Has The NASDAQ Topped? (Uncut) 02-24-2025

Has The NASDAQ Topped? Is NVIDIA Still A Buy? Analyst Answers | John Belton

But I think the West is still, and specifically the U.S., is still at the leading edge when it comes to generative AI technologies. And then I think the U.S. is still far ahead of China when it comes to companies in the real economy having adopted and having started to build out useful AI applications, which I think is sort of the next phase of the technology. Is NVIDIA still a buy? John Belton, portfolio manager of Gabelli Funds, is here with us to talk about the tech landscape, tech earnings and what’s next for investors.

Welcome to the show, John. Good to see you. Thank you, David.

Good to be with you virtually. Good to see you virtually as well. Let’s talk about talking about the overall macro backdrop for tech.

The S&P 500 and the Nasdaq have seen volatility up until basically the second week of February. Now, as you know, a lot of the scare from the tech side came from DeepSeek’s introduction a couple of weeks ago, which shook roughly a trillion dollars off of market cap of global stocks, especially from NVIDIA on that day. Now, much of those losses were recovered, but we were still trading range bound.

What is going on with the Nasdaq right now? Is this the top? Yeah, so I would I think there’s a there’s a bunch of moving pieces under the surface here, I think. First, understand the context, you know, what’s the starting point for this market? We entered 2025 following back to back 20 percent plus years total returns in the S&P 500. We entered with sort of a top decile type valuation, talking next 12 months price earnings ratio at kind of the index level.

No shortage of sort of uncertain dynamics happening in the background with the new U.S. administration. Unclear what kind of the policy platform is going to be. So I think we entered this year with stocks priced kind of perfect, kind of for perfection, despite a bunch of uncertainty.

So I think that’s important context to have now. Obviously, Deep Seek, you mentioned, you know, we can talk more in more detail about that, but that is that was a big event for sort of the tech space specifically within the market. You know, my opinion is Deep Seek.

It wasn’t a it was more evolutionary than revolutionary. I’m not sure how much radically shifted within kind of the AI landscape. I think it was more of just a slight breakthrough was reached.

It happened against sort of a nervous market backdrop to begin with. And it became more of like a valuation event for the market than something that, you know, radically shifts the fundamental outlook for a lot of these companies. So last part of your question, is this the top? I mean, I think it really depends on your time horizon.

I think a lot of the dynamics I touched on at first are still at play, still not a particularly cheap market valuations historically haven’t had too much correlation to short term returns. But, you know, I do think your starting point is hard to bet on multiple expansion here. But at the same time, the economy is very resilient, a lot of reasons to believe in and be excited about AI.

There’s a lot of new things the new administration is doing should be kind of more business friendly. So I think if you have a long term perspective, a long term time horizon, stocks are still a good place. Hard to call a nearer term.

How significant is the deep risk, deep seek risk to American tech stocks if there is still such a risk? Yeah, so I would I would frame it this way, you know, step back. What did Deep Seek do? I think Deep Seek came up with a couple of really creative engineering techniques that. Successfully allowed them to train a very capable set of models at a without the same type of eye popping investments that a lot of Western peers have have done.

So, you know, I think the question is, how significant were their breakthroughs from a financial standpoint? How repeatable are those breakthroughs going to be for other companies? And, you know, I guess how out of the ordinary was this to begin with? Let me take that last point first. I think in every computing cycle, historically, increasing efficiencies and therefore lower costs. Is it just that’s a given that’s that’s that’s been around since Moore’s Law many decades ago? I think you’ve seen the same thing in the AI space, in the LLM and transformer model space.

More recently, you know, there’s a really good blog post that the CEO of Anthropic posted a couple of weeks ago where he’s he said he’s seeing something like 4x cost or efficiency improvements in model training and model inferencing costs annually is sort of the the run rate he’s been observing. Sam Altman, CEO of OpenAI, put out a blog post today. He’s seeing more like 10x cost improvements per year.

So I guess the point of all that is to say, you know, what DeepSeek did, part of it is just on this natural evolution, the natural curve of growing efficiencies in AI computing broadly. So while this is to be expected now, to be fair, I do think they did reach some breakthroughs that sort of accelerated that curve a bit for them. I don’t think it’s 100 or 1000x improvements, but I do think they did discover some things that other companies can use to drive down computing costs.

