Meta to Deploy “Millions” of Nvidia Processors

0:00Hello, I'm Stephen Carroll. I'm in Brussels where many of Europe's biggest decisions get made. And I'm Caroline Hepker in London with the hosts of the Bloomberg Daybreak Europe podcast. We're up early every weekday keeping an eye on what's happening across Europe and around the world. We do it early so the news is fresh, not recycled and so you know what actually matters as the day gets going. From Brussels, I'm following the politics, policy and the people shaping the European Union right now. And from London, I'm looking at what all that means for markets, money and the wider economy. We've got reporters across Europe and around the globe feeding in as stories break.

0:37So whether it's geopolitics, energy, tech or markets, you're hearing it while it happens. It's smart, calm and to the point. And it fits into your morning. You can find new episodes of the Bloomberg Daybreak Europe podcast by 7am in Dublin or 8am in Brussels, Berlin and Paris. On Apple, Spotify, YouTube or wherever you get your podcasts.

1:02Bloomberg Audio Studios. Podcasts, radio, news.

1:11Bloomberg Tech is live from coast to coast with Caroline Hyde in New York and Ed Lovellow in San Francisco. This is Bloomberg Tech. Coming up, Meta agrees to deploy millions of NVIDIA processors over the next few years, reinforcing its ties with the AI chip giant. Plus, Meta's CEO Mark Zuckerberg is set to take the stand later today in a landmark social media addiction trial in Los Angeles. And we discussed some news in software and energy tech with the CEOs of Autodesk and Heron later this hour. Welcome to Bloomberg Tech. The story right now in public markets when it comes to tech is we are rebounding a little bit.

1:55The Nasdaq 100 up more than a percentage point. Largely, that's on strong or positive economic data. But I put the Philadelphia Semiconductor Index in there or SOX because actually movements in chips and the Mag7 is the news driver, the catalyst in some of what's happening at the index level. That manifests in a deal between Meta and NVIDIA for millions of NVIDIA's processors over a number of years. But this is clearly a driver of markets, It's at least in NVIDIA's case up almost 3%. Meta had been lower, is now modestly higher. Let's get the details of Bloomberg's Ian King, who leads our coverage of semiconductors.

2:33So the reality is, Meta's already the number two customer for NVIDIA, but this is an agreement over time. Millions of chips, and we're not just talking Blackwell GPU, we're talking later generation Rubin, and then the CPUs in each case. What are the kind of facts that we need to know? Yeah, and networking as well. I mean, the frustrating part about it is we don't have a definitive time period. But if we sort of back it out, a million chips, their average selling price for NVIDIA of one GPU for the data center is about$15 ,000,$16 ,000, probably as much as twice as that for a leading-edge one. So you're already talking as much as Meta spends on NVIDIA gear in a year in just the GPUs, add on top of that the networking, add on top of that the CPUs.

3:21And obviously it feels like an extension. It feels like a bigger commitment to NVIDIA, but they wouldn't say for sure. You had a discussion with one of NVIDIA's leaders in the data center and infrastructure group. And part of the discussion was this is the first big example of NVIDIA selling the Grace CPU to market as a kind of standalone product. Simple question is like, what would you use a Grace CPU for other than, you know, what's happening in AI? And that is why the number is probably tens of billions over that time period, right? Yeah, I mean, it's definitely a step forward for them. It's definitely a market expansion for them.

4:00You remember Jensen told us this a while ago. He dropped a strong hint that this was coming, and here it is. The key thing to look at is, this is Intel's territory. This is AMD's territory, the general purpose processor for a data center that does a lot of the general computing tasks. what NVIDIA is saying, hey, we've got this CPU that was buried in the middle of all of this gear, but you can just use it on its own as well. That's a market extension for them if we see widespread adoption. Obviously, Meta is a good adoption case. One of the other movers in the market this morning is AMD, which was to the downside.

