Nvidia’s Estimate-Topping Forecast Fails to Impress Investors

0:00Some people treat ChatGPT like some kind of smart search engine, and some use it to get work done. ChatGPT Work is a new way of working in ChatGPT that can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work. It's designed to help you move from a chaotic starting point to a reviewable first version. So all the source materials, briefs, and scattered information that you have to grind through to turn into something useful can just become something useful. Put ChatGPT to work on your most ambitious ideas and projects. Get started at ChatGPT.com by selecting Work Mode, available on Plus and Pro plans.

0:42Hi, I'm Tom Keen, inviting you to join me for the Bloomberg Surveillance Podcast. It's about making you smarter every business day. I'm Paul Sweeney. We bring you complete coverage of stocks, bonds, commodities, even crypto. All the information you need to excel in the markets. And I'm Alexis Christophorus. Listen to us for essential conversations with the smartest names in economics, finance, investment, and international relations. That's the Bloomberg Surveillance Podcast. Subscribe today on Apple, Spotify, or anywhere you listen. Bloomberg Audio Studios. Podcasts. Radio. News. This is Bloomberg Business Week Daily, reporting from the magazine that helps global leaders stay ahead with insight on the people, companies, and trends shaping today's complex economy, plus global business, finance, and tech news as it happens.

1:37The Bloomberg Business Week Daily podcast with Carol Masser and Tim Stenevek on Bloomberg Radio. I want to bring in Kunjan Sabani of Bloomberg Intelligence. He's senior semiconductor analyst. Kunjan, I'm wondering, you know, we haven't had a chance to talk to you. It's been about 13 minutes since these numbers crossed. Obviously, you were prepared for them. Apart from that outlook on the top line, what sticks out to you from this print? Yeah, there's a slight miss on the gross margin. I think that is going to get a lot more questions. You know, in our preview, we talked about the three key areas people were looking at.

2:13One is that sort of circular financing angle that Ed really laid out very clearly. Second was this gross margin. Remember, Nvidia has announced price increases. That implies that they need to increase their price to protect that 75 % gross margin. They only came in a tad bit below, but it will be interesting how the outlook looks when the Vera Rubin ramp starts fully in the fourth fiscal quarter. So that is something we're going to pay very close attention to. Kuchin, I've got to ask you, their shares have kind of been on this wild ride, right? They sink in the winter. In the spring, they kind of bounce back.

2:46In the summer, it's been kind of a little bit of a little bit of a little bit of a rate. bit all over the place. They're coming off a seven-day losing streak as well. So we look at NVIDIA, it's up about 14 % in 2026. It's a decent showing, a little bit of far cry from past performance. What can you say about, is that all concerns about the AI trade? Is that simply exactly what it was or is there more to the story? No, it's simply that is, you know, look, NVIDIA is the overall AI trade right now, right? So that's what the concerns are. And right now, I think we in that bubble gap timing-wise this quarter and next quarter, where everyone's trying to figure out what happens in 28.

3:22And until you don't get the clarity that this AI wave continues or accelerates in 28, people are trying to sort of stay on the sidelines. And what is up with gross margins, that bottom line figure there? Yeah. Yeah. I mean, I'm just, I'm reading back over it. I'll just recount it. So in the quarter gone, adjusted gross margins was 75%, bang in line with expectations. For the fiscal third, the guide is 74 % plus or minus 50 basis points. I think consensus was 74.8 % on the adjusted gross margin. Kunjan just said that NVIDIA has announced price increases. Maybe he knows something I don't. But all I'll do is say that Bloomberg had reported that the server makers were informing customers that starting in January 2027, that any system shipped from that point, both Blackwell and Vera Rubin would be subject to higher prices of 15 % or more.

4:15And so like right now, the environment is hard right now. You know, memory pricing, we've talked about the supply commitment expanding, but I don't get a sense that there's an explanation of what the impact on margin was contemporaneously, right? I don't see that in the release. So, you know, on the call, maybe there'll be just some discussion of that. You You know, what are the real-time impacts of higher memory? What are the real-time impacts of fighting for fab capacity? You know, we just don't have answers there. Hey, Kunjan, we're getting questions from people around the world right now who are tuning into this.