So I think the natural next question to ask is, what is the fallout from cheaper compute when it comes to AI? And I think it’s very early to say. I think the impacts are going to be different for different companies. But one thing that is for sure, going back to sort of historical patterns in computing cycles, is cheaper compute has never meant less usage or less spending or less investment.

Cheaper compute, on the contrary, has always meant more usage, more investment, more spending. I don’t see why that’s not going to be the case here. I think this is a development that’s very good for anyone who’s using AI technologies, anyone who’s spending on AI compute, whether that’s consumer Internet companies or enterprise software companies or companies across various enterprises that are trying to make value out of AI and trying to create useful AI applications.

For the sort of arms dealers, the pick and shovel companies, the NVIDIA’s of the world, the Broadcom’s of the world, some of the data center industrials, I think we’re going to see how this evolves. I think for me, this is a little bit more of a technical point, but I think the days of sort of endless investments in massive training data centers are probably we’re closer to the peak in that investment cycle. But I think at the same time, we’re just on the precipice of this new big investment cycle to build out inferencing data centers and inferencing infrastructure.

So I guess the short answer to your question would be, I think this is more evolutionary than revolutionary. I think it is for sure a positive for anyone who’s using AI technology, investing in AI technology for the chip companies, the data center companies. It should be a positive, but we’ll see.

Let’s talk about NVIDIA itself then. Is NVIDIA still a buy? Yeah, so I think NVIDIA from my seat really still has no real competition. I think for any customer who wants sort of the full stack Gen AI compute solution, there is no alternative to NVIDIA.

I don’t see that changing anytime soon. I think what they’ve built with CUDA, what they’ve built with their networking products, what they’ve built with kind of their modular computing equipment set up, I just don’t think that is going to be there’s no real competitive threat there. I don’t see that changing.

I think the valuation on NVIDIA is still reasonable. I think nearer term, pretty good conviction in numbers. I think the hyperscale AI investment cycle very clearly coming out of this earnings is alive and well.

I think that is at least continuing through this year into next year. NVIDIA, obviously the big beneficiary there. I do think NVIDIA is a semiconductor company.

Semiconductors, by definition, it always is a cyclical space. So I think, you know, just got to be cognizant of putting a peak, peak type of multiple on peak earnings. But I don’t think we’re anywhere near the peak in terms of earnings yet.

So I’m still comfortable holding NVIDIA kind of over the medium term, medium to long term here. Chinese tech overall, there’s been some mutter or rumors in the in the web sphere that that Chinese AI has now surpassed the likes of American AI. We have Alibaba’s Quen that came out as well.

That’s made headlines. Apparently its model is more advanced. And of course, we talked about DeepSeek.

Do you think that Chinese tech is is basically the end of American tech dominance? In other words, investors should start focusing their their capital deployment to the east? I think absolutely not. I do think DeepSeek and I guess for that matter, also some of what Alibaba released last week. It’s a bit of a wake up call to suggest that development in AI technology in China has been actually much more robust and advanced than we had been thinking.

Just to clarify with DeepSeek, I think it’s important to say, you know, for sure DeepSeek had some has some really brilliant engineers, has has achieved some really impressive things from an engineering standpoint. But these models that DeepSeek has released, both their sort of language model, V3 and their their reasoning model, they don’t really I’d say they don’t really stack up to the state of the art frontier models from Western companies. They’re still a little bit behind.

So just getting that out there. And then where China really is behind, I’d say, is more on sort of the the manufacturing side, the chip design side and the tighter and tighter that U.S. export restrictions go. I think the harder it is going to be for China to continue to keep up.

And they’re just, I think, going forward, not going to have access to the most advanced compute resources like Western companies have. So I think it shouldn’t be discounted. But I think the West is still and specifically the U.S. is still at the leading edge when it comes to generative AI technologies.

And then I think the West, the U.S. is still far ahead of of China when it comes to companies in the real economy having adopted and having started to build out useful applications, which I think is sort of the next phase of the technology. Let’s talk about tech earnings now. So on Monday, the S&P closed higher.

The Nasdaq popped up nearly one percent. Investors were shrugging off tariff warnings. A lot of this had to do with optimism, I think.

Walk us through some of the major earnings updates that you were following. Yeah, I’d say by and large, large mega cap tech earnings have been pretty pretty constructive so far this earnings season. A couple of themes that we’re noticing.