4:35I guess not surprising, the biggest story is NVIDIA kind of keeps rolling on and dominating the spend of the hyperscalers plus Meta, which is not a hyperscaler, but that has the data center footprint of one. I was interested to go into the terminal and see like Meta about 9 % of revenues for NVIDIA. But Microsoft's number one, you know, double that. But the big picture is there, right? NVIDIA, it seems dominant. Yeah, I mean, you know, the background here is Microsoft, Meta, Amazon. They're all making their own CPUs. They're all trying to make their own accelerators. Any commitment to either NVIDIA or indeed AMD for exactly those parts is obviously something that investors want to see and is a strong sign for the chip makers themselves.

5:19Bloomberg's Ian King on a story that, again, news driven catalyst for markets this morning. Investors are looking also to scoop up beaten down names after a stretch of choppy AI driven volatility. Morgan Stanley says the recent sell off, especially in software and services, was largely indiscriminate with some companies hit harder despite strong fundamentals. Here's break down the AI impact across all sectors is Morgan Stanley, Global Research Director, Katie Huberti. I mean, it's such a detailed report that looks across stocks with a very wide lens. But in the here and now, I want to go to that word indiscriminate and just that paragraph of the report where you summarize the last few weeks.

6:03What do you think was happening? Right. Right. So I think it's important on the back of the announcement that you just discussed to understand that we're in the very early innings of what is likely to be a 10 trillion dollar CapEx investment cycle. That investment cycle is underwritten by a productivity story. As the technology diffuses into the market, there is real productivity acceleration, which is important because the last two decades we've been running at labor productivity that is half of normal levels. Because we knew this adoption story was going to be critical and it was going to touch all sectors of the market, that we have now five times over mapped the 3 ,600 stocks that we cover to understand rate of change, exposure, materiality, pricing power.

6:59And it's on the back of that data that, yes, we've been able to identify where there are companies that have been sold indiscriminately. And there's a real opportunity. We see that in software and services. in research, we're a heavy adopter of AI technology. And the conversations we're having right now with large language models is how do we connect those models to the applications and data that we use every day to unlock the real potential? And so you can see that the list that we published last week, the thread across the software and services companies are where there is a data mode, whether that's credit or it's market data or systems of record in financials or customer data.

7:50This is the fifth time you've done this exercise of mapping global AI stocks, right? And you list margin expansion as the leading piece of evidence of ROI. I'm thinking back to earlier in earnings season when we had names like Salesforce ServiceNow and that part of the research report I flagged at the beginning, fundamentals you know they didn't really get the benefit of the doubt from this the market they showed actually quite a lot of real world deployment of that technology um but something quite you know isn't quite believed by the by the street what is that so what's really important in this data set is is tracking rate of change so if you go back two years ago, the big rate of change in exposure materiality was energy and utilities.

8:39That sector outperformed the most in 2024. Last year, it was financials. At the beginning of 2025, we highlighted as the biggest rate of change. Financials, by many accounts, even outperformed parts of the technology sector last year. Right now, the rate of change is actually greatest in non-tech categories, particularly in consumer and apparel, in durable goods, in autos. And so, yes, we see areas of software that have been sold indiscriminately where there is a data mode and there's an opportunity, but it's broader than just technology. What we see in the data is that the AI adopters are experiencing margin expansion that's double the market indices, double the MSCI World Index, double the S &P 500.

9:32And so we believe that there's a regime shift happening in the market away from a purely tech and specifically AI enabler leadership to now broadening of AI adoption and leadership, not just in technology, but across all sectors. Katie, tied to this, one of the big moments of last year was Morgan Stanley launching its private companies coverage, going much deeper on private companies who are so active in this world, right? Have you shifted existing analysts from public to private companies? Are you onboarding specific private analysts on specific domains? We just have 30 seconds. Right. So we believe that the domain experts that cover the public companies are typically and best positioned to also cover private sectors.

