 0:03:36.000
<v Speaker 1>with Connect. For the last ten years, we've been offering

0:03:36.040 --> 0:03:39.440
<v Speaker 1>applications in the top layer, and we've always said there's

0:03:39.480 --> 0:03:42.160
<v Speaker 1>millions and millions and millions of customers who build applications

0:03:42.200 --> 0:03:44.520
<v Speaker 1>of that top layer. We thought, we always think that

0:03:44.560 --> 0:03:47.880
<v Speaker 1>we'll build a few some of those will show customers

0:03:48.400 --> 0:03:51.280
<v Speaker 1>things that they're able to do with that technology. Sometimes

0:03:51.280 --> 0:03:54.160
<v Speaker 1>it's to expose areas where we're experts and we think

0:03:54.200 --> 0:03:57.360
<v Speaker 1>that we can provide value to customers, but we always

0:03:57.440 --> 0:03:59.120
<v Speaker 1>know that there's going to be lots of players up

0:03:59.160 --> 0:03:59.920
<v Speaker 1>there at the application.

0:04:00.440 --> 0:04:04.000
<v Speaker 2>What AWSS too, it's about is AI revenue run rate,

0:04:04.080 --> 0:04:06.880
<v Speaker 2>which getting into detail, seems to focus a lot on

0:04:06.880 --> 0:04:09.920
<v Speaker 2>the frontier models, how they're access to, how they use

0:04:10.040 --> 0:04:13.640
<v Speaker 2>But if you think about quick Connect applications, is that

0:04:13.720 --> 0:04:16.960
<v Speaker 2>something Matt that you see making meaningful contribution in the

0:04:17.000 --> 0:04:18.279
<v Speaker 2>AI revenue bucket?

0:04:18.480 --> 0:04:19.640
<v Speaker 1>Absolutely we see.

0:04:20.480 --> 0:04:21.240
<v Speaker 3>You're absolutely right.

0:04:21.279 --> 0:04:23.360
<v Speaker 1>A lot of our AI revenue today is on Amazon

0:04:23.400 --> 0:04:27.560
<v Speaker 1>Bedrock and it's through accessing frontier models. But we see

0:04:27.560 --> 0:04:29.479
<v Speaker 1>that every application in the world is going to be

0:04:29.520 --> 0:04:32.200
<v Speaker 1>remade with AI. We think agents are going to totally

0:04:32.360 --> 0:04:36.159
<v Speaker 1>change how work is done, how jobs are efficiently done,

0:04:36.680 --> 0:04:39.560
<v Speaker 1>and we think we have some opportunities to have some

0:04:39.720 --> 0:04:43.000
<v Speaker 1>unique offerings in that space that really change the game.

0:04:43.240 --> 0:04:45.240
<v Speaker 1>And so we think that Amazon Quick is one of those.

0:04:45.240 --> 0:04:47.719
<v Speaker 1>We think Connect is one of those. We think Hero

0:04:47.839 --> 0:04:49.960
<v Speaker 1>on the coding side can really help people in and

0:04:50.000 --> 0:04:52.920
<v Speaker 1>when you have some of these applications, that often inspires

0:04:52.960 --> 0:04:54.880
<v Speaker 1>others to build around them. And we think it's a

0:04:54.880 --> 0:04:58.159
<v Speaker 1>great opportunity for us where we think we have differentiated

0:04:58.160 --> 0:05:00.040
<v Speaker 1>ideas to really change how things have been done for

0:05:00.080 --> 0:05:00.599
<v Speaker 1>a long time.

0:05:01.120 --> 0:05:05.240
<v Speaker 2>Open AI, Yeah, what's new, what's changed in your relationship?

0:05:05.320 --> 0:05:07.640
<v Speaker 1>Yeah, well, we're quite excited about the partnership that we're

0:05:07.640 --> 0:05:12.080
<v Speaker 1>announcing with them today. So yeah, So for a long time,

0:05:12.560 --> 0:05:15.360
<v Speaker 1>when we built Amazon Bedrock as a way for our

0:05:15.360 --> 0:05:19.640
<v Speaker 1>customers to access frontier models and access AI models, we've

0:05:19.680 --> 0:05:22.320
<v Speaker 1>always started with the position that we wanted to offer

0:05:22.400 --> 0:05:24.880
<v Speaker 1>choice and we wanted to offer all of the best

0:05:24.920 --> 0:05:27.919
<v Speaker 1>models available out there. And today we're excited to be

0:05:27.920 --> 0:05:31.960
<v Speaker 1>bringing open AI into Bedrock. It's I think it's it's

0:05:31.960 --> 0:05:33.840
<v Speaker 1>something that our customers have asked for for a really

0:05:33.920 --> 0:05:35.280
<v Speaker 1>long time, and.

