Nvidia CEO Slams US Chip Rules, Trump’s AI Advisor Discusses Action Plan
42:15 Share

This is an iHeart Podcast. Discover how one of China's largest financial services companies serves 240 million customers with AI-powered solutions. Hear from Ping An's co-CEO and chief scientist. We're a goldmine of data integrated to provide tailored solutions to our customers. Public domain tax 3.2 trillion tokens, 7.5 billion images. Tune in to our technology-powered growth podcast.

Find it at group.pingan.com, Spotify, or Apple Podcasts.

Developers like you are building the future, but you need the right tools to move fast and go further, right? That's where Microsoft comes in. With tools like GitHub Copilot, VS Code, and Azure AI Foundry, you have everything you need to push the limits and bring your ideas to life faster. And with security, compliance, and responsible AI built in, you can focus on what matters most, building the next big thing. Learn more at developer.microsoft.com.

Bloomberg Audio Studios. Podcasts. Radio. News. From the heart of where innovation, money, and power collide. In Silicon Valley and beyond. This is Bloomberg Technology with Caroline Hyde and Ed Ludlow. ♪

Live from San Francisco, this is Bloomberg Technology. Coming up, Nvidia CEO Jensen Wang slams US restrictions on AI chip sales to China and calls them a failure. Plus, President Trump vows to make a Golden Dome missile defense shield fully operational by the end of his term. And President Trump invites Elon Musk to meet with South African leader Cyril Ramaphosa with Starlink, a point of discussion.

Let's get straight to today's top story. Nvidia CEO Jensen Huang slamming US restrictions on AI chip sales to China, calling them, quote, a failure and urging the White House to lower barriers, saying, quote, the US should maximize the speed of AI diffusion because if we don't, the competition will come. Bloomberg's Ian King, who leads our coverage of semiconductors, is here in San Francisco today.

What is Jensen Huang's grievance and how does it relate to NVIDIA's business? Yeah, no, that's the key point here is that China is the largest market for semiconductors. NVIDIA obviously has faced increasing regulations from Washington, which is restricting its ability to do business there, the latest of which we saw was the

H20 chip, which you can no longer ship there. And this was a chip, remember, that he scaled down so that it wouldn't trigger those restrictions. So life is getting harder and harder for Jensen and for Nvidia in China. Scaled down in the sense that he lowered the performance of that chip so that it met parameters outlined by the U.S. government. It's kind of an interesting change of tone because it was just literally seven days ago that Jensen Wang was with the president in the Middle East. And at

At that time when the Commerce Department rescinded the diffusion rule, Jensen was quite happy about it. But there's a distinction between NVIDIA's ability to sell semiconductors to the Gulf states and its hope to do business directly with China. That's right. I mean, he's trying to parlay that sort of momentum into a fresh look at what's going on in China.

The diffusion rule was basically going to restrict an area of the world where there's a lot of money going to be devoted to AI infrastructure. And there was a victory that the US companies are going to be allowed to be part of that. His argument is that should be the case everywhere because then that will give US technology the opportunity

the position that it needs in the position it should occupy which is that they're a center of this revolution and a reminder that invidia has its quarterly earnings report next week bloomberg zine king thank you very much now last week there were multiple deals between u.s technology companies including invidia and the gulf states as part of president trump's visit to the region

In the week that's followed, China's access to U.S. technology via third countries or channels like the Gulf has since become a concern for the technology industry. To discuss, we're joined by White House Senior Policy Advisor for AI, Sriram Krishnan, who was with the President in the Middle East and participated in the formation of some of those deals. Mr. Krishnan, welcome to Bloomberg Technology. Thank you for your time and for joining us. You heard there the reporting overnight about

NVIDIA CEO Jensen Wang's complaint about the state of American business and being able to do business with China. You're familiar with that story. What is your reaction to start to Mr. Wang's comments and his line of argument? First, Ed, thank you for having me on the show. I'm a big fan, excited to finally be here. It may be useful to set some context about how we got here.

In the first week of President Trump taking office, he issued an executive order, which did away with the old Biden executive order on AI. And President Trump's executive order had very clear marching orders. It said, we need

dominance for American AI. And since that first week, that's what me, the AIs are David Sachs, Michael Katsios, the head of OSDB and many, many others have been working towards. And what you saw last week in the Gulf as part of this historic trip was a milestone as part of the dominance of American AI. And if I could maybe quickly recap what we announced last week. So under the Biden regime,

it was pretty much impossible for our allies to get access to our best and brightest advanced semiconductor technology, whether it be from NVIDIA or AMD or a whole set of other companies. We basically divided the world into a set of GPU haves and a bunch of GPU have-nots and essentially really told our allies to go pound sand. And last week's trip, which was historic, changed all that.