10:24But there are areas of thematics, think humanoids and AI, where we have shifted analysts from stock coverage to covering those thematics, where there will be a lot of private company innovation. But most of the time, that coverage comes from our existing public company domain experts. Katie Hubert here, Morgan Stanley, it's great to have you on Bloomberg Tech. Thank you very much. Now, coming up on the show, we're going to speak with Andrew and Agnos, Autodesk President and CEO. That conversation's next. This is Bloomberg Tech.

11:07This is Special Agent Regal, Special Agent Bradley Hall. The time is approximately 11.15am. I'm about to start a consensual telephone call with Dr. Daiwa Zhang.

11:25China's Ministry of State Security is one of the most mysterious and powerful spy agencies in the world. But in 2017, the FBI got inside.

11:48I've never seen that much evidence in my entire career, and I don't think we'll ever see that much evidence again. I now have several terabytes of an MSS officer, no doubt, no question, of his life. And that's a unicorn. This is a story of the inner workings of the MSS and how one man's ambition and mistakes opened its vault of secrets. Listen to The Sixth Bureau from Bloomberg Podcasts starting on February 13th on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.

12:32autodesk is leaning into the ai-powered physical world the company announced a 200 million dollar strategic investment in ai research firm world labs its largest startup investment to date here to discuss autodesk president and ceo andrew and eggnauss very interesting good morning to you I'm very excited about today. We've been looking into this idea a lot, you know, through the beginning of 2026, particularly in the context of industrial manufacturing, the future role of humanoid robots, how they come to learn the environment that they're in. People sometimes say digital twins. You have chosen what you believe to be the best player, right, for the sort of AI and software underpinning of that.

13:18To start, why World Labs? Yeah, well, there's a couple of really important reasons. First, we kind of share a common ethos. We believe that AI can enhance and expand human ingenuity, not replace it. And that's something that both World Labs and Autodesk share in common. But more importantly, Fei-Fei Li, Autodesk, others believe that the next frontier are world models. This is the ability to simulate, predict, and react to real-world constraints and virtual changes in real-world constraints, virtual changes of design constraints within a virtual environment. That means you really have to partner with someone that's got deep expertise in spatial reasoning for AI.

13:58That's why we picked Working World Labs. And they picked us because we have deep expertise and deep data knowledge of what happens in a design and make process. The technical challenge is to apply the physics of the real world into the virtual world, particularly when you're trying to simulate. I'm assuming that beyond the investment, there is a commercial arrangement here, right, where you're able to do something with World Labs technology. How does that manifest? Yeah, so we're going to be working together with using World Labs technology with our tools. They're also going to get access to some of the proprietary models that we've built so that we can help them expand and create their ability to do deep spatial reasoning.

14:40So you'll see us initially lean into the media and entertainment space. But, you know, as you said, this is much more expansive than that. At Autodesk, we're really interested in the mid-market manufacturing space and mid-market factories, small and mid-sized factories in particular. This is going to be a high growth area. And these factories are going to need to be able to operate like the biggest factories in the world, but at a much smaller scale. And they need technology from us and technology from World Labs to do that. You put out the announcement about the$200 million part of the strategic investment.

15:11And actually, later in the morning, World Labs confirmed they did a$1 billion fundraise overall. So you're a big portion of it. What was the valuation that World Labs was assigned in that? You'd have to talk to World Labs about that. I'm not going to discuss what their valuations were, but you can reach out to that team. The question is, when does this become a real commercial opportunity for you then, Andrew? You know, when is it that you're going to be able to say in an earnings report, this is something that resulted from that strategic investment? You know, it's important, Ed, for us to recognize that in our space, AI is already returning value.

15:49I know there's a lot of hype out there, especially with the infrastructure buildup and what's happened with large language models. But a lot of software companies, particularly Autodesk, have been investing in AI incrementally for a decade. So we're already delivering value today around the ability to simulate certain things like flooding, lighting studies, the ability to simulate air and wind and noise studies using AI, the ability to build preliminary models with AI that take into account structural constraints or the need to remove structural constraints and helping people make better decisions in the construction cycle.