0:05:35.200 --> 0:05:39.599
<v Speaker 2>We're talking specifically about frontier models, not just open weighted models,

0:05:39.600 --> 0:05:39.960
<v Speaker 2>which was.

0:05:39.960 --> 0:05:42.440
<v Speaker 1>That's right, We've we've had open weight models for a while,

0:05:42.480 --> 0:05:44.960
<v Speaker 1>but yes, this is the frontier model. So starting in

0:05:44.960 --> 0:05:49.880
<v Speaker 1>preview today we have open AI's model five point four

0:05:50.000 --> 0:05:51.760
<v Speaker 1>and five to five is coming in the next couple

0:05:51.760 --> 0:05:56.479
<v Speaker 1>of weeks. We're also collaborating on a new offering which

0:05:56.480 --> 0:05:59.279
<v Speaker 1>we called Managed Agents featuring open Ai, which is a

0:05:59.279 --> 0:06:02.760
<v Speaker 1>complete manage Asian capabilities, so customers can really easily make

0:06:02.880 --> 0:06:07.160
<v Speaker 1>stateful agents and build agentic applications together with Bedrock and

0:06:07.200 --> 0:06:09.839
<v Speaker 1>open Ai. And it's really just the start of a

0:06:09.880 --> 0:06:13.280
<v Speaker 1>long term partnership that we've established together. As the teams

0:06:13.320 --> 0:06:15.680
<v Speaker 1>have gotten together, we see opportunity for us to really

0:06:15.720 --> 0:06:19.400
<v Speaker 1>invent new capabilities for our customers to go and build

0:06:19.440 --> 0:06:24.080
<v Speaker 1>interesting applications together. And really excited about working together with

0:06:24.160 --> 0:06:28.760
<v Speaker 1>the open ai team and unlocking more things that customers

0:06:28.760 --> 0:06:29.800
<v Speaker 1>can build on AWS.

0:06:30.000 --> 0:06:34.920
<v Speaker 2>Think back to twenty twenty three, open AIS models were

0:06:34.920 --> 0:06:40.200
<v Speaker 2>only available in Azure. Right since then until now, what

0:06:40.279 --> 0:06:44.560
<v Speaker 2>kind of behaviors did you recognize in the AWS customer base?

0:06:45.040 --> 0:06:48.120
<v Speaker 2>Did they literally open it as your accounts so that

0:06:48.160 --> 0:06:52.120
<v Speaker 2>they could have access to something that wasn't available in AWS?

0:06:52.160 --> 0:06:54.599
<v Speaker 2>And that was kind of the indicator to you that

0:06:54.720 --> 0:06:56.239
<v Speaker 2>you needed to make our codels.

0:06:56.520 --> 0:06:58.760
<v Speaker 1>Customers have long told us they want to be able

0:06:58.760 --> 0:07:01.320
<v Speaker 1>to use the absolute best cloud with the best set

0:07:01.320 --> 0:07:03.800
<v Speaker 1>of models available, and they wanted to run all their

0:07:03.800 --> 0:07:07.080
<v Speaker 1>applications in AWS, and they weren't able to run and

0:07:07.120 --> 0:07:09.479
<v Speaker 1>they did every and they did oftentimes, our customers run

0:07:09.480 --> 0:07:11.880
<v Speaker 1>every part of their application in AWS, and then they

0:07:12.240 --> 0:07:14.520
<v Speaker 1>access if they wanted to use open AA models, they'd

0:07:14.520 --> 0:07:16.400
<v Speaker 1>either use them direct from them in their first party

0:07:16.440 --> 0:07:18.800
<v Speaker 1>or through Azure or somewhere else. Now they don't have

0:07:18.880 --> 0:07:22.080
<v Speaker 1>to make that trade off. And with the open AI

0:07:22.160 --> 0:07:25.520
<v Speaker 1>team and the AWS team, we're really excited to not

0:07:25.800 --> 0:07:27.960
<v Speaker 1>have to force customers into that choice. And if they

0:07:27.960 --> 0:07:31.040
<v Speaker 1>want to build all in the AWS set of tools

0:07:31.040 --> 0:07:34.200
<v Speaker 1>and capabilities, it's where their applications run, it's where most

0:07:34.200 --> 0:07:37.680
<v Speaker 1>customers store their data, it's where they feel most secure,

0:07:38.440 --> 0:07:41.440
<v Speaker 1>and they can now run that together with open AI models,

0:07:41.440 --> 0:07:43.800
<v Speaker 1>together with the whole suite of models that we have

0:07:44.000 --> 0:07:49.200
<v Speaker 1>in AWS, including Anthropic models, including open models, including Nova models,

0:07:49.280 --> 0:07:51.280
<v Speaker 1>a whole set of models, and they can combine the

0:07:51.320 --> 0:07:52.560
<v Speaker 1>best of what they want to do in order to

0:07:52.560 --> 0:07:53.520
<v Speaker 1>build their applications.