This deal, these deals that we struck in the Middle East has three essential components which are very, very important to understand.

The first is they are America first in the sense that every one of these investments is going to have a matching investment in American data centers and American AI infrastructure right here in the United States. So if there's a UAE or any of the countries are building data center capacity over there, as a part of this framework, we are going to get equal

build out here in the United States. And I think that is a very, very critical piece of this in terms of how this deal is America first. Number two, these deals and these GPUs are pre-existing.

are predominantly going to be run by American hyperscalers and American cloud service providers and American companies. So for our businesses, a lot of whom are probably watching the show right now, this represents an amazing expansion opportunity, a new channel, if you will, for a region which has obviously resource rich. So that's number two. A lot of these GPUs, most of these GPUs are going to be run, hosted, controlled by American companies.

And the third part, and this is maybe the most important part because I think Ian and you touched on it, is on security. So we took security incredibly seriously, and this agreement have two key security protections, which is about how do we stop the physical diversion of these GPUs to countries of concern, and two, how do we make sure there is no illegal or illicit

remote access. So that's kind of the deal and that's why it's historic. And the reason why this is important, and again, this is, we are in a business show, this will make sense, is what is the history of Silicon Valley teaches? It teaches us that the companies which have the biggest ecosystem, the biggest network effect, they wind up winning.

And America AI Inc. is no different. So with this piece of AI diplomacy, what we are doing is essentially locking in the American AI stack in the Middle East. And I think that's going to set us up for years and decades to come. Mr. Krishnan, what...

Jensen Huang is essentially calling for is to maximize, speed up diffusion. And his argument is predicated on the idea that there's a lot of activity in China. He has consistently said like more than 50% of the research or AI development is done there. What is your current thinking and what will you advise the president on either ending limits to China's ability to buy US chips or adjusting the current rules?

Look, we obviously have huge respect for Jensen. I was with him and several other founders and CEOs in the Middle East last week. When it comes to, I think it's important to make a distinction between advanced semiconductors being used inside China and everywhere else. When it comes to inside China, I do think there is still...

bipartisan and broad concern about what can happen to these GPUs when they are physically inside because a lot of these are obviously dual use. They can be used for a broad range of applications. And we've tried to kind of thread the needle when it comes to export regulation.

But one of the things to point out is I think Jensen is right on a couple of these items. The first is our previous export controls have failed from the Biden regime. So if those export controls had worked, we would not be looking at deep sea coming out the first week of President Trump's

being an office, which basically kind of skirted around a lot of our export controls. So the Biden regime did not enforce his export controls correctly, and we are taking measures to stop that. But again, it is important to distinguish between GPUs inside China to the rest of the world. And when it comes to the rest of the world, we want American AI stack, starting from the GPUs to the models to everything on top,

We want our companies, a lot of them obviously sitting very, very close to where you are now in Silicon Valley, to go diffuse around the world. And I think on that, Jensen and I and us are in a violent agreement.

You raise an interesting point that many technologists or just technology company founders in the Bay Area particularly make to me, which is why even bother with the infrastructure build out in the Middle East? Why not retain the full stack and capacity here in America, both not just from a national security point of view, but just the market have all of the workload availability here on domestic soil?

I'm happy you asked that because I saw this come up on X a couple of times. And here's what I would say. The first is, going back to the first thing I said, as a part of these deals, every bit of investment that goes into the Middle East, into infrastructure, is going to be matched.

So if we have a lead, we're going to continue having that lead and we're going full throttle on build, baby, build as what maybe the president would say in making sure we have enough infrastructure and enough capacity. But on the other hand, you have to think about the fact that these and other countries in the world are our allies. They are resource rich. They have ambitions. They want to use American tech.

What do you think winds up happening if we basically say, hey, we can't do business with you. Go look elsewhere. They're going to go look elsewhere. We would much rather have them locked in to the American stack for years and decades. And what that would represent to the founders and CEOs watching this is an opportunity, an opportunity for partnerships, an opportunity for market expansion, rather than a market which is locked to an adversary.

An AI action plan is due in mid-July. There is a new AI policy document. Could you give any preview for our audience of what you, Mr. Sachs, the president, will ask Congress to legislate on in the field of AI and what policy is coming down the pipe? Well, I know you had...