4:50This one's from pretty close to home, actually. Geary in Edison, New Jersey. This is actually about what Ed was just talking about, sending a question to us about NVIDIA informing its customers that the prices of servers with its chip are going up by 15%. That's what you were just speaking about, right? Yeah. Okay. That was a Bloomberg report last week, or I can't remember. Yeah. Okay. So Geary asks, and I'll pose this question to both of you. I'll start with you, Kunjan. What percentage of it is directly going to its top line, and how much is going to DRAM producers, Micron, Samsung, and the like, and what percent is going to server manufacturers?

5:24So they're just curious about who the beneficiaries of this is. Is it CoreWeave? Is it Nebius? Is it Dell? or is it the Samsung and microns, Samsung's and microns of the world? Yeah, from what it seems like, I mean, the company has not publicly discussed, but our estimate would be it's all going to offset the rise in memory prices. Remember, NVIDIA purchases the memory from the memory suppliers and has a pass-through model. So in order for it to protect its margin and the cost of its input, which is memory, and wafers from TSMC as well, that has also increased in prices. It needs to raise its prices.

5:58So most of this is going for that input rise from wafer, so to TSMC and the rest to the memory makers. Yeah, do you have anything to add to Giri's question about this sort of, again, not just this, we don't get this information from the company. But just based on your reporting and you report on a lot of the Neo cloud providers as well as the DRAM producers. Well, the way that I put it, and I kind of throw it back as a question to Kunjan if I can, is just remember how the world works. So like if you are a NeoCloud or a hyperscaler or you're a large software company that has their own sort of data center on prem, you don't walk into Santa Clara and say, hi, Jensen, can I have a hundred of those?

6:38It's not how it works. There are companies like Dell, HPE, Supermicro that assemble the servers. And historically and traditionally, they are also the sales channel, right? And so like our reporting was based on the idea that those server makers told NVIDIA customers like prices are going up. NVIDIA owns more of the content of the server design, but I'm trying to understand where and when this price hike gets initiated. Do you see what I mean, Kunjan? I do. And again, the flow through of the whole bomb, starting from re-adding memory, NVIDIA chips, and the work that the OEMs, the Dells put in.

7:13So this is going to flow through everywhere. And again, it's coming from the back front end of the supply chain. So rising input wafer cost has to be passed through from the chipmaker. Rising memory prices has to be eaten by the chipmaker, which then in turn is increasing prices, which forces or necessitates the Dells and OEMs to increase their prices as well, unless some party within the supply chain decides to eat that cost. Kunjgan, can you talk about looking ahead, right? Can you talk about where NVIDIA's next phase of growth is coming from and how important is China to their growth? Yeah, I mean, look, growth is sort of being driven mostly from the data center.

7:52But within the data center, the ACIE segment is something we estimate in the long term is going to grow faster from a growth rate perspective. Hyperscaler still being a big piece. China, again, is going to be a key question still today. Zero dollar estimate of China in current consensus. We have heard news about some H200 being shipped to the sort of numbers of a few billion dollars, but nothing has been clarified there yet. So anything on China will be an upside to that numbers. However, what I want to point out, you know, I think there's a lot more being made around China, even if they can open up and start shipping initial samples.

8:30That's about sort of 10, 20 or so billion dollars per year, what they used to get from China. Given the size of their business today, I don't think that's going to move the needle a lot. NVIDIA earnings are out. Okay, so we've been speaking with a great panel of people. First of them, there was Kunjan Subhani, who is a Bloomberg Intelligence Senior Semiconductor Analyst. We also have Bloomberg Tech's Ed Ludlow, who we want to bring back in before he has to scooch out of here as well. Ed, we've gotten all the numbers. We've had the breakdown of it. I want to pose a different kind of question for you.

9:00Is NVIDIA selling chips or is it effectively selling the infrastructure that's going to kind of power that next generation of the internet. Yeah, it's interesting. I give Kunjan a lot of credit because he was ahead of the game on this a few quarters ago. When you say Blackwell or you say Rubin, you are talking about the specific design of a GPU. But in the real world, you're talking about a server design where it's a number of GPUs in combination with the CPU. And in that tray, lots of trays go into one cabinet, which is a server. Increasingly, NVIDIA has more ownership of the content of it. And Kunjan posed a really smart question a few quarters ago, which is where does NVIDIA sit with that?