0:07:53.680 --> 0:07:57.080
<v Speaker 2>It's early days I appreciate that, But what is the

0:07:57.120 --> 0:08:00.920
<v Speaker 2>sort of one metric you could point me too that

0:08:01.040 --> 0:08:04.400
<v Speaker 2>gives you the conviction there is absolutely demand there at

0:08:04.400 --> 0:08:07.880
<v Speaker 2>the enterprise level or otherwise for open AIS models.

0:08:08.840 --> 0:08:13.560
<v Speaker 1>Well, look, I think the team announced a statistic recently

0:08:13.920 --> 0:08:17.600
<v Speaker 1>that Codex, which is also available in AWS in previews

0:08:17.600 --> 0:08:20.400
<v Speaker 1>starting today in the last couple of weeks, has grown

0:08:20.440 --> 0:08:23.560
<v Speaker 1>from two million users to four million users in just

0:08:23.560 --> 0:08:24.800
<v Speaker 1>a really short amount of time.

0:08:25.120 --> 0:08:26.160
<v Speaker 3>That is incredible.

0:08:26.240 --> 0:08:29.000
<v Speaker 1>I mean, that is really unprecedented growth and it just

0:08:29.040 --> 0:08:31.840
<v Speaker 1>shows that they're building really, really capable.

0:08:31.800 --> 0:08:32.800
<v Speaker 3>Opportunities out there.

0:08:33.200 --> 0:08:35.680
<v Speaker 1>But our customers still say, I would love to access

0:08:35.760 --> 0:08:39.000
<v Speaker 1>and power Codex through Bedrock, so I have the security

0:08:39.040 --> 0:08:42.160
<v Speaker 1>of Bedrock and AWS. Now with the launches today, all

0:08:42.200 --> 0:08:44.440
<v Speaker 1>of those millions and millions of customers that are loving

0:08:44.480 --> 0:08:47.880
<v Speaker 1>Codex can choose Bedrock as the provider and get the

0:08:47.920 --> 0:08:49.560
<v Speaker 1>AWS security that comes behind that.

0:08:49.640 --> 0:08:53.280
<v Speaker 2>I think about Claude is a comparisonal case study, right, Yeah.

0:08:53.280 --> 0:08:56.760
<v Speaker 2>What was so interesting about claud through Bedrock is anthropic

0:08:56.800 --> 0:08:58.280
<v Speaker 2>got the benefit of a platform.

0:08:58.480 --> 0:08:58.800
<v Speaker 3>That's right.

0:08:59.240 --> 0:09:02.439
<v Speaker 2>You channeled business to them, that's right. But equally people

0:09:02.440 --> 0:09:05.680
<v Speaker 2>want to use Claude, so that channeled business to Bedrock.

0:09:05.720 --> 0:09:06.120
<v Speaker 3>That's right.

0:09:06.360 --> 0:09:09.040
<v Speaker 2>Do you see that playing out similarly with open Ai?

0:09:09.400 --> 0:09:11.040
<v Speaker 1>That's exactly what one of the things that the Opening

0:09:11.120 --> 0:09:13.280
<v Speaker 1>It team is so excited about is partnering with AWS.

0:09:13.640 --> 0:09:15.480
<v Speaker 1>Is that exact partnership And I think it's one of

0:09:15.520 --> 0:09:18.280
<v Speaker 1>the things that we do well at AWS is partner

0:09:18.520 --> 0:09:21.679
<v Speaker 1>with folks that we're building together with and just like

0:09:21.760 --> 0:09:24.880
<v Speaker 1>with Anthropic, which we have a fantastic partnership with, and

0:09:24.920 --> 0:09:27.840
<v Speaker 1>we've really built this business together with Anthropic where we

0:09:27.960 --> 0:09:31.000
<v Speaker 1>lean into them, we help their customers build applications on

0:09:31.040 --> 0:09:33.040
<v Speaker 1>top of AWS and find out the best ways to

0:09:33.120 --> 0:09:36.120
<v Speaker 1>use their claud models, and we've seen their business really

0:09:36.160 --> 0:09:38.760
<v Speaker 1>take off. And that's what we anticipate doing with open

0:09:38.760 --> 0:09:39.560
<v Speaker 1>a team as well.