White House Office of Science and Technology Policy Director Michael Kratios on the show who talked about this, and him, David Sachs, A.I. Zarr, and me and many others are working on this. Now, I don't want to sort of give away what is going to be announced in a month and a half, but it's a season of summer trailer. But the broad principles, please, Sriram, the broad principles. Yes, I would...

I would say a few things, right? And I think Michael talked about promote and protect. I think here's what we would say. First, it's critical to have innovation and domination when it comes to this technology. And what we saw in the Biden regime was an incredibly complicated executive order, an incredibly complicated diffusion rule, a

patchwork of state laws. We need less of that. We need more of entrepreneurs like the ones who are on your show or watching now. We need them to go build, right? And that's what we need. And I think that's probably going to be a big theme. Second, build, baby, build. We need infrastructure. We need to make sure we get...

bureaucracy and red tape out of the way in making sure this infrastructure is built out as quickly as we can. And the third thing I would say is AI diplomacy. We need our allies with resources who want to work with us. We want them to pick the American stack and not be frustrated and wait for somebody else to go build a competing solution. So I think those would be three kind of mini previews of what's going to be in the AI action plan.

Sriram, I think there's a lot of interest in your day-to-day work with the administration. There are a number of names from Silicon Valley, so to speak, you, David Sachs, but also Elon Musk. Has Elon been engaged with you and had a seat at the table specifically with the president's considerations around AI?

I asked because yesterday I was at Dell Technologies World talking to them about the rapid build-out of Colossus and the XAI data center, and I wonder how much influence he has on your thinking and on the president's thinking. Well, look, I mean, Elon is a long-term friend of mine,

mine, of David Sachs, of the president, and he's obviously one of the greatest entrepreneurs of our era. And if he's hanging around the White House, we would be stupid to not seek out his advice. So he's definitely had useful input. But at the end of the day, I think this comes from the president himself. The second day in office, his executive order said, we need American AI dominance. And there's a lot of people

from Michael Katzias, David Sachs, obviously the Secretary of Commerce, Howard Lutnick, who was instrumental in seeing these deals happen. Many, many, many others who are all working towards this goal. Sri Ramakrishnan, the White House Senior Policy Advisor for AI, thank you very much for joining us on Bloomberg Technology. Now coming up, President Trump fast-tracks his Golden Dome defense plan. We're going to discuss that more

And some breaking news: Bitcoin has hit a fresh record high, largely driven by optimism around coming US regulations. This is Bloomberg Technology.

Discover how one of China's largest financial services companies serves 240 million customers with AI-powered solutions. Hear from Ping An's co-CEO and chief scientist. We're a goldmine of data, integrated to provide tailored solutions to our customers. Public domain tax, 3.2 trillion tokens, 7.5 billion images. Tune in to our technology-powered growth podcast.

Find it at group.pingan.com, Spotify, or Apple Podcasts. The world is built on code. From the apps we use every day to the systems powering industries, developers like you are the architects of tomorrow. But let's be real. The road to innovation can get a little tricky. You need the right tools to move fast, but you also need a community to help you go further.

That's where Microsoft comes in. Microsoft has the tools to help you move at lightning speed, like GitHub Copilot, VS Code, and a ton of AI resources to keep you on the cutting edge. But here's the best part. You can build with confidence, knowing that Microsoft's security and compliance are already taken care of. No more worrying about vulnerabilities or threats while you focus on your craft.

And with Azure AI Foundry, you can build your way. The future is yours to build, no strings attached. From ready-to-code tools to full flexibility, it's all in one place. The future's in your hands. So learn more at developer.microsoft.com slash AI.

President Trump has vowed that a new Golden Dome missile defense would be fully operational in three years' time, saying it would be able to protect the U.S. from ballistic, hypersonic and advanced cruise missiles. I want to bring in Bloomberg's Michael Shepard in D.C. Michael, what are the details of the president's pledge?

