WEBVTT - Dell CEO Michael Dell & Nvidia CEO Jensen Huang Talk Agentic AI, Memory Demand & China

0:00:02.520 --> 0:00:07.800
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news.

0:00:07.880 --> 0:00:13.040
<v Speaker 2>Gentlemen, Good morning, Hello Michael in the court of gone.

0:00:13.480 --> 0:00:17.200
<v Speaker 3>One thousand new clients for AI server, for AI factory.

0:00:17.640 --> 0:00:18.520
<v Speaker 2>It's a hell of a jump.

0:00:19.760 --> 0:00:22.599
<v Speaker 3>What is it that those new clients fight thousand total

0:00:22.640 --> 0:00:26.000
<v Speaker 3>are actually building now different to one year ago.

0:00:26.040 --> 0:00:27.680
<v Speaker 2>I think that's probably a good place to start.

0:00:29.400 --> 0:00:32.960
<v Speaker 4>I think the change we see is it's kind of

0:00:33.040 --> 0:00:37.640
<v Speaker 4>moved from testing and evaluating into production. And we showed

0:00:37.640 --> 0:00:41.239
<v Speaker 4>some great examples on stage right with Eli Lilly with

0:00:41.280 --> 0:00:44.960
<v Speaker 4>one thousand gpussal world. It's Samsung, and these are not

0:00:45.760 --> 0:00:47.720
<v Speaker 4>things that are on the screen, right, This is in

0:00:47.760 --> 0:00:50.680
<v Speaker 4>the real world with the largest companies in the world,

0:00:50.800 --> 0:00:57.160
<v Speaker 4>and so it's propagating broadly across all customers in every

0:00:57.200 --> 0:01:02.000
<v Speaker 4>industry and every country. And you know, you see the

0:01:02.120 --> 0:01:05.400
<v Speaker 4>proovement in all the models, and now we have the

0:01:05.440 --> 0:01:09.400
<v Speaker 4>augentic capabilities, and so while it is exciting, there's been

0:01:09.400 --> 0:01:12.039
<v Speaker 4>a tremendous amount of growth, I still think it's just

0:01:12.080 --> 0:01:16.240
<v Speaker 4>the beginning of this wave, particularly when it comes to enterprise,

0:01:16.800 --> 0:01:20.440
<v Speaker 4>which is really where you know, we have an enormous opportunity.

0:01:20.520 --> 0:01:25.200
<v Speaker 3>What's so fascinating, Jensen, is you spent four years telling

0:01:25.200 --> 0:01:27.440
<v Speaker 3>me that we needed to change the definition of the

0:01:27.520 --> 0:01:30.200
<v Speaker 3>computer in context of accelerating computing.

0:01:30.520 --> 0:01:33.080
<v Speaker 2>But the big focus was on the Hyperscali's right cloud.

0:01:33.520 --> 0:01:36.440
<v Speaker 3>What I took from Michael's presentation was this is happening

0:01:36.600 --> 0:01:41.160
<v Speaker 3>you said locally, but on trend. What's the nvidior interpretation

0:01:41.319 --> 0:01:43.840
<v Speaker 3>of that part of this site.

0:01:44.440 --> 0:01:50.480
<v Speaker 5>Intelligence has to be performed produced at the point of context,

0:01:51.800 --> 0:01:56.360
<v Speaker 5>and so wherever the context is, wherever the action is,

0:01:56.880 --> 0:01:57.440
<v Speaker 5>that's where you.

0:01:57.400 --> 0:01:58.720
<v Speaker 1>Want to produce the intelligence.

0:01:59.160 --> 0:02:01.880
<v Speaker 5>For most of the early applications of AI, it was

0:02:01.920 --> 0:02:02.480
<v Speaker 5>in the cloud.

0:02:02.920 --> 0:02:05.040
<v Speaker 1>A lot of consumer services are in the cloud.

0:02:05.520 --> 0:02:11.120
<v Speaker 5>However, for Lily, Samsung, the future manufacturing, a lot of companies,

0:02:11.639 --> 0:02:14.480
<v Speaker 5>you want the agents to be on prem because that's

0:02:14.520 --> 0:02:17.400
<v Speaker 5>where all of your data is, where all your secure

0:02:17.480 --> 0:02:20.840
<v Speaker 5>data is, You're proprietary data, and all of the skills

0:02:20.840 --> 0:02:23.600
<v Speaker 5>associated with your company is. And so now we have

0:02:23.960 --> 0:02:27.600
<v Speaker 5>agents that are here, AIS that can do work right.

0:02:27.880 --> 0:02:31.239
<v Speaker 5>Chat GPT was fantastic a launched generative AI.

0:02:31.080 --> 0:02:33.040
<v Speaker 2>But you just hate content. That was it.

0:02:33.680 --> 0:02:38.320
<v Speaker 5>Making content is very important, but doing work is really valuable.

0:02:38.360 --> 0:02:41.680
<v Speaker 5>And now we're doing productive work incredibly well, that's why

0:02:41.720 --> 0:02:43.680
<v Speaker 5>they're called agentic AI in this.

0:02:43.800 --> 0:02:46.440
<v Speaker 3>New era, what everyone is trying to work out is

0:02:46.760 --> 0:02:50.280
<v Speaker 3>aren't all the GPUs locked up at the hyperscalers. How

0:02:50.360 --> 0:02:54.360
<v Speaker 3>is Michael Dell gonna service those one thousand new clients

0:02:54.919 --> 0:02:58.000
<v Speaker 3>with the GPUs to build their own on frame local

0:02:58.040 --> 0:03:00.080
<v Speaker 3>AI factory.