9:42And the answer that I had got from Jensen Wang at the time, and I don't recall which specific quarter it was, but let's say it was in the middle of last year, was that NVIDIA now is well north of 55 % of ownership of the content of a server design. They've tried to do a lot more of it. It's one reason why the server makers have found margin pressure and not been able to make much money from this exercise, because NVIDIA just does so much of that. And so this is the bit I've been trying to unpick what's happening in the world. Why are margins impacted? Why is the growth outlook such that it is?

10:12Because NVIDIA relies on server makers to get these products into the real world. But increasingly, it just kind of wants to do it all itself. I have to ask you about Jensen Wong again, because I asked you before about him. One market strategist called Jensen Wong the kind of Pope of AI. They said he gets to bless any of the deals. Is that how a lot of Wall Street is seeing it? Do you see him kind of as his founder, who blesses all these deals that are happening? Yeah, I mean, people call him the godfather of AI. And, you know, a lot of people would make the point that where we are to date, NVIDIA powered all of the training of the AI models that put AI into everyone's consciousness, right?

10:56Now we're in a phase where we are transitioning to inference. That is running models that have already been trained. And that is where there will be more scrutiny and debate, because to this point, Jensen Wang has sort of dismissed the threat of other chip makers. That could be an AMD that has a GPU-based platform. It could be custom silicon that the Frontier Labs are working on themselves. And in a world where it's more about running AI models, those custom silicon designs on paper are more competitive with NVIDIA's technology. And so, you know, maybe that's a question on the call. One which to this point, again, to answer your question, Jensen Wong has been so dismissive of those.

11:36He would say NVIDIA has the best technology and the software library to go with it, but it increasingly is being looked at by the world of what the benefit of using other chips that are not NVIDIA might have for the world of AI. They say you can't be in more than one place at one time. Ed Ludlow proves that is not the case as he joins us here on Bloomberg Business Week Daily as well as on the live blog, TLIV Go, on the Bloomberg Terminal. Ed's got to go. He's the Bloomberg tech host. Join him every single day, Monday through Friday at 11 a.m. Wall Street time on Bloomberg Tech. We've got with us Kunjan Sabani, Bloomberg Intelligence Senior Semiconductor Analyst.

12:11Kunjan, I want to talk China with you. We're getting some CFO commentary and getting some specific numbers about China. I'm curious about the China story here right now and the opportunity in China. Colette Kress, NVIDIA's CFO, says that sales of Hopper products in China in the quarter were less than 1 % of data center revenue. In the previous quarter, there were no such sales to China. Big deal that it's now 1 %? Not at all. And like, you know what I was saying, right? Even if they open up the doors and go back to their previous peak China revenues. Again, look, we're talking about a company that will probably is on target to do a trillion in about two years.

12:54even if you go bring back that 10, 20, even 30 billion back China revenues. Fundamentally for us, it doesn't move the needle. Yes, it's a good storyline. It's a good top line of opening up the market again. But right now, the growth that they're getting from outside China is significant. One additional thing that just crossed the wire is the 4.5 gigawatt Ohio plant arrangement where they're backstopping the sort of that financing. you know, the newswire is saying the revenue exposure there could be as high as$200 billion, right? So one deal here like that in the U.S. bringing in other$150 to$200 billion really offset anything that China could bring in in the near term.

13:37Well, on that note, I want to also talk about competition in the chip space. I mean, you heard about OpenAI talking about their jalapeno chips, right? You have AMD, Google's TPUs, custom chips by Amazon, Microsoft, Meta, all that said, What's your take on the competition and where NVIDIA stands in that space? Yeah, look, the truth is competition is rising. I mean, the initial specs that the OpenAI chip announced this week were outstanding and very surprising to us that they are able to deliver on this speed, not being a semiconductor company. So there's definite competition rising. We don't expect that to go away.

14:15we expect the unit share volume for the ASICs, which are the chips that NVIDIA's customers are designing for themselves, to continue to rise. However, the demand environment is so strong right now, that doesn't mean that NVIDIA's revenue growth will stall. The tide is lifting all boats. NVIDIA is trying to use its balance sheet, use its free cash flow to sort of become, I'm stealing this from an investor I heard from last night, but sort of the bank of Jensen. So they're going to continue doing what they're good at, leveraging and out-innovating everything else. But we expect competition to continue rising.