Well, you know, actually, Ed, the details are a little bit short. He did outline what he expected would be the cost for this system, which would be space-based and designed to protect a huge area of the United States territory from Alaska to Florida to Hawaii from ballistic and hypersonic missile attacks. He gave that three-year timeline that you outlined. Many believe that to be optimistic. His price tag for it is $175 billion.

over the next several years but the congressional budget office has a different view on the cost it sees it going up as much as five hundred forty two billion dollars in taking as many as twenty years to deploy and there are a lot of uncertainties about this technology ed uh... including how it would get deployed and who would be

in the lead to do it. We see some of the biggest names that we know from American Defense, including Raytheon and Lockheed Martin and L3 Harris getting name-checked at the event yesterday at the White House, where parts of this were unveiled together with Defense Secretary Pete Hegseth. But we also know that there are other startups that are trying to get involved in this as well. And, of course, there is Elon Musk. SpaceX is one of the top Pentagon contractors when it comes to defense

satellites and launches of satellites. And they've also done a significant amount of work for the intelligence sector here in Washington as well. They would be a likely contender. But there's a lot that we don't know about this system. And the three-year timeline, according to many experts, is really optimistic, given the task that the president

is outlining and the expectations he's putting on it. Remember, this is a campaign pledge that he made much in the same way that he is promising this big, beautiful tax cut and all those tariffs. And so this missile defense program is another pillar of what he had promised the voters.

Bloomberg's Michael Shepard, thank you very much. Staying in Washington, President Trump is meeting with South African leader Cyril Ramaphosa, with Elon Musk also set to join. Bloomberg's reported that the South African government is set to offer Musk a workaround of black ownership laws, allowing his Starlink service to operate in the country. Musk has been a staunch critic of Ramaphosa's government and what he calls racist laws. Here's what Musk told Bloomberg yesterday at the Qatar Economic Forum.

It's improper for there to be racist laws in South Africa. All races should be treated equally, and there should be no preference given to one or the other. Whereas there are now 140 laws in South Africa that basically give strong preference to if you're a black South African, and not otherwise. And so now I'm in this absurd situation where I was born in South Africa but cannot get a license to operate in Starlink because I'm not black.

For more, we're joined by Bloomberg's Max Chafkin, a member of our Elon Inc. team as well. And at the crux of that soundbite was a question about whether Trump's relationship with Musk opens doors for Starlink in certain jurisdictions because those countries want to get close to President Trump and this administration.

Yeah, absolutely. And look, there are reasons why South Africa might be willing to do this outside of the influence of Donald Trump and perhaps an effort to make the White House happy. Starlink, of course, offers a compelling service. We've seen a lot of countries and people, especially if they can afford it, embrace this. On the other hand, this is clearly a case where Elon Musk

is benefiting from his connection with the White House. We're seeing this not just in South Africa, but lots of countries over the last couple of months changing their rules, essentially allowing this company to operate, you know, telecom, satellite telecom in particular is a heavily regulated field, not just in South Africa, but around the world. So this is part of a larger pattern, I would say.

Musk born in Pretoria, South Africa. This is just one case study, right? Elon Musk has met with several heads of state, either directly because he's with Trump in person or on the president's behalf since Trump took office. Yeah, and what's interesting about this meeting today, I mean, you know, besides the dynamic with Musk as a critic, is just that, you know, Musk has said in various ways, "I'm backing away from the White House." You know, he said he's spending less time there.

He's maybe going to donate less money, at least according to his comments yesterday. On the other hand, here he is at the White House with Donald Trump. Also yesterday, he said he was going to spend two days this week. I think he mentioned a cabinet meeting as well. So, you know, on one hand, he may be making noises about backing away, but it sure doesn't seem like he's doing a ton of that right now. Bloomberg's Max Chafkin. Thank you very much.

Google is rolling out its experimental AI agent, Project Mariner, to subscribers in the United States. The research prototype can browse the internet and use websites. Bloomberg's Jackie Davalos heard more about it from Jacqueline Kondelman, the director of product management at Google Lab.

Project Mariner is a research prototype that's exploring the future of human-agent interaction starting with browsers. And what we launched today was a web app that allows anybody to give Project Mariner a task and it will be able to go and navigate across the internet to complete it.

and the specific version that we launched today can also handle up to 10 tasks that it can do at the same time, allowing people to really try to multitask. Agents are really kind of the next frontier of AI. We talk a lot about how artificial intelligence can boost productivity, and agents seem to be really at the forefront of that. But I'm curious

about how this moves from the innovation lab to prime time, especially because if you're asking an AI to do something, you really want to make sure it gets it right. Absolutely. So I think that that's part of the power of launching Project Mariner as this research prototype, as an experiment right now. We are putting users at the forefront of being able to try this out firsthand. We are

pairing nicely the ability for Project Mariner to multitask, but also a user at any time can watch everything that it's doing. It'll get a live view of the task as it's happening, and a user can take over the task at any time. They can also pause the agent from doing the task and resume it at a later point in time, or

Or they could choose to trust Project Mariner if they're giving it something that's a simpler task and go back to whatever it is that they're doing. So we've really tried to blend user control with pushing forward what these agents are capable of and launching it as that early experiment kind of lets us take it to that next phase of bringing it to people. That was Jacqueline Konzelman, Director, Product Management at Google Labs, speaking to Bloomberg's Jackie Devalos. Look at shares of Google, by the way.