So one, I’d say the big Internet advertising companies clearly benefiting from a robust consumer spending backdrop, talking meta and Google in particular here, even some of the midcap advertisers, though, have also been reporting pretty good numbers this cycle. But what’s really clear to me about meta and Google is AI is having a an amazing, amazingly positive impact on their core advertising businesses. When you look at some of the engagement trends that they’re reporting, which is being helped by AI powered content recommendation algorithms.

When you look at some of the ad pricing data that they’re reporting, which comes from some of these ad targeting engines they’ve built with AI, these businesses are using AI to drive strong revenue growth. It’s working and they’re further benefiting from a robust consumer spending backdrop. So those companies, I’d say on the top lines, have looked really good this earnings cycle.

I think in terms of the public cloud companies, the themes coming through there are every major U.S. public cloud infrastructure provider. So Amazon, AWS, Microsoft, Azure and Google Cloud, every one of them is reporting still being supply constrained when it comes to the AI services side of their business. So they don’t have enough data center capacity.

They don’t have enough GPU capacity to fulfill all the demand that’s out there. So I think that’s pretty bullish signal for for AI services, despite not being able to keep up with that demand. These companies are reporting extraordinary growth in AI revenues.

So looking at Microsoft, Microsoft, Azure or sorry, Microsoft across the company has already reached a 13 billion dollar AI revenue run rate. That’s up from 10 billion dollars a quarter ago. So still seems pretty early in the AI revenue growth curve for most of these companies.

And then I guess the third point to make about mega cap tech earnings, clearly one of the big drivers of these stocks over the last few years that probably gets under talked about, given all the exciting stuff that’s happening in these business is cost controls and margin expansion and really profitability trends, particularly thinking about Meta, Google, Amazon, these companies that have radically re-engineered their cost bases and transformed their P&Ls and been reporting extraordinary earnings growth. I think that continued this earnings cycle, but was a little bit more mixed. Meta put out a big initial 2025 expense guide number.

Google missed on margins. The market really didn’t like that. So I think that story is probably in later innings.

And in fact, a lot of these companies that are signaling they’re kind of going to be picking up particularly capital spending this year. So that’s kind of the other theme is later innings in that profit margin expansion story for a lot of these mega cap tech companies. I wonder whether or not the government could affect the tech landscape.

Take a look at this. Open AI looks across U.S. for sites to build. It’s Trump-backed Stark at AI data centers.

As you know, the news came out a couple of weeks ago that Trump has announced its Stargate partnership, up to 500 million dollars of investment into large scale data centers and energy generation needed for further AI development and partnership with a lot of different companies, including SoftBank. What is going on here? Will this actually feed into better earnings for some of these tech companies that we talked about? Well, one amazing thing about this project is that this was unveiled, I want to say it was one or two days after the inauguration, happened to be unveiled, I think, three or four days after DeepSeek released their newest model and a little over a month after DeepSeek released kind of the initial V3 model with the initial paper outlining all the breakthroughs that they made. So point of that is this massive investment project was announced after all the DeepSeek news was very much known by tech executives, policymakers, investors, et cetera.

I think that’s just one sort of interesting wrinkle here. To your questions, you know, what is this? Do they have the money? How big is this going to be? I think there’s a lot of unknowns. There’s a lot that hasn’t been disclosed yet.

How is this project going to be capitalized? What exactly are those $100 and $500 billion numbers? Are those CapEx numbers? Are those total cost of ownership numbers? Are those annual? Are those cumulative? We don’t know. I think a lot is not known. What we do know is there is an operational facility in Texas that’s being operated by Oracle Cloud, where I think they’ve called that kind of the first Stargate training facility for OpenAI’s models.

There’s been reporting that OpenAI is the only customer for Stargate. I think that’s another very important thing to monitor. Who are the customers? Is it just OpenAI or is this going to be sort of a brand new AI cloud that’s going to be open to other third parties? I think what this signals to me is the fact that the press conference happened in the White House with Donald Trump.

That’s very symbolic to me. That signals the administration is taking AI very seriously, is prioritizing AI, is hoping to champion companies that are ready to double down and triple down on AI investments. I think that’s very significant.

I think this is also a clear positive for Oracle, kind of gets Oracle a seat at the table with the biggest cloud infrastructure platforms. We’ll see. I think it’s hard to have real conviction on how this is going to unfold, other than to say, I think that first $100 billion number, you can pretty readily paint a picture of how that gets funded.