0:03:00.440 --> 0:03:04.360
<v Speaker 4>Well, the supply chain that Jensen has built, we've built together,

0:03:04.680 --> 0:03:07.640
<v Speaker 4>is continuing to scale up. And while it's true that

0:03:07.680 --> 0:03:12.720
<v Speaker 4>there's more demand than supply, there's more supply that's being added,

0:03:12.840 --> 0:03:18.560
<v Speaker 4>and you know, customers are figuring out how they start

0:03:18.600 --> 0:03:22.919
<v Speaker 4>to scale these systems up. So, you know, I think

0:03:22.960 --> 0:03:28.799
<v Speaker 4>what's also happening is companies are understanding that when they

0:03:29.320 --> 0:03:34.800
<v Speaker 4>reimagine their workflows using this technology, they don't get ten

0:03:34.880 --> 0:03:37.640
<v Speaker 4>or twenty or thirty percent improvement. They get ten times

0:03:37.800 --> 0:03:41.040
<v Speaker 4>or twenty times or one hundred times, And that is

0:03:41.120 --> 0:03:44.120
<v Speaker 4>really the speed that matters to make up as a

0:03:44.160 --> 0:03:47.520
<v Speaker 4>successful We're doing it ourselves, and Video is doing it,

0:03:47.920 --> 0:03:50.440
<v Speaker 4>and so it's not a secret anymore that these things

0:03:50.440 --> 0:03:55.280
<v Speaker 4>are possible, and every company wants to capture that speed

0:03:55.800 --> 0:03:58.360
<v Speaker 4>and translate it into competitive advantage and outcomes.

0:04:00.160 --> 0:04:03.240
<v Speaker 3>L was the sales channel, right Jensen. Michael's company is

0:04:03.320 --> 0:04:08.480
<v Speaker 3>very good to selling technology to America's biggest companies. How

0:04:08.520 --> 0:04:10.560
<v Speaker 3>is that going to change things from VideA going forward?

0:04:10.640 --> 0:04:13.080
<v Speaker 3>Like the makeup of the types of companies we're talking

0:04:13.120 --> 0:04:16.800
<v Speaker 3>about are at scale, but there's also that kind of

0:04:16.839 --> 0:04:20.799
<v Speaker 3>middle market of data center that's being filled different kinds

0:04:20.800 --> 0:04:22.560
<v Speaker 3>of in the industrial.

0:04:22.080 --> 0:04:23.720
<v Speaker 2>Space in healthcare.

0:04:24.400 --> 0:04:27.279
<v Speaker 3>Is that something that puts a video into new territory

0:04:27.520 --> 0:04:30.839
<v Speaker 3>away from the frontier labs, away from the hyperscalers.

0:04:32.360 --> 0:04:38.440
<v Speaker 5>Well, in videos a technology company, right, the hyperscalers have

0:04:38.520 --> 0:04:40.640
<v Speaker 5>the ability to take our technology.

0:04:40.279 --> 0:04:43.039
<v Speaker 1>And integrate them, operate them into a service.

0:04:43.720 --> 0:04:46.640
<v Speaker 5>Dell has the ability to take our technology turn them

0:04:46.680 --> 0:04:49.880
<v Speaker 5>into a solution that the lover's impact the customers.

0:04:50.200 --> 0:04:50.960
<v Speaker 1>If you look at what.

0:04:50.920 --> 0:04:54.480
<v Speaker 5>Has happened, agentic AI has completely As we were talking

0:04:54.520 --> 0:04:55.640
<v Speaker 5>about earlier.

0:04:55.440 --> 0:04:58.600
<v Speaker 1>Reinvent a computer, we had to do several things together.

0:04:59.600 --> 0:05:01.679
<v Speaker 1>The first, of course, we have to build the brain.

0:05:02.200 --> 0:05:05.479
<v Speaker 5>This is the Grace Blackwell ENDLINGK seventy two, the Barrel

0:05:05.520 --> 0:05:09.760
<v Speaker 5>Ruben Mulling seventy two, giant large language models. The second

0:05:09.760 --> 0:05:13.159
<v Speaker 5>part now is the VERA CPU that we're now in

0:05:13.200 --> 0:05:16.520
<v Speaker 5>the process of launching. The fact the highest performance CPU

0:05:16.520 --> 0:05:20.400
<v Speaker 5>in the world. It's designed for agentic AI and now

0:05:20.600 --> 0:05:25.359
<v Speaker 5>this will be the harness running the agent itself using

0:05:25.400 --> 0:05:25.960
<v Speaker 5>the tools.

0:05:26.560 --> 0:05:28.760
<v Speaker 2>The third part, what does harness mean?

0:05:28.839 --> 0:05:32.960
<v Speaker 1>What harness harness is what UH puts. A puts a.

0:05:32.880 --> 0:05:36.160
<v Speaker 5>Harness around the large language model so that it can

0:05:36.200 --> 0:05:40.520
<v Speaker 5>access memory, access the network, use tools.

0:05:40.360 --> 0:05:44.719
<v Speaker 1>Have local scratch, bad memory, working memory, access long term memory.