This is easily the best performer on the NASDAQ 100, outperforming the other MAG7 names. And it's all reaction to what was happening at Google I.O., on track for its biggest jump since early April, the stock trading at its highest level since early March. Analysts are very positive about what they heard, and they note that the company's updates have basically eased concerns about its ability to win in this crowded AI space. We will continue to talk about that.

Google throughout the hour and what we learned out of Google I.O. Now coming up, we speak with Crusoe CEO Chase Lockmiller as the company sees the second phase beginning for its Texas AI data centers. A really interesting and timely conversation coming up here from San Francisco. This is Bloomberg Technology.

Welcome back to Bloomberg Technology. Ed Ludlow here in San Francisco. Let's talk a little bit more about what's going on in financial markets and the tech sector. Martin Norton, Chief Investment Strategist at Empire, joins us now. I'm trying to find a common thread of what's going on because when I woke up this morning, a lot of the headlines on the terminal were about economics and concerns about growth. There's been a lot of Fed speak.

And then tech brings us back, baby. Google I/O, a developers conference, headlines, Alphabet's up 5%. What is it that you're paying attention to right now? Well, I think you're making an interesting point because really not just this morning, but the past few weeks, the past few months, maybe you could say the entire 2025 year so far has been focused on news out of Washington and kind of that macroeconomic and fiscal policy dynamic.

What we're seeing at least for the moment is a return back to the fundamentals, the individual story and the AI narrative, which has been lost for a while as we've dealt with deep seek and as we've turned macroeconomic. So I think what we're seeing here is a move back to the individual companies a bit.

Right. And then it reminds me of all of the focus we had earlier in the year on Mag7. Perhaps actually more to the downside. But, you know, if you have a stock like Alphabet and it is up that much, just purely from a points perspective, it's giving us a little lift this morning. That's the moment in time. Yeah. Sorry, I don't mean to interrupt. All I was going to go on to say is like.

It was a moment in time where investors wanted to say, what is it you're going to do in AI? And they gave enough evidence, I think, over in Mountain View yesterday. Well, I think that's absolutely the case. And I think what's also interesting is when we look back to earlier in the year and really up

up until the present moment, a lot of the focus has been on the spend for AI and questions on whether that's reasonable. And at least with the most recent news, and I think this has been kind of filtering out over the course of the year, we're seeing a bit more of that use case. And that gives kind of a shinier tinge to what we can think about when it comes to AI.

So it's shifting the narrative a bit. Now, whether this is long lasting, I think there's a lot more to see on that front. But it certainly is shifting the narrative to the use case rather than just the spend on AI on the CapEx front. Liberation Day.

is kind of, it feels like it's in the rear view mirror a bit, but if there's one thing that's consistent about this administration and trade, it's that anything can happen. Is the market still braced for that? Is it checking over its shoulder for some uncertainty in the tech sector? Well, this is what I'm wrestling with. I think Liberation Day, you know, the seismic change in trade where we don't really even fully know where it lands quite yet,

It feels like, to your point, distant history, a blip on the radar, and markets have rallied really powerfully. That's at the broad market level, and that's also within technology. If we're looking at technology as a sector on a valuation basis relative to its history, it's back in those upper echelon of deciles where the valuations are really stretched.

gives me some concern that there is a bit of complacency or as we're thinking about the future, the range of outcomes, people are really focused on the positive and less focused on what could go wrong. That is the concern. But Ed, I think a really important point here is that the Mag 7 actually have looked cheaper. They've been harder hit over the course of 2025 and that's brought their valuations down. So maybe there's a bit more margin of safety in those mega cap tech names.

Right, that takes us to a broader conversation about valuations, right? That on any given day you can forget that the technology sector is more than the Mag7. Is there anywhere that you see opportunity or that you're super cautious on outside of that bucket of mega cap names?

I think I'd more characterize my view as cautious outside of the Mag7. If you're breaking down the technology sector, going industry by industry, most of the industries look a bit expensive. And it's really that, and it's such a strange world to be talking about the Mag7 as the valuation opportunity within tech, within the market. But that's kind of where our analysis is pointing us today.