14:50Forget the bank of Jensen. I want the bank of Kunjan Sabani because there is just a wealth of information there. We love it when you join us, Kunjan. Stay with us. More from Bloomberg Businessweek Daily coming up after this.

15:05The thing about AI for business, it may not automatically fit the way your business works. At IBM, we've seen this firsthand. But by embedding AI across HR, IT, and procurement processes, we've reduced costs by millions, slash repetitive tasks, and freed thousands of hours for strategic work. Now we're helping companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create smarter business, IBM. At Radiolab, we love nothing more than nerding out about science, neuroscience, chemistry. But we do also like to get into other kinds of stories.

15:46Stories about policing or politics, country music, hockey, sex of bugs. Regardless of whether we're looking at science or not science, we bring a rigorous curiosity to get you the answers. And hopefully make you see the world anew. Radiolab, adventures on the edge of what we think we know. Wherever you get your podcasts. You're listening to the Bloomberg Business Week Daily Podcast. Catch us live weekday afternoons from 2 to 5 Eastern. Listen on Apple CarPlay and Android Auto with the Bloomberg Business App. Or watch us live on YouTube. Well, Wall Street certainly focused on NVIDIA. We're getting news in another part of the semiconductor ecosystem, and that's from Lamb Research, the nearly$400 billion market cap developer of machines that are used in the chip manufacturing process.

16:33The company out with news this afternoon that it's breaking ground on a more than$600 million expansion of its Tualatin campus in Oregon. It's just south of Portland, part of a more than$3 billion plan investment in its labs around the world. We've got with us the CEO of Lamb Research, Tim Archer, who joins us alongside Ed Ludlow, the host of Bloomberg Tech. Ed's out there in San Francisco. Tim is out there in Tualatin, Oregon. Tim, I want to get to this project. We'll get to it. But before we do, shares of Lamb Research just surged at the end of last month. This was just after you reported results closing higher by close to 18%.

17:07You reported fourth quarter results in an outlook for adjusted earnings that beat expectations. You said in a statement at the time that record revenue, operating margin, and earnings per share at a record as AI-driven demand continues to reshape the semiconductor industry. Can you give us some color of what you've seen day to day over the past four weeks that show what you're seeing with demand right now? Well, thanks very much for having me today. It's, look, I think those results, you know, just indicate we're in an accelerating demand environment for semiconductors. And, you know, what LAM does, I think most people realize, but, you know, we make the machines that help make all of those chips.

17:50And especially our strengths are in some of the most advanced manufacturing processes for chip making. So we're seeing tremendous demand with the AI build out and all the chips required to support AI. And I think that just week by week by week, we continue to just see not only really we don't talk a lot about capacity. In fact, the announcement we're making here today is about a new R &D lab. We're just seeing an acceleration of the pace of innovation, the capabilities of each of those chips. And it's something that we're investing to stay ahead of that demand. Tim, it's Ed in San Francisco. And you're right.

18:27We don't really talk about at least lamb in the context of more capacity. Your challenge is to make increasingly tiny three-dimensional structures at atomic level precision. And that's only getting more difficult in the AI age. How do these lab investments help you get there? You know, one of the keys to really accelerating innovation is working very closely with our customers. And so one of the things that's very unique about the announcement and, you know, the groundbreaking we have here in Oregon. I mean, Oregon has been kind of the heart of the Silicon Forest for more than three decades. And so a lab here in Oregon puts us close to a lot of the advanced chip making that happens very close by here, by some of our very large customers.

19:12And we also have labs in Asia and Europe as well to put us close to those customers. We think that by shortening that distance and really building incredibly tight collaboration with our customers who understand what those next problems are to be solved, that accelerates the velocity of innovation with them for those next generations. So that's what's unique about our lab strategy and the type of work that's done. So, Tim, why is the timing on this? I mean, why is LAM investing so heavily now? What is semiconductor manufacturing technology matter? Yeah, if you think about this, this lab here is part of a much bigger investment that we announced.