0:05:45.080 --> 0:05:49.440
<v Speaker 5>And so that harness basically turns, if you will, the

0:05:49.480 --> 0:05:53.520
<v Speaker 5>brain into an agent, okay, into a digital robot.

0:05:53.560 --> 0:05:55.920
<v Speaker 1>If you will, that could do work. And so now

0:05:56.160 --> 0:05:59.040
<v Speaker 1>the agent runs on a CPU. We also worked with.

0:05:59.040 --> 0:06:01.960
<v Speaker 5>Dell to create a new type of long term memory

0:06:01.960 --> 0:06:06.680
<v Speaker 5>for agents called the Dell AI Data platform that's built

0:06:06.680 --> 0:06:09.479
<v Speaker 5>on in video. The networking to scale it out is

0:06:09.480 --> 0:06:13.440
<v Speaker 5>built on in video. So the agent, the brain, the

0:06:13.520 --> 0:06:17.080
<v Speaker 5>long term memory, all of the networking necessary to scale

0:06:17.120 --> 0:06:21.400
<v Speaker 5>it up, as well as the agent run time itself.

0:06:21.440 --> 0:06:23.800
<v Speaker 1>We call Nemo CLAW running in a.

0:06:24.000 --> 0:06:29.560
<v Speaker 5>Secure and governed container called open shelf. All of that

0:06:29.600 --> 0:06:33.120
<v Speaker 5>has been put together, and now the technologies are the

0:06:33.200 --> 0:06:34.320
<v Speaker 5>technology parts.

0:06:34.839 --> 0:06:37.080
<v Speaker 1>What Dell has to do is turn it into a

0:06:37.120 --> 0:06:38.479
<v Speaker 1>solution that people can use.

0:06:38.800 --> 0:06:42.760
<v Speaker 5>Dell will do for the world's enterprises what the clouds

0:06:42.839 --> 0:06:45.360
<v Speaker 5>do for the clouds makes perfect sense.

0:06:45.560 --> 0:06:48.920
<v Speaker 3>What is the Dell story, Michael, around CPU and sort

0:06:48.960 --> 0:06:52.640
<v Speaker 3>of like general purpose computing in the agentic era. We've

0:06:52.640 --> 0:06:57.320
<v Speaker 3>talked a lot about the AI factory offering the GPU,

0:06:57.800 --> 0:07:02.400
<v Speaker 3>but actually there's potential for you in more general purpose workloads.

0:07:02.640 --> 0:07:05.679
<v Speaker 2>The buildouts happening either way.

0:07:05.600 --> 0:07:08.600
<v Speaker 4>It is, and and the demand is exceeding to supply

0:07:08.680 --> 0:07:11.600
<v Speaker 4>there as well as you know. And look as you

0:07:12.040 --> 0:07:18.360
<v Speaker 4>move to these agent frameworks inside companies.

0:07:18.760 --> 0:07:20.320
<v Speaker 1>You use a lot more CPUs.

0:07:20.720 --> 0:07:25.480
<v Speaker 4>Yeah, and uh, you know, that's that's just the reality

0:07:25.520 --> 0:07:26.760
<v Speaker 4>of what's what's happening.

0:07:26.960 --> 0:07:30.679
<v Speaker 1>And I think, I think that's only gonna good increase.

0:07:30.720 --> 0:07:36.840
<v Speaker 5>So instead of humans using tools, it's now agents using tools,

0:07:37.360 --> 0:07:40.840
<v Speaker 5>and agents, as you were talking about earlier on stage there,

0:07:40.840 --> 0:07:43.560
<v Speaker 5>we're gonna have we have a billion people, will have

0:07:43.920 --> 0:07:48.120
<v Speaker 5>hundreds of billions of agents. People use tools every now

0:07:48.160 --> 0:07:51.000
<v Speaker 5>and then agents are gonna use tools all the time,

0:07:51.360 --> 0:07:54.360
<v Speaker 5>and agents use tools very quickly, and so we're gonna

0:07:54.400 --> 0:07:57.240
<v Speaker 5>need a lot more CPUs. And those CPUs are connected

0:07:57.240 --> 0:08:00.880
<v Speaker 5>to GPU brains so that the CPUs know how to think,

0:08:01.080 --> 0:08:03.240
<v Speaker 5>how to reason, how to plan, and how to use

0:08:03.280 --> 0:08:03.920
<v Speaker 5>those tools.

0:08:04.360 --> 0:08:05.720
<v Speaker 1>So that's basically how it works.

0:08:06.040 --> 0:08:09.440
<v Speaker 2>Gentlemen, What is the biggest supply constraint. What will meant

0:08:09.440 --> 0:08:10.120
<v Speaker 2>for you right now?

0:08:12.360 --> 0:08:15.880
<v Speaker 4>Well, certainly, you know memory is a challenge.

0:08:15.400 --> 0:08:16.480
<v Speaker 2>I think it is memory.

0:08:16.680 --> 0:08:19.560
<v Speaker 4>The advanced node semi conductors are still challenging.

0:08:20.560 --> 0:08:23.960
<v Speaker 1>You know. It's it's really I mean, we think about

0:08:24.000 --> 0:08:24.840
<v Speaker 1>it from the things that.

0:08:24.800 --> 0:08:31.720
<v Speaker 4>We're producing, and the semiconductor supply chain is ramping, but

0:08:31.920 --> 0:08:34.480
<v Speaker 4>the demand's growing faster than the supply.