Marta, I know that this isn't an entirely fair question, but I'd ask you to have a crystal ball. What happens in the second half of this year for the technology sector? Well, you know, I think a big part of what we're going to see over the second half of the year is growth.

maybe not resolution, but additional clarity on what's happening on the tariff front for technology, what happens on the regulation front, what happens in terms of, as you were recently chatting about, the administration's focus on building out kind of the US as an AI superpower. I think all of those different dynamics are going to have

a role to play in terms of how technology performs over the course of 2025. And what we have to ask ourselves as investors is what of those outcomes are embedded in the price and where is there a margin of safety? And so that's the kind of chemistry that we have to do as we look at the different movers coming over the course of 2025. Marta, real quick, a big moment in the show earlier, Bitcoin at almost 109,000 US dollars per token. You interested there?

It's tough with this. What I really struggle with is where is the fair value here? And I think that's really hard to determine. So I think, yes, we're getting a lot of news on the regulation front that's going to charge Bitcoin, that's going to charge crypto. But I think what we have to ask ourselves is can we predict what's coming next? And I think that's a bit harder question.

Martin Orton and Power, Chief Investment Strategist, great to have you back here on Bloomberg Technology. Thank you very much. Now, Crusoe, Blue Owl Capital and Primary Digital Infrastructure announced funding for the second phase of its AI data center joint venture, raising now $15 billion in total for the project. For more, Chase Lockmiller, Crusoe CEO, joins us here in San Francisco.

Actually, you featured in the show 24 hours ago because Emily Chang went to the Texas Stargate site and brought us that kind of deep look at its construction and your involvement there. Very simply, explain your role and this joint initiative that has raised this funding and where the funding goes.

Sure. The announcement we had this morning was $15 billion in funding for developing a very large purpose-built AI data center. We're calling them AI factories, these large-scale data centers with chips that can manufacture intelligence. So our role in all of this is we are basically...

We developed the data center itself and we're in charge of building it and we will own it in a joint venture partnership with Bulal. Data centers, to some,

are large buildings with concrete foundations and metal around the outside and no one really knows what's inside them. The other way of looking at it is that these are like massive infrastructure projects where it's not just the compute inside, all of the associated infrastructure around. Which part are you more responsible for? So Crusoe as a business will handle the entire stack. For this specific project, we're focused on the design, build, construction,

of this large-scale data center.

I think it's worth appreciating that data centers as a product have changed very dramatically in line with the changes in computer architectures over the last 20 years. So if you look 20 years ago, a standard data center rack might have been 2 kilowatts, maybe 4 kilowatts. Today, for the GB200 and VL72 configuration, we're budgeting 130 kilowatts per rack, all with direct-to-chip liquid-cooled components.

water cooling. And if you look forward to the next generation, what Jensen announced recently at GTC with the Vero Rubin Ultra, those are 600 kilowatts per rack. So you're looking at more than 100x

increase in overall power density. That fundamentally shifts what you're building in terms of when you're thinking about building and constructing a data center. The other piece I'll highlight is that these data centers are no longer a bunch of individual computers all plugged in, operating independent workflows and tasks.

You really should be thinking about the data center is the computer, right? All of these chips are interconnected on the same high-performance RDMA fabric so that they can really be thinking together and sharing information. The thing is, in aggregate, we're talking about 10,000 megawatt facilities, plus or minus, up to the future generation, maybe 20,000 megawatt.

Something that came out of earnings of the hyperscalers is that they raise capital expenditures in some cases like meta. But there's real evidence that that wasn't just like a bigger commitment. It's because the cost of building a data center is higher. Do you see that in your activities?

The cost of building it is higher because... Materials, I mean labor. Materials and labor are kind of like... Labor is honestly the biggest expense here. And what we're seeing because there's such a big boom, there's a shortage of labor, right? There's a shortage of construction workers. There's a shortage of electricians that can really make these facilities happen at scale. Crystal ball for you as well. You know, you've kind of put Crusoe on the map here, but this is one project, right?

Are we seeing year on year on year all the way through the 2030 just capital expenditure increase, more data centers, there's no sort of plateauing of the activity? You know, I can speak for what we're seeing, which is no slowdown. We are seeing incredible demand from across a diverse set of customers that want to build scaled infrastructure to support their own AI computing needs.