0:09:39.880 --> 0:09:41.880
<v Speaker 2>Have you Matt had to make an assessment of how

0:09:41.960 --> 0:09:45.880
<v Speaker 2>much of the spending that Amazon's committed to is literally

0:09:45.920 --> 0:09:49.920
<v Speaker 2>dedicated to open Ai infrastructure wise to support the ramp

0:09:50.040 --> 0:09:51.359
<v Speaker 2>up of availability.

0:09:51.720 --> 0:09:55.280
<v Speaker 1>Yes, well, we absolutely are expecting a lot of growth there,

0:09:55.320 --> 0:09:58.440
<v Speaker 1>which we're quite excited about. And Opening Eyes made a

0:09:58.480 --> 0:10:02.000
<v Speaker 1>large commitment to AWS right to use Trainium in particular,

0:10:02.240 --> 0:10:05.040
<v Speaker 1>but across our AWS services to to grow their business

0:10:05.040 --> 0:10:08.920
<v Speaker 1>and so so yes, they're they're strongly committed to growing

0:10:08.960 --> 0:10:12.200
<v Speaker 1>their footprint on top of a WS and we expect

0:10:12.200 --> 0:10:13.800
<v Speaker 1>bedrock to grow really rapidly as well.

0:10:13.880 --> 0:10:17.080
<v Speaker 2>Does the demand meet the infrastructure that's been put in place?

0:10:17.360 --> 0:10:20.800
<v Speaker 2>Where are we in the balance between future supply and

0:10:21.280 --> 0:10:22.079
<v Speaker 2>current demand?

0:10:22.280 --> 0:10:24.560
<v Speaker 1>There is still I think more demand out there than

0:10:24.559 --> 0:10:26.200
<v Speaker 1>there is supply, right for sure?

0:10:26.559 --> 0:10:28.880
<v Speaker 3>And is that true of open AI. I can appreciate

0:10:28.920 --> 0:10:31.000
<v Speaker 3>were in preview as you speak.

0:10:30.760 --> 0:10:32.680
<v Speaker 1>Well, yes, it's it's it's the first day, so I

0:10:32.679 --> 0:10:34.720
<v Speaker 1>can't tell you that, yes, but but yes, the the

0:10:34.760 --> 0:10:37.880
<v Speaker 1>open A team, I think would would gladly take more

0:10:37.920 --> 0:10:39.760
<v Speaker 1>capacity from us this year and next year in the

0:10:39.840 --> 0:10:42.360
<v Speaker 1>year after that as we add it. And so we're

0:10:42.360 --> 0:10:46.239
<v Speaker 1>we're still working really hard to continue to add more capacity,

0:10:46.240 --> 0:10:49.400
<v Speaker 1>whether it's power, whether it's chips, whether it's memory, all

0:10:49.440 --> 0:10:51.880
<v Speaker 1>around the world. And I think that you know, we're

0:10:52.000 --> 0:10:57.000
<v Speaker 1>we're planning for success across many dimensions there. And but yes,

0:10:57.000 --> 0:11:00.360
<v Speaker 1>today the world is still largely supply constrains when it

0:11:00.360 --> 0:11:01.480
<v Speaker 1>comes to AI capacity.

0:11:01.720 --> 0:11:05.240
<v Speaker 2>We are in a strange situation where open AI models

0:11:05.240 --> 0:11:09.880
<v Speaker 2>can now run on aws, but Microsoft will benefit financially

0:11:09.920 --> 0:11:14.679
<v Speaker 2>because the latest terms of their agreement they've ended exclusivity,

0:11:14.720 --> 0:11:17.439
<v Speaker 2>but OpenAI continues to make payments to Microsoft.

0:11:17.559 --> 0:11:19.319
<v Speaker 3>Ye, how do you think about that?

0:11:19.760 --> 0:11:25.000
<v Speaker 1>Oh, that's okay. Look, Microsoft has benefited from the growth

0:11:25.000 --> 0:11:27.960
<v Speaker 1>of AWS since the very early days. In fact, we've

0:11:28.000 --> 0:11:31.199
<v Speaker 1>supported Windows licenses as an example, I think since two

0:11:31.240 --> 0:11:33.439
<v Speaker 1>thousand and seven, two thousand and eight, I can't remember

0:11:33.440 --> 0:11:36.880
<v Speaker 1>the exact year. But so Microsoft has benefited. They build

0:11:36.880 --> 0:11:39.520
<v Speaker 1>great software, they have good partnerships, and they should benefit

0:11:39.559 --> 0:11:42.880
<v Speaker 1>from those, but customers really want to use those technology.