0:08:34.920 --> 0:08:39.760
<v Speaker 5>In our case, we provide the technology integrating and so

0:08:39.840 --> 0:08:43.080
<v Speaker 5>the memory comes with our technology. We've been planning our

0:08:43.120 --> 0:08:45.240
<v Speaker 5>supply chain for a couple of two three years. We

0:08:45.320 --> 0:08:48.280
<v Speaker 5>have the largest supply chain in the world. Our partners

0:08:48.320 --> 0:08:52.200
<v Speaker 5>have done a great job securing supply for us, and

0:08:52.280 --> 0:08:55.480
<v Speaker 5>so all of the pieces go together. The co OSS

0:08:55.600 --> 0:08:57.960
<v Speaker 5>is lined up with the HBM, which is lined up

0:08:58.000 --> 0:09:00.760
<v Speaker 5>with a grace black weals, and the CPU as a

0:09:00.880 --> 0:09:03.800
<v Speaker 5>co OS are the COSS L the COSS. All of

0:09:03.840 --> 0:09:06.840
<v Speaker 5>it is all lined up. The silicon photonics is lined up,

0:09:06.960 --> 0:09:09.280
<v Speaker 5>or everything is all lined up. It's just that the

0:09:09.320 --> 0:09:13.240
<v Speaker 5>demand is much greater than the overall capacity of the world.

0:09:13.320 --> 0:09:16.720
<v Speaker 3>So the overall capacity Jensen, should I put my textbook away,

0:09:16.760 --> 0:09:18.920
<v Speaker 3>because if I get my textbook out, it tells me

0:09:19.080 --> 0:09:23.240
<v Speaker 3>that memory historically is cyclical, it's boom and bus, and

0:09:23.280 --> 0:09:27.720
<v Speaker 3>so you both kind of have to convince the memory

0:09:27.760 --> 0:09:31.600
<v Speaker 3>makers of the permanency of this to build the capacity

0:09:32.080 --> 0:09:34.679
<v Speaker 3>that won't sort of fool away. Is that the right

0:09:34.720 --> 0:09:36.560
<v Speaker 3>way of looking at it, that this is not a

0:09:36.600 --> 0:09:39.520
<v Speaker 3>boom and bus cycle, it's just a complete change in

0:09:39.559 --> 0:09:42.080
<v Speaker 3>the structure of that market.

0:09:42.240 --> 0:09:43.760
<v Speaker 1>Well, Michael and I do this all the time. We

0:09:43.840 --> 0:09:45.320
<v Speaker 1>spent a lot of time with the supply chain.

0:09:45.679 --> 0:09:48.840
<v Speaker 5>I mean, if you ask Sanjay Metro over a over

0:09:48.880 --> 0:09:52.959
<v Speaker 5>a Micron, they'll tell you three years ago during a meeting,

0:09:53.000 --> 0:09:55.160
<v Speaker 5>I explained the future to them exactly as it is

0:09:55.200 --> 0:09:58.680
<v Speaker 5>happening right now, and I was really grateful that the

0:09:58.840 --> 0:10:01.240
<v Speaker 5>Micron and video really lined up to.

0:10:01.320 --> 0:10:02.720
<v Speaker 1>Line up all of our roadmap.

0:10:03.080 --> 0:10:05.800
<v Speaker 5>Tony will tell you over at sk that we did

0:10:05.840 --> 0:10:09.560
<v Speaker 5>the same thing years before. And so it's our job

0:10:09.600 --> 0:10:11.880
<v Speaker 5>to make sure that the vision of the future the

0:10:12.000 --> 0:10:16.400
<v Speaker 5>industry we convey upstream to our supply chain so that

0:10:16.440 --> 0:10:19.520
<v Speaker 5>they are building for it. We also have to convey

0:10:19.520 --> 0:10:22.880
<v Speaker 5>it downstream to people who have power generators and land

0:10:22.960 --> 0:10:25.120
<v Speaker 5>and financing and so on and so forth, and so

0:10:25.559 --> 0:10:27.840
<v Speaker 5>we have to make sure that the supply chain upstream

0:10:27.840 --> 0:10:28.600
<v Speaker 5>and downstream are.

0:10:28.559 --> 0:10:29.600
<v Speaker 1>Prepared for this future.

0:10:29.679 --> 0:10:33.880
<v Speaker 5>It is true that the simple logic is this that

0:10:33.960 --> 0:10:37.880
<v Speaker 5>we have now reached a level of agentic AI useful

0:10:37.920 --> 0:10:42.319
<v Speaker 5>AI productive AI capability, and the way to think about.

0:10:42.040 --> 0:10:44.800
<v Speaker 1>These agents is kind of like just digital workers.

0:10:44.960 --> 0:10:48.280
<v Speaker 5>Right, we have hundreds of millions of digital workers in

0:10:48.320 --> 0:10:52.280
<v Speaker 5>the world. We're going to have billions of AI agents

0:10:52.280 --> 0:10:54.160
<v Speaker 5>in the world, and they're going to be working twenty

0:10:54.160 --> 0:10:56.720
<v Speaker 5>four to seven. And so just as we give every

0:10:56.840 --> 0:11:00.959
<v Speaker 5>digital worker a laptop and a small to the data center,

0:11:01.240 --> 0:11:04.280
<v Speaker 5>we're gonna have to give every agent essentially a computer

0:11:04.559 --> 0:11:06.199
<v Speaker 5>and a little bit of stories in the data center

0:11:06.240 --> 0:11:06.559
<v Speaker 5>to use.