And we see that scaling by orders of magnitude. So from our perspective, this is the largest infrastructure investment in human history and it's not slowing down. - How big a challenge is securing energy and also the utility infrastructure to support your sites? - So energy is the core of this in terms of, we've never seen data center infrastructure built at this scale before. If you look at Northern Virginia,

As of the end of last year, a report from JLL shared that it was 4.5 gigawatts of total capacity in Northern Virginia. What we're building in Abilene, Texas is 1.2 gigawatts. So you're talking about one facility that's a quarter of the size of all of the data center capacity that's been built in the center of data center mecca in Northern Virginia.

And now we're looking at facilities that are five times bigger. So energy and the amount of energy required to support AI infrastructure at scale is truly mind-boggling. So it is a critical component to solving. And I know we're focused on doing it here in America and expanding also internationally.

Crusoe CEO Chase Lockmiller, great to have you on the show. Thank you very much. Real quick, CoreWeave shares soaring today, by the way, on pace for a four straight day of gains after Citi raised its price target on the data center builder to $94 from $43, keeping a neutral rating. Citi says the company's first quarter was, quote, out of the gate, though the rest of results were a mix.

Analysts are saying the earnings report reinforces Corweave's high growth status, especially with that recent $4 billion OpenAI expansion deal. But more progress on profitability is to be seen, something we discussed with the company this week. All right, coming up, these smart glasses are X-real and Google's answer to Meta's Ray-Ban smart glasses. We're going to hear from the CEO of the startup next. This is Bloomberg Technology. ♪

Discover how one of China's largest financial services companies serves 240 million customers with AI-powered solutions. Hear from Ping An's co-CEO and chief scientist. We're on a goldmine of data integrated to provide tailored solutions to our customers. Public domain tax, 3.2 trillion tokens, 7.5 billion images. Tune in to our technology-powered growth podcast.

Find it at group.pingan.com, Spotify, or Apple Podcasts.

The world is built on code. From the apps we use every day to the systems powering industries, developers like you are the architects of tomorrow. But let's be real, the road to innovation can get a little tricky. You need the right tools to move fast, but you also need a community to help you go further. That's where Microsoft comes in. Microsoft has the tools to help you move at lightning speed, like GitHub Copilot, VS Code, and a ton of AI resources to keep you on the cutting edge.

But here's the best part. You can build with confidence, knowing that Microsoft security and compliance are already taken care of. No more worrying about vulnerabilities or threats while you focus on your craft.

And with Azure AI Foundry, you can build your way. The future is yours to build, no strings attached. From ready-to-code tools to full flexibility, it's all in one place. The future's in your hands. So learn more at developer.microsoft.com.

Google's host of AI announcements yesterday included plans for augmented reality glasses that could rival Meta's upcoming smart glasses releases. To do this, the search giant has partnered with Xreal, a Chinese startup which makes popular AI sunglasses. And this new partnership, called Project Aura, would be the first extended reality glasses to run Google's Android XR operating system.

Let's bring in Zhizhu, the CEO and founder of XREAL. Welcome back to the program. And thank you for having me. I don't know that I saw this one coming. This is a very big deal for XREAL with a big partner. What was the logic behind doing it? Okay, so first of all, I don't think it's...

not just a big deal for us, I think it's a big deal for the future of AR on Android. Finally, speaking of the spatial computing, I feel like all the pieces are ready. The hardware is ready, form factor is right, and the platform is ready, ecosystem is there, and most importantly, Gemini.

AI, we believe, is going to make the whole XR experience so much better to the next level. So that's why we're so excited to see this coming. When's it coming? Tell me about the product. How real is this in terms of when a consumer or a developer can get their hands on the hardware? I think a developer will get them as soon as the end of this year. Is that it?

No, this is the current generation, XR1 Pro, the one you tried last time. For our audience, there's a set of glasses on the desk in front of us, but that's not it. Keep going. So we believe for the consumer, it's going to be early 2026.

Has the relationship with Google allowed you to accelerate your go-to-market on a new product? Absolutely, absolutely. And this is the first time, I believe, you're going to have this, for spatial computing, you're going to have this end-to-end experience. And you get a chance to try all the first-party Google applications on XR as well.