19:55It's$3 billion across that global lab network. And why now? The pace of innovation. AI has fundamentally changed the rate at which we have to be innovating. You know, I've looked back just a few years ago, and when the markets were driven by things like smartphones and PCs, that cycle was operating, you know, kind of on every one to three years, you needed to really innovate something new. AI has changed that requirement such that with new AI models coming every few months, and each of those demanding more computational power, even better high bandwidth memory, even more storage. You know, LAM is looking to meet those needs.

20:36We are a manufacturer of machines that make each and every one of those types of devices. And so there's a lot of demand for us to continue to come out with the next great innovation and process and manufacturing technology to keep up. We've also seen a tremendous boom in the what's called advanced packaging. It used to be that in most cases you built a single chip and it had a certain function and you increased the performance of that chip by making it smaller and introducing new materials. Advanced packaging changes the game because now it is about putting many different types of chips together to create system level performance.

21:16And putting all of these chips together and stacking them into a three-dimensional architecture, it requires a lot of etch and deposition equipment. We said that our business in advanced packaging this year would be growing more than 70 % year on year. And one of the particular tools, the copper electroplating tool that plays such an important role, that tool is actually developed and manufactured here in Oregon, which makes this such an important site for us. Tim, you know, it's really important to underscore NVIDIA, as an example, is not your direct customer, right? It is TSMC and the memory makers.

21:51But, you know, it is the demand for lead edge chips across logic and memory, HBM, that is the story out in the world. How much of that growth in demand directly reads through to LAMM? And how do you respond to it? Yeah, if you look at advanced chips today, one of the trends that just is, it's been around for a few years, and AI has really put it into hyperdrive, is the three-dimensional scaling of every device. And so, of course, that's been true in foundry logic with technologies like gate all around, introduction to gate all around, which is a three-dimensional transistor structure. But really, it comes to bear when you think about what high bandwidth memory is.

22:32High bandwidth memory is the stacking of multiple DRAM chips on top of each other to create higher bandwidth performance for AI. And so clearly a three-dimensional architecture that requires a lot of etch and deposition equipment and a lot of innovation for us to be able to support the manufacturer of those structures. NAND has had a 3D architecture since more than 10 years. And LAM has been a very strong player. In fact, it has traditionally been our strongest market because of that 3D architecture. And what we're seeing now is as other devices go 3D, it just drives a lot more demand for our specific types of equipment.

23:15Hey, Tim, we only have 20 seconds left, but you mentioned this is part of a$3 billion effort to expand your global lab network over the next five years. You made that announcement just a couple of weeks ago. I'm curious where you think we are in the cycle right now and why you're so confident that demand will still be high. in five years? Yeah, well, the need for innovation never ends, regardless of where you are in the cycle. And so, one thing we've seen, I've been in the industry more than 30 years, lots of cycles. And so, regardless, the investment in innovation, the investment in R &D labs, putting yourself close to the customer to accelerate collaboration, clearly we think we're in the early stages of a big build-out.

23:54But regardless, we need to invest in innovation, and that makes the company stronger throughout any cycle. Tim, appreciate your time this afternoon. Tim Archer, CEO of Lamb Research. Also, always appreciate Ed Ludlow's time. He's the host of Bloomberg Tech. Stay with us. More from Bloomberg Business Week Daily coming up after this.

24:17You're listening to the Bloomberg Business Week Daily podcast. Catch us live weekday afternoons from 2 to 5 Eastern. Listen on Apple CarPlay and Android Auto with the Bloomberg Business app, or watch us live on YouTube. Watch and share is a meta up right now by about one and a half percent. They bounced around quite a bit today. They were up as much as 4%, down as much as 1.4%. This after the company said it agreed to pay up to$18 billion in landmark settlements to resolve allegations from multiple U.S. states that the company deliberately designed Facebook and Instagram to encourage compulsive use among young users.

24:54For more, we're joined by Alexandra Levine, Bloomberg News, social media and technology reporter. Also with us, Stacey Steinberg. She's a professor and director of the program on children and families at the University of Florida Levin College of Law. Alex is here on set with us in New York. Stacey is with us from Gainesville, Florida. Alex, I want to start with you. I mentioned the$18 billion and up to$18 billion in landmark settlements. What else do we know about the details of this? Well, we always knew or we were always expecting that there could be some big monetary piece of this. I think perhaps The most interesting piece, though, is that there are these measures that would essentially force the company to change the way that Instagram and Facebook operate for people who are under the age of 18.