0:11:06.720 --> 0:11:07.640
<v Speaker 1>Think about it this way.

0:11:07.720 --> 0:11:11.360
<v Speaker 4>You know you do individual work, you know as a person,

0:11:11.960 --> 0:11:13.720
<v Speaker 4>and you send it on to somebody else, and you

0:11:13.720 --> 0:11:14.800
<v Speaker 4>know there's interactions.

0:11:15.080 --> 0:11:18.120
<v Speaker 1>Well, now you might have hundreds or thousands.

0:11:17.640 --> 0:11:21.720
<v Speaker 4>Of you know, digital agents working for ED right supervised

0:11:21.960 --> 0:11:24.679
<v Speaker 4>that you supervise, and that's going to help you be

0:11:24.800 --> 0:11:29.480
<v Speaker 4>way more productive, get way more things done, expand your creativity. Now,

0:11:29.480 --> 0:11:32.400
<v Speaker 4>it does require a lot more computing and memory and

0:11:32.440 --> 0:11:35.440
<v Speaker 4>storage and networking and all the things that we're doing together.

0:11:35.520 --> 0:11:38.360
<v Speaker 3>Last one on this site, Jensen outline being the Micron

0:11:38.440 --> 0:11:41.080
<v Speaker 3>and the ESK example, three years ago, you gave them

0:11:41.160 --> 0:11:41.719
<v Speaker 3>the heads up.

0:11:42.000 --> 0:11:44.480
<v Speaker 2>Do they believe you? Are they sort of acting.

0:11:44.160 --> 0:11:45.280
<v Speaker 1>On that they're investing?

0:11:45.320 --> 0:11:48.920
<v Speaker 4>I mean, it's we're managing through it. But these things

0:11:48.920 --> 0:11:52.040
<v Speaker 4>are very hard to predict. Right if you tried to predict,

0:11:52.600 --> 0:11:54.880
<v Speaker 4>you know, in twenty twenty three, what the demand was

0:11:54.920 --> 0:11:57.440
<v Speaker 4>going to be in twenty twenty seven, you would have

0:11:57.440 --> 0:11:59.560
<v Speaker 4>a hard time doing that. So it does take a

0:11:59.559 --> 0:12:02.440
<v Speaker 4>long time to build these factories. But we've got great

0:12:02.480 --> 0:12:07.280
<v Speaker 4>relationships with these partners we have for decades. That's helping us,

0:12:07.520 --> 0:12:09.720
<v Speaker 4>and they see that we're winning and so they want

0:12:09.760 --> 0:12:13.760
<v Speaker 4>to work with us even more. And it's really a

0:12:13.800 --> 0:12:17.160
<v Speaker 4>great long term partnership, even though we'd like more.

0:12:17.320 --> 0:12:20.079
<v Speaker 5>Right now, we're in the beginning of the AI build out.

0:12:20.400 --> 0:12:23.400
<v Speaker 5>This is literally the very beginning of the agentic AI

0:12:23.440 --> 0:12:25.880
<v Speaker 5>build out. We're gonna be building this out for a

0:12:25.920 --> 0:12:29.360
<v Speaker 5>decade maybe more, because after this digital agents will be

0:12:29.400 --> 0:12:30.280
<v Speaker 5>a physical agents.

0:12:30.320 --> 0:12:31.640
<v Speaker 1>When we go to the physical AI.

0:12:32.160 --> 0:12:34.080
<v Speaker 4>We haven't even started that. I mean, you saw some

0:12:35.120 --> 0:12:38.160
<v Speaker 4>examples of that, you know, in the keynote, But that

0:12:38.280 --> 0:12:42.080
<v Speaker 4>is a way bigger market and it will require all sorts.

0:12:41.840 --> 0:12:44.560
<v Speaker 1>Of new infrastructure capabilities. We're gonna for the very first.

0:12:44.320 --> 0:12:48.120
<v Speaker 5>Time bring it to the world's ninety trillion other industry,

0:12:48.880 --> 0:12:51.160
<v Speaker 5>and so there's a giant industry ahead of us to

0:12:51.160 --> 0:12:51.840
<v Speaker 5>build towards.

0:12:52.160 --> 0:12:52.319
<v Speaker 1>Now.

0:12:52.360 --> 0:12:55.720
<v Speaker 5>Meanwhile, the supply chain is more than doubling every year.

0:12:56.080 --> 0:12:59.120
<v Speaker 5>I mean, it's probably quadrupling every year. But we'll still

0:12:59.160 --> 0:13:01.400
<v Speaker 5>have a hard time keep up with the build out

0:13:01.720 --> 0:13:02.720
<v Speaker 5>for at least a decade.

0:13:02.760 --> 0:13:07.560
<v Speaker 3>My sense, China, Jensen, you just returned from China on

0:13:07.679 --> 0:13:11.079
<v Speaker 3>Friday on Air Force one. The President said that H

0:13:11.160 --> 0:13:14.280
<v Speaker 3>two hundred came up, but that China's position is it

0:13:14.360 --> 0:13:17.400
<v Speaker 3>wants to support its own industry. Could I just ask

0:13:17.679 --> 0:13:20.760
<v Speaker 3>what the net outcome was of your trip to China

0:13:20.800 --> 0:13:23.920
<v Speaker 3>and your understanding of what is not or is allowed

0:13:23.920 --> 0:13:26.600
<v Speaker 3>with H two hundred in the customers that you have

0:13:26.840 --> 0:13:28.320
<v Speaker 3>or do not have in China.