I am a Ray-Ban Metas user regularly. That is not a VRAR set of glasses. They're smart glasses. But as you know, I spent a bit of time with Project Orion, right? Augmented reality glasses. Why...

or why not? Is your offering going to compete with that? Is it similar? Why is it dissimilar? - Okay, I won't say it's a competition, but I really enjoy working with Google 'cause I believe the last missing pieces will be ecosystem. And Google always offer a really open, developer-friendly kind of ecosystem. And the theory line I like about Android XR is if you already develop on Android,

you develop on Android. - Well, how simple should we make this for the Bloomberg Technology audience, right? Android from a smartphone perspective or, you know, in cars, the OS is everywhere around the world. So you're basically arguing that I'm gonna make the Android smart glass simple. - That's right. I think, you know, if you look at the success of Android, Google just wanna repeat that for XR.

And this time they recognize they need to work with a lot of other OEM partners. And starting with Samsung, Project Wuhan will be the first XR devices. And we have X-Reel, we have even Gentle Monster, Wabi Parker announcing yesterday as well. I see we're putting alliance, working together to deliver a variety of different kind of devices for Android XR.

I've got a question for you from our audience when I said on social media you're coming on the show. This is from @vr_tonio on X. Can you ask them if the new X1 Pro needs the beam puck for a spatial experience? It is vague on your website.

Well, okay, so for X-Real, we always have two different product visions. One is what we call spatial display. It's basically an easy-to-connect accessory that can plug into virtually any devices, right? And then there's another one we call spatial computing, which is end-to-end experience, just similar to Orion or maybe even Apple Vision Pro.

Real quick, financials. How much money are you making right now? Do you need to raise money? How's that looking? Okay, so we do have a plan to go public in 26. Okay, where? Definitely in the US. All right, that was Shiju, CEO of XREAL, with up-to-the-minute updates, not just on the latest technology, but what's happening for the company too.

I want to go back to our top story. Nvidia shares are actually now markedly higher, having opened under pressure. CEO Jensen Huang overnight slammed the US over chip sale restrictions to China. For more, I want to bring in Mandeep Singh of Bloomberg Intelligence. Mandeep, help our Bloomberg Technology audience understand

the state of the market in China. Basically, NVIDIA, the world's leader in AI accelerators, but others like AMD, right? They can't access that market, but it's in that market where a lot of the activity of AI development is taking place and there are data centers being built as well.

Yeah, look, I think we all know about DeepSeq and what it did at the start of the year. And all these models, when you think about the training of DeepSeq or Alibaba's models, it's happening within China. And a lot of the NVIDIA GPUs are still used for training of LLMs. So the fact that NVIDIA won't be able to sell to China, which is about 20% of the overall

GPU spend of around $250 billion. It tells you that they don't have that sovereign AI opportunity which everyone is very excited about with NVIDIA. NVIDIA did sign up a few deals in the Middle East, but when I look at the large language model landscape, I mean, the four or five of the best models are here in the U.S., and then

Then you have the remainder of the other most prominent models in China. All the other countries are still investing, building the infrastructure, but they don't have the LLMs like these two countries have.

Mendy, when you publish research on the Bloomberg terminal, I print it and I put it on my desk. And I had this piece on my desk from March 17th, DeepSeek to recast $525 billion CapEx for hyperscalers by 2032. In the context of the NVIDIA story, just explain that thesis at the time and what's happened as a result of the DeepSeek volatility.

Yeah, I mean what DeepSeek did was really forced everyone who was on the training side to look at how much they wanted to spend upfront on pre-training and really pivot more towards reasoning and inferencing. And that's what we have seen with OpenAI, you know, 01 and 03 models and now with Google 2.5 Pro, a lot of emphasis on reasoning. And our thesis at that time was there is this shift from training to inferencing that's happening now.

And what will happen as a result of that is that inferencing opportunity will be a lot bigger. That part will grow a lot faster. And the net result is that overall AI market will continue to grow, but there will be that shift from training to inferencing. And I think we are seeing that pan out.

And in China, a part of it may be domestic names using domestic champion Huawei for some of their training chips. Mandeep Singh of Bloomberg Intelligence. Thank you so much. That does it for this edition of Bloomberg Technology. Don't forget, check out our podcast. So many of you listen to the show in podcast format. You can find it on the terminal and online on Apple, Spotify and iHeart. From San Francisco, this is Bloomberg Technology.

Developers like you are building the future, but you need the right tools to move fast and go further, right? That's where Microsoft comes in. With tools like GitHub Copilot, VS Code, and Azure AI Foundry, you have everything you need to push the limits and bring your ideas to life faster. And with security, compliance, and responsible AI built in, you can focus on what matters most, building the next big thing. Learn more at developer.microsoft.com.

Let's talk about.

This is an iHeart Podcast.