00:00Helios requires innovation at every single level, hardware, software, and systems.
00:07It starts with our engineering teams who designed our next-generation Instinct MI455 accelerators
00:13to deliver the largest generational performance increase we've ever achieved.
00:19AMD CEO Lisa Su currently speaking on stage.
00:23Ray Wang, founder of Constellation Research, joins us from Ground Zero.
00:27Ray, good to have you with us.
00:28We heard from Jensen Huang, Vera Rubin.
00:32It does seem like NVIDIA is doubling down on AI beyond data centers.
00:37How significant is this?
00:40This is the next trillion-dollar market opportunity, which is physical AI.
00:44Jensen's been talking about it for quite some time.
00:47That means autonomous vehicles.
00:48That means robotics.
00:49That means other devices that are controlled and actually move using AI and chips that reason on their own.
00:55This is the game-changer that we are looking for, which is driving the trillion-dollar and additional market cap that people are going to put in terms of valuations.
01:03And then, of course, the sovereign AI that's going to be powered by Vera Rubin.
01:07I think the game-changer here was the fact that we didn't think it was in production yet, and they accelerated timelines.
01:13This is an announcement they normally give at their GTC event in March, not at CES.
01:19So, this was a surprise to everybody who was there.
01:23So, you say it is a game-changer, and Vera Rubin has come earlier than anticipated.
01:29Does it change, perhaps, the valuations on how investors look at NVIDIA right now?
01:34I think it changes because what they're always worried about is, is there going to be another announcement?
01:40Is there going to be something new on the pipeline?
01:43And what they've learned over time is Jensen has a full pipe, at least for the next three to five years.
01:48Why this chip is important, it handles a trillion parameters.
01:51It's five times the performance with only 1.6x more transistors, which means it's even more efficient than what was going on with Blackwell.
01:59And, of course, they're now in market and in production, which means they're driving demand for new orders as we speak.
02:08So, how is Jensen Huang positioning NVIDIA right now?
02:14It's no longer just a chip company.
02:16It's now gone into autonomous vehicles.
02:18How do you view this company now?
02:22I think it's important to think about every layer, from chips all the way down to software, all the way up to devices.
02:28What NVIDIA has is an ecosystem in play.
02:32It's got all the chips that are there.
02:34It's got the software that powers the chips and the ecosystem, and it's coming across.
02:38And this is really what they mean when they're talking about full stack for the physical AI ecosystem.
02:43That means the partnership with Mercedes.
02:45That means the partnership with what's going on with Siemens.
02:48It's basically about putting AI into action.
02:51A lot of this has been theoretical.
02:52And what they've been able to show is that they've got everything across the full stack that companies and customers can take to market and taking advantage of the exponential efficiency and infinite possibility with AI.
03:06Ray, we know that Lisa, let me just say, Lisa Su of AMD is speaking on stage right now where you are.
03:16I'm just wondering, how do the rivals compare?
03:19How does AMD compare?
03:20How does Intel compare?
03:23Well, they're in different leagues at the moment, right?
03:26NVIDIA is dominating the GPU market.
03:28And AMD is about to launch their new GPUs to the world so that there's a competitor in the market or an alternative.
03:34And Intel, of course, is getting a lot of federal funding and a lot of funding to actually get manufacturing in the U.S.
03:41So it's kind of seen as a more state-sponsored kind of approach.
03:45All three are playing in different parts of the market and have different roles.
03:48And, of course, NVIDIA's latest announcement with Grok for our aqua hire and licensing deal gives them the ability to get into the market Google was in, which is about TPUs.
03:58So as we move from training to inference, we're moving from GPUs to TPUs.
04:03And, of course, NVIDIA is working hand-in-hand to be able to make that happen and be able to take the efficiencies of both GPUs and TPUs with that acquisition of Grok.
04:14As for AMD, Ray, what would you like to hear from Lisa Su that would perhaps, you know, boost the outlook for the company?
04:23I think on the AMD side, you're going to see a lot of opportunities to talk about more efficient chips, more availability of those chips and more applications.
04:34You're going to see a competitor on the other end in terms of ability to get production in play.
04:39And so I think it's an important piece for AMD to show that they are the alternative to NVIDIA on the GPU side and, of course, some of the partnerships that they're doing at the data center side.
04:48And so you're seeing because of the demand in the marketplace, it isn't a winner-takes-all market yet.
04:53It's still a lot of demand and a lot of chips that are needed to power the revolution and power the commitments that have been made for data centers.
05:01We're talking almost $61 billion in data center commitments in terms of the Western world this year and, of course, more going forward.
Comments