0:13:28.440 --> 0:13:33.040
<v Speaker 5>The President wants America to win everywhere, right, the President

0:13:33.040 --> 0:13:37.480
<v Speaker 5>wants America to lead the AI revolution, and so H

0:13:37.520 --> 0:13:42.480
<v Speaker 5>two hundreds are licensed to sell to China. The Chinese

0:13:42.520 --> 0:13:45.920
<v Speaker 5>government has to decide how much of their local market

0:13:45.960 --> 0:13:47.840
<v Speaker 5>do they want to protect and how much of their

0:13:47.880 --> 0:13:51.640
<v Speaker 5>local market do they want to expand with more AI capacity.

0:13:52.320 --> 0:13:55.239
<v Speaker 5>My sense is that the demand in China is so incredible,

0:13:55.640 --> 0:13:56.839
<v Speaker 5>just like it is here.

0:13:57.200 --> 0:14:01.040
<v Speaker 1>Agentic Ai is also making enormous product us there. My

0:14:01.160 --> 0:14:04.480
<v Speaker 1>sense is that over time the market will open.

0:14:05.440 --> 0:14:08.320
<v Speaker 5>President she was very clear that he wants China to

0:14:08.360 --> 0:14:13.880
<v Speaker 5>be an even wider open market. Premier Lee Chang was

0:14:14.040 --> 0:14:19.040
<v Speaker 5>very straightforward and to explain very eloquently that that China

0:14:19.080 --> 0:14:20.960
<v Speaker 5>will be an open market. So I'm looking forward to

0:14:21.080 --> 0:14:21.920
<v Speaker 5>China being a more.

0:14:21.800 --> 0:14:24.480
<v Speaker 3>Openly clarified as you were able to meet with those

0:14:24.520 --> 0:14:28.120
<v Speaker 3>officials directly to discuss whether or not you can sell

0:14:28.200 --> 0:14:30.080
<v Speaker 3>to those Chinese tech companies.

0:14:30.120 --> 0:14:30.400
<v Speaker 1>I did.

0:14:30.440 --> 0:14:33.600
<v Speaker 5>I didn't discuss directly with him about age two hundred, right.

0:14:33.640 --> 0:14:35.720
<v Speaker 5>I was there to represent the United States, and I

0:14:35.720 --> 0:14:37.600
<v Speaker 5>was honored to do so. I was there to support

0:14:37.680 --> 0:14:40.240
<v Speaker 5>President Trump and really glad to do so.

0:14:41.000 --> 0:14:44.480
<v Speaker 1>But that was really the focus of my trip. President

0:14:44.560 --> 0:14:45.200
<v Speaker 1>Trump had.

0:14:45.040 --> 0:14:49.520
<v Speaker 5>Some conversations with the leaders, and I'm looking forward to

0:14:49.520 --> 0:14:50.600
<v Speaker 5>to what they decide.

0:14:50.880 --> 0:14:52.880
<v Speaker 3>Michael, you did not go to China, But I think

0:14:52.960 --> 0:14:55.360
<v Speaker 3>what's interesting is you are a member of the President's

0:14:55.360 --> 0:14:59.840
<v Speaker 3>Council Advisors for Science and Technology, as is Jensen. You'll

0:15:00.120 --> 0:15:03.040
<v Speaker 3>or net conclusion on whether or not China will become

0:15:03.160 --> 0:15:06.920
<v Speaker 3>open to American technology companies to do business there.

0:15:08.560 --> 0:15:11.560
<v Speaker 1>You know, we have a bus in China. Obviously, we

0:15:12.040 --> 0:15:12.960
<v Speaker 1>comply with all.

0:15:12.920 --> 0:15:18.320
<v Speaker 4>The restrictions and you know, various controls that are in place.

0:15:19.080 --> 0:15:23.320
<v Speaker 4>But I hope that there's more economic collaboration between the

0:15:23.400 --> 0:15:26.160
<v Speaker 4>United States and China that ultimately as well lead to

0:15:27.160 --> 0:15:30.920
<v Speaker 4>greater outcomes of prosperity for everyone and you know, a

0:15:30.960 --> 0:15:35.360
<v Speaker 4>greater likelihood of you know, a successful relationship between the

0:15:35.400 --> 0:15:38.280
<v Speaker 4>countries and you know, around the world.

0:15:39.160 --> 0:15:42.080
<v Speaker 3>The final question on that trip, Jensen, is the sharpest

0:15:42.160 --> 0:15:46.400
<v Speaker 3>rhetoric was probably on Taiwan. We've talked about the supply chain,

0:15:46.880 --> 0:15:49.240
<v Speaker 3>but but what did you take from those comments from

0:15:49.280 --> 0:15:51.320
<v Speaker 3>from President g On on.

0:15:51.240 --> 0:15:52.320
<v Speaker 2>The issue of Taiwan.

0:15:52.440 --> 0:15:56.280
<v Speaker 3>Of course, from a manufacturing capacity standpoint, TESSEMC as a

0:15:56.320 --> 0:16:00.000
<v Speaker 3>critical partner. You and I have discussed it in the past,

0:16:00.040 --> 0:16:03.359
<v Speaker 3>about it at this moment in time, how top of mind.

0:16:03.160 --> 0:16:05.760
<v Speaker 2>Is a view the security of supply from Taiwan.