16:22So AI is delivering value in our space today. And regardless of what happens out there in the world, we're going to continue to invest over the next two, three, four, five years. So you'll see us continue to roll out incremental real value. So it's happening now. This is only going to allow us to accelerate deeper value into additional areas and into deeper parts of the design and make lifecycle. That's where I want to end it. You know, a lot of people ask, what's the net result of all of this? We focus a lot on digital twins, virtual worlds. We've discussed that. Is the result that, you know, big facilities, manufacturing sites, plants get built quicker, the people inside them get trained more quickly, humanoid robots are deployed in the real world more quickly?

17:04Here's the big story in our space, Eddie. There is a massive capacity problem. There is not enough money, people and materials to build and rebuild everything in the world right now. Right now, people are building lots of data centers, a lot of infrastructure that serves day-to-day people in their lives, road, rail, all sorts of things, doesn't get built while that's getting built. We need to execute more projects with the same amount of people in the same capacity we have today, and we need to do it at an accelerated rate. That's something that's very unique about the design and build world, this fundamental capacity problem.

17:40AI is going to be enabled to unlock that. Andrew and Agnes, Autodesk CEO, really appreciate your time on Blumbo Tech. Thank you. In other earnings news, Palo Alto Network shares plunging, having a difficult day after the cybersecurity company released a forecast for adjusted earnings that was weaker than anticipated, though it also boosted its full-year forecast for other key metrics, including revenue and remaining performance obligations. CEO Nikesh Arora told us why he thinks the market is wrong about its reaction to the earnings. Listen to this. I think the market is not paying attention to our numbers carefully.

18:18We just completed our largest acquisition in CyberArk. We had a great quarter. Our numbers are across every known consensus metric. The market is not understanding that our guidance for the next two quarters includes CyberArk. If you take the consensus of CyberArk and ours, you're actually getting above the collective consensus, but the market hasn't understood the dilution of shares. So I think they have it wrong. Coming up, Apple fast tracks AI wearable devices. We have the Bloomberg reporting next. This is Bloomberg Tech.

18:57Jeff Bezos' space startup Blue Origin has shifted resources from space tourism to lunar projects. CEO Dave Lim spoke with Bloomberg's Anne-Marie Horderne about the move while at the Defense Tech Summit. We're at such a pivotal moment around where space is. And there's never been more demand. We have years worth of demand for our launch business, as does everybody that's launching. We're very constrained as a country in terms of the number of launch vehicles that we have right now. And our adversaries around the world are not standing still. They have, they're putting hundreds of billions of dollars into launch and into space.

19:41And especially as it relates to the moon, which, you know, we haven't been on the moon for 50 years. We think it's really important for blue for the country. And it just made prudent sense from our standpoint to take the brilliant people that we're working on New Shepard, pause that for a while. We're planning for at least two years and then repurpose them to even further accelerate our efforts in lunar and launch. There's all this excitement, I think, about New Shepard, especially because Katy Perry went to space. And for civilians who maybe want to have that moment, like Mark Bezos had. So do you think you'll go back to that at some point?

20:20Is it really just a pause? I think we will. I think there's still, you know, we had multiple years of backlog and it was the easiest ticket to sell was that. And so I think we'll likely go back into that business. But again, at the moment in time right now, it just makes more sense to focus on the moon. So it really is a space race right now. I don't, you know, I'm very much pro-America and a capitalist, but I don't think we want another Sputnik moment. And we were the first to put boots on the moon. And I really feel like we want to put, you know, we want to put boots on the moon. And this time, the name of our group inside of Blue is Lunar Permanence.

21:05So we're not only racing to the moon to get there back again for the first time, but we want to keep people on the moon. Blue Origin CEO Dave Limp speaking with Bloomberg's Anne-Marie Horder. In other news, it's time for Talking Tech. And first up, Warren Buffett's Berkshire Hathaway cut its Amazon stake by 75 % in the fourth quarter, reducing it to about 2.3 million shares, according to a regulatory filing. At the same time, it built a new position in the New York Times with 5.1 million shares and continued trimming, according to the 13F document, its other holdings in Bank of America and Apple.