0:11:42.960 --> 0:11:45.640
<v Speaker 1>SQL server is a good example that runs great on ABS.

0:11:45.679 --> 0:11:47.760
<v Speaker 1>In fact, many of our customers tell us that sql

0:11:47.760 --> 0:11:50.240
<v Speaker 1>server runs better, way better on AWS than it does

0:11:50.280 --> 0:11:54.160
<v Speaker 1>on Azure. And so for those customers, running an ABS

0:11:54.200 --> 0:11:56.800
<v Speaker 1>is where they get the best reliability, the most security,

0:11:56.840 --> 0:11:58.079
<v Speaker 1>and the broadest set of features.

0:11:58.120 --> 0:11:59.240
<v Speaker 3>And so that's great.

0:11:59.240 --> 0:12:02.800
<v Speaker 1>And if Microsoft benefits because their software or their partnerships

0:12:02.840 --> 0:12:05.319
<v Speaker 1>or their licenses are being used inside of ABS, that

0:12:05.520 --> 0:12:08.920
<v Speaker 1>that's great. And in this case, we partner with Microsoft

0:12:09.160 --> 0:12:12.319
<v Speaker 1>just like we partner with Oracle and other providers that

0:12:12.600 --> 0:12:15.720
<v Speaker 1>build software, and we want to make all capabilities available

0:12:15.720 --> 0:12:17.000
<v Speaker 1>on ABS for people to build.

0:12:17.320 --> 0:12:19.559
<v Speaker 3>You built bedrock and.

0:12:21.800 --> 0:12:26.400
<v Speaker 2>Stateful run time at a time where access to models

0:12:26.600 --> 0:12:30.439
<v Speaker 2>was constrained. Now, if open ai is the case, it's

0:12:30.440 --> 0:12:36.119
<v Speaker 2>available broadly everywhere. You know, what's the benefit of accessing

0:12:36.160 --> 0:12:38.800
<v Speaker 2>it through AWS? What's the pitch really? I mean for

0:12:38.880 --> 0:12:41.920
<v Speaker 2>the and we're calling this Managed Agents is managed Eliden.

0:12:42.040 --> 0:12:44.360
<v Speaker 2>We called it Stateful Runtime before, but the launch name

0:12:44.400 --> 0:12:47.280
<v Speaker 2>of it is called Managed Agents, and it's just we're

0:12:47.320 --> 0:12:48.960
<v Speaker 2>just as excited about that as we've ever been. By

0:12:49.000 --> 0:12:51.079
<v Speaker 2>the way, so that that that product.

0:12:50.760 --> 0:12:52.400
<v Speaker 1>Is something that we Why does a customer need to

0:12:52.480 --> 0:12:54.480
<v Speaker 1>use ABUS? That's the simple question, why do you want

0:12:54.480 --> 0:12:56.240
<v Speaker 1>to use AWS? Well, this is the only place that

0:12:56.280 --> 0:12:59.400
<v Speaker 1>we'll have this managed Agent's product together with with open Ai,

0:13:01.240 --> 0:13:03.199
<v Speaker 1>and we think it's going to be the single best

0:13:03.400 --> 0:13:05.559
<v Speaker 1>and easiest way for you to go and build agentic

0:13:05.600 --> 0:13:08.400
<v Speaker 1>applications that learn with you over time, that get better

0:13:08.440 --> 0:13:12.120
<v Speaker 1>over time. It's a really simple way to It includes

0:13:12.160 --> 0:13:14.000
<v Speaker 1>kind of this agentic harness and a bunch of these

0:13:14.000 --> 0:13:17.000
<v Speaker 1>pieces together that makes it really easy to build agents

0:13:17.040 --> 0:13:20.160
<v Speaker 1>that leverage the latest frontier models from open Ai. And

0:13:20.200 --> 0:13:22.640
<v Speaker 1>so we're very excited about that product. We think that

0:13:22.640 --> 0:13:24.800
<v Speaker 1>that is going to take off. Like investors, it's launching

0:13:24.840 --> 0:13:27.319
<v Speaker 1>in preview today as well, and it'll be ga here

0:13:27.320 --> 0:13:29.800
<v Speaker 1>in the next couple of weeks, and we think that

0:13:30.679 --> 0:13:32.720
<v Speaker 1>customers are going to really love how easy it is

0:13:32.760 --> 0:13:34.920
<v Speaker 1>to build agentic applications with that product.