0:16:05.960 --> 0:16:08.320
<v Speaker 5>We none of us were involved in any of those

0:16:08.320 --> 0:16:10.800
<v Speaker 5>conversations except for President Trump.

0:16:11.800 --> 0:16:12.760
<v Speaker 1>With respect to Taiwan.

0:16:13.040 --> 0:16:17.360
<v Speaker 5>Obviously, Taiwan is still epicenter of the world's technology manufacturing

0:16:17.360 --> 0:16:21.120
<v Speaker 5>and technology development. The supply chain is rich in Taiwan.

0:16:22.000 --> 0:16:26.040
<v Speaker 5>We're also, of course reindustrializing the United States, bringing manufacturing

0:16:26.080 --> 0:16:29.120
<v Speaker 5>back to the United States. We're doing so at a

0:16:29.160 --> 0:16:32.040
<v Speaker 5>time when demand for AI and this beginning of this

0:16:32.080 --> 0:16:36.400
<v Speaker 5>new computer revolution is happening, and so demand is extraordinary.

0:16:36.800 --> 0:16:39.640
<v Speaker 5>So as a result, we're building more factories here in

0:16:39.640 --> 0:16:45.360
<v Speaker 5>the United States, chip factories, packaging, computer factories, AI factories

0:16:45.400 --> 0:16:48.000
<v Speaker 5>of course, so we're building factories of all kinds here.

0:16:48.520 --> 0:16:51.840
<v Speaker 5>They're also ramping up capacity, and the reason for that

0:16:51.920 --> 0:16:53.560
<v Speaker 5>is because the demand is just so.

0:16:53.600 --> 0:16:54.600
<v Speaker 1>Great across the board.

0:16:55.560 --> 0:16:58.080
<v Speaker 5>I think the answer is that we want to have

0:16:58.560 --> 0:17:02.320
<v Speaker 5>it is possible to have supply chain diversity and resilience,

0:17:02.600 --> 0:17:06.320
<v Speaker 5>and we everybody should be seeking to improve that. And

0:17:06.359 --> 0:17:09.119
<v Speaker 5>it's also very true that Taiwan will continue to be

0:17:09.800 --> 0:17:12.000
<v Speaker 5>one of the epicenters of the world's technology hub.

0:17:12.960 --> 0:17:14.720
<v Speaker 2>Michael, I grew up using a del com Peter.

0:17:14.880 --> 0:17:18.440
<v Speaker 3>You know that we discussed it in the past, desktop laptop.

0:17:18.760 --> 0:17:21.760
<v Speaker 3>You and I never talked about computers in that context.

0:17:22.080 --> 0:17:24.680
<v Speaker 2>We're always talking about supercomputers accelerating computing.

0:17:25.240 --> 0:17:27.399
<v Speaker 4>But no, and I think you should Upbrade. I mean,

0:17:27.440 --> 0:17:30.200
<v Speaker 4>now we have the new XPS fourteen or sixteen. That

0:17:30.440 --> 0:17:31.560
<v Speaker 4>would be my choice for you.

0:17:31.800 --> 0:17:32.760
<v Speaker 2>So what is the story?

0:17:32.800 --> 0:17:34.600
<v Speaker 1>These are the best notebooks we've ever had.

0:17:34.440 --> 0:17:37.560
<v Speaker 3>Talked about AIPC. But we're gonna get Gensen's take to finish.

0:17:37.640 --> 0:17:39.840
<v Speaker 3>But what is the role of the PC in this

0:17:39.920 --> 0:17:42.560
<v Speaker 3>agentic age? Like I'm using a computer at my desk

0:17:42.600 --> 0:17:43.200
<v Speaker 3>to do work?

0:17:43.880 --> 0:17:47.359
<v Speaker 4>Yeah, well, look, I mean it's still the device that

0:17:47.600 --> 0:17:52.600
<v Speaker 4>is the center of productivity for knowledge work and it

0:17:52.720 --> 0:17:55.560
<v Speaker 4>is right there in front of everyone, and you know,

0:17:55.680 --> 0:17:58.439
<v Speaker 4>we have a great business there. And those devices are

0:17:58.440 --> 0:18:01.200
<v Speaker 4>evolving too. You saw on stage we're you.

0:18:01.160 --> 0:18:03.919
<v Speaker 1>Know, embedding the.

0:18:03.160 --> 0:18:05.960
<v Speaker 4>Ability to run the small models and the local models

0:18:06.280 --> 0:18:11.879
<v Speaker 4>inside your PC. And you know what's happening is customers

0:18:11.920 --> 0:18:15.920
<v Speaker 4>are wanting more powerful PCs because they want to be

0:18:15.960 --> 0:18:19.080
<v Speaker 4>able to do all this this this great hybrid AI.

0:18:19.720 --> 0:18:22.920
<v Speaker 4>And so it's it's a it's it's a great business.

0:18:23.000 --> 0:18:26.000
<v Speaker 4>It's still very much alive, and it also gives us

0:18:26.040 --> 0:18:29.399
<v Speaker 4>incredible scale and strengthen our supply chain which helps us

0:18:29.440 --> 0:18:32.879
<v Speaker 4>secure all the you know, needed ingredients.

0:18:32.520 --> 0:18:33.040
<v Speaker 1>That we need.

0:18:33.480 --> 0:18:36.359
<v Speaker 3>So you spent thirty one years working on the services

0:18:36.400 --> 0:18:38.240
<v Speaker 3>design together accelerated computing.