21:41Plus, Uber plans to invest more than$100 million to expand fast charging infrastructure for autonomous EVs in the US. The company will build high-capacity charging hubs in San Francisco, Los Angeles, and Dallas, the same markets where it's preparing to roll out a public robo-taxi service. And Apple is breaking away from its tech peers. Bloomberg data show the iPhone maker's 40-day correlation with the NASDAQ 100 has dropped to its lowest level since 2006 as it largely sits out the AI arms race. With rivals pouring money into AI-related capex, Some investors are starting to see Apple as an appealing alternative.

22:21Let's power on and stay with Apple. The company's fast-tracking work on three new wearables as it races to compete with AI-powered hardware. Sources say the tech giant is developing smart glasses, a pendant, and AirPods with expanded AI capabilities. Bloomberg Consumer Tech and Apple managing editor Mark Gurman broke the story and is with us. Okay, three distinct products, three distinct categories. Let's go through what you've learned from sources. You know, Apple, they are sitting out the AI race in a way, but they're going to enter in an Apple way. And that means hardware blended with AI, blended with software, blended with services.

23:00And they're going to get there if they're able to pull off this new Siri, which I believe they will before the end of the year. And so three hardware products. One is smart glasses. These are going to be quite significant. They're going to compete with the meta product. They're likely going to be a bit pricier. They're going to have dedicated cameras, one for media capture, one for computer vision. In terms of functionality, you'll get things like if you're taking a walk and you have turn-by-turn directions on, instead of it telling you make a right in 450 feet, it might say make a right at the gray building because it sees your surroundings or make a right past the white car.

23:36AirPods with cameras, very similar AI functionality without the media capture functionality built into the earbuds that people already know and love. And then the pendant, that's really an alternative product. That is an accessory for the phone. Your eyes and ears for the phone, for AI in a sense. Similar functionality to the glasses and the AirPods, but not everyone wants to wear smart glasses. And so if Apple wants to provide that eyes and ears to the world, it's going to need some of their product. And that's what the pendant's for. People aren't going to have AirPods in their ears at all times.

24:11People aren't going to want to put glasses on their face in all cases, especially if they don't have vision correction needs. So you need some sort of pin or necklace. So that's what they're developing as well. Just 15 seconds, Mark, literally. But do these come this year? When do they get released? I would expect the glasses and the pendant to be released next year and the AirPods as early as this year. Bloomberg is Mark Gurman, who leads all our consumer tech coverage. Thank you very much. There's some interesting stuff going on in private markets. And actually, the team at Bloomberg News do a really good job at going into those markets, looking at the documents and finding out stuff that actually all markets want to know.

24:52Sally Bakewell leads our finance team. And in this case, the reporting is about what we know from sources from some of those private names in software who are coming to us a little bit earlier than normal and saying, here's what's under the hood. Here's what we're learning. Just explain the reporting to us. Yeah, that's exactly right. So these are private firms. They are owned by private equity firms, which obviously borrowed in order to buy them. So these are company software firms that have bonds and loans outstanding. Now, a lot of debt tied to software has sold off in recent weeks, particularly software as a service, because investors are very worried about the threat of AI disruption, potentially supplanting some of those tasks.

25:34So we have seen billions of dollars of loans tied to U.S. software and U.S. tech slipping into distressed territory. Now, amid all of this concern, we have had a couple of private software firms actually opening their earnings books a little bit early to investors in a bid to calm nerves to say that our earnings and our revenue are trucking along just fine. So we had the likes of McAfee, which showed that its fourth quarter revenue held steady at about$626 million compared to the year earlier. We had Rocket Software, which actually saw a 5 % increase in 2025 revenue. And we had Perforce Software, which actually saw a slight decline, but it was just a slight decline.