0:18:38.320 --> 0:18:40.600
<v Speaker 2>That's the scale we're talking about. Let me just be reading, Well,

0:18:40.640 --> 0:18:43.600
<v Speaker 2>we started with the PC, but why don't.

0:18:43.440 --> 0:18:45.320
<v Speaker 1>We just team up? I was I was trying to

0:18:45.359 --> 0:18:46.320
<v Speaker 1>sell them a gaming GP.

0:18:46.640 --> 0:18:48.280
<v Speaker 2>So what's going to happen between the two of you?

0:18:48.400 --> 0:18:51.840
<v Speaker 3>A PC with a powerful GPU inside it?

0:18:52.040 --> 0:18:53.159
<v Speaker 2>Why doesn't have that?

0:18:53.440 --> 0:18:53.520
<v Speaker 1>Ye?

0:18:53.920 --> 0:18:55.439
<v Speaker 2>And what's the plan going forward for that?

0:18:56.080 --> 0:18:57.960
<v Speaker 1>Well, we can't tell you the plan right now.

0:18:58.240 --> 0:19:00.439
<v Speaker 5>Tell me very very soon, we're like to tell you

0:19:00.520 --> 0:19:03.640
<v Speaker 5>there's there's a Well, let's let's think think.

0:19:03.560 --> 0:19:07.840
<v Speaker 1>About think about the arc, think about them. I'm interested

0:19:07.880 --> 0:19:11.040
<v Speaker 1>in computing, no doubt. Uh, think about the arc of computing.

0:19:11.080 --> 0:19:13.640
<v Speaker 5>It When when Michael and I came into the industry,

0:19:14.000 --> 0:19:15.639
<v Speaker 5>it was at it was kind of at the tail

0:19:15.800 --> 0:19:18.520
<v Speaker 5>end of mainframes. Not that it was a tail end

0:19:18.520 --> 0:19:21.280
<v Speaker 5>of mainframes because mainframes go away. There was a tail

0:19:21.400 --> 0:19:25.160
<v Speaker 5>end of of its growth and it was the beginning

0:19:25.200 --> 0:19:26.520
<v Speaker 5>of personal computers.

0:19:27.040 --> 0:19:27.159
<v Speaker 1>Uh.

0:19:27.400 --> 0:19:29.960
<v Speaker 5>We're now seeing the beginning of of course AI in

0:19:30.040 --> 0:19:32.320
<v Speaker 5>the cloud, and that's going to continue to grow. But

0:19:32.320 --> 0:19:35.480
<v Speaker 5>we're also going to see personal AI instead of personal computers,

0:19:35.560 --> 0:19:38.080
<v Speaker 5>my personal AI. So the question is, and the reason

0:19:38.119 --> 0:19:40.000
<v Speaker 5>for that is just we were talking about earlier.

0:19:40.400 --> 0:19:42.919
<v Speaker 1>AI needs to be where the context is.

0:19:43.600 --> 0:19:45.359
<v Speaker 5>If all the information that I have is on my

0:19:45.480 --> 0:19:49.040
<v Speaker 5>laptop and I need I need help. I need AI

0:19:49.119 --> 0:19:51.159
<v Speaker 5>to help me do work on my laptop. Then I

0:19:51.200 --> 0:19:54.679
<v Speaker 5>need AI to run kind of locally. And if I

0:19:54.760 --> 0:19:57.520
<v Speaker 5>have UH, if I have a factory, then I need

0:19:57.560 --> 0:19:58.680
<v Speaker 5>agents to run.

0:19:58.520 --> 0:19:59.399
<v Speaker 1>In the factory.

0:19:59.560 --> 0:20:01.600
<v Speaker 5>If I have if I have a hospital, I need

0:20:01.680 --> 0:20:02.800
<v Speaker 5>agents to run the hospital.

0:20:02.880 --> 0:20:06.240
<v Speaker 4>That's the operating room, in the operating It can't can't

0:20:06.280 --> 0:20:07.680
<v Speaker 4>be running somewhere else.

0:20:07.560 --> 0:20:09.879
<v Speaker 5>Right, because that's where the context is, That's where the

0:20:09.920 --> 0:20:10.399
<v Speaker 5>action is.

0:20:10.520 --> 0:20:14.400
<v Speaker 4>Yeah, if you've got an autonomous vehicle, right, THEI has

0:20:14.440 --> 0:20:17.159
<v Speaker 4>to be running the car inside the vehicle. And so

0:20:17.800 --> 0:20:25.080
<v Speaker 4>this idea of distributed intelligence and unmetered intelligence right where

0:20:25.119 --> 0:20:27.240
<v Speaker 4>you can generate as many tokens as you want ed

0:20:27.359 --> 0:20:29.040
<v Speaker 4>on your new XPS sixteen.

0:20:30.080 --> 0:20:31.880
<v Speaker 1>You just have to get Blueberg to get you one.

0:20:32.000 --> 0:20:34.440
<v Speaker 1>You know, I'm sure we can.

0:20:35.040 --> 0:20:38.280
<v Speaker 3>Michael Dell taman Is CEO of Dell Technology is tens

0:20:38.280 --> 0:20:41.960
<v Speaker 3>and one CEO and video live in Las Vegas once again.

0:20:42.359 --> 0:20:44.200
<v Speaker 2>Dell Technology is Well twenty twenty six