Beyond that, I think it’s going to depend on a lot of factors that are still kind of unfolding. Which companies, if any, will be the biggest beneficiaries of this, besides Oracle? Well, NVIDIA, I think, just effectively gets a fourth or fifth big hyperscale customer here. I think this is probably a good thing for OpenAI.

Sam Altman has been on a world tour for the last few months trying to raise as much money as he can to go out and invest in infrastructure, to go out and invest in data centers and computing kit for his kind of model building aspirations. I think this is a clear, huge win for him. And then, yeah, I think those are kind of the obvious… And then some of the other companies that kind of go into the AI cluster, computing cluster building equation.

So whether that’s Broadcom, who it seems like they’ll be providing some of the networking equipment here. I think those are sort of the obvious winners. But then, again, I think the only other point I’d make is this is a big endorsement for AI technology more broadly coming from the top, coming from the White House.

And I think that should be a good endorsement for the space. Some other tech names we can take a look at. So here’s Microsoft.

And Microsoft’s been trading range-bound basically ever since last summer. And it hasn’t had a good start to 2025. What is Microsoft doing to compete on the AI front right now? And have there been disappointments in recent developments for you as an analyst? Yeah, Microsoft was… I’d put their December quarter earnings in sort of the mixed bucket.

There was a lot of positives to take from that earnings report. They raised their operating margin guidance. They, as I mentioned earlier, announced AI revenues have reached a $13 billion annualized run rate up from $10 billion a quarter prior.

Generally reporting pretty strong fundamentals across the company. There was a slight miss in the quarter versus expectations at the Azure business. And the miss was specific to the non-AI side of the business.

And within that, specific to some of their smaller customers, where they’ve kind of reshuffled a bit of the go-to-market motion. And so it was a little bit disappointing because they’ve been kind of preparing the street for this nice acceleration in Azure revenues in the first half of calendar 2025. And given this sort of go-to-market execution mishap, it doesn’t seem like they’re going to deliver that.

So I think the street or the investors were not happy about that. I still think when you look at Microsoft over a one to three-year time horizon, I think it’s a matter of time before the Azure business starts to reaccelerate. I think the core sort of cloud migration story is intact.

I think Azure is very well positioned from an AI infrastructure standpoint, particularly given all we talked about earlier with supply being constrained on the AI side. And then where Microsoft really stands out is they’ve got the ability to potentially start monetizing AI on the application side. So they’re already doing that with GitHub Copilot, which is basically probably the industry leading coding assistant.

Many million developers using it, several hundred million or more in ARR for that effort last I checked. And then the real dream is if they can start to monetize generative AI at Office as well with the Office Copilot suite, which has definitely been slow to gain uptake and slow to start monetizing. But I think that still exists as a long-term opportunity.

So I still think Microsoft is a blue chip. Any tech portfolio, that’s sort of a cornerstone. I don’t see that changing.

I think there’s a lot of reasons to be excited, just may have to wait for another quarter or two of kind of cleaning things up on the Azure side. Yeah. And here’s Google as well, Alphabet.

Gemini has not been, I guess, the core driver of growth, at least for the share value perspective. Walk us through whether or not Gemini is still positioned to be a competitive product and with the future of Google overall. Yeah, I think that the conversation there, you probably break into two buckets.

One, how can they use Gemini to help their core existing businesses? And then two, what types of potential future business models can you conceive of for Gemini? And I’d start by saying Gemini generally scores at or near the top on most sort of AI model tests. So I do think Gemini is regarded in the industry as one of the leading foundation models. And so on the point about Google’s core business, Gemini specifically, and just their use of AI models more broadly, they’ve clearly had a lot of success bolstering their core business.

Things like YouTube content recommendations, both search and YouTube ad targets across the company with more than two billion monthly active users, which is kind of an amazing stat. And they’re using every one of those services right now. So I think that the point there is they have this huge opportunity to just integrate Gemini, integrate generative AI into multiple aspects of their business to improve usage, improve monetization.

I think they’re doing that. I think that’s working. The second sort of bucket, can Gemini become a standalone business in its own right? And can Gemini start to really move the needle for Google by creating new revenue streams, new monetization? I’m a little bit less optimistic about sort of the business model of these foundation models as standalones.

That’s not to say that Gemini is not a great asset that again, they’re going to be able to use to bolster their core businesses, but I think it’s very much wait and see in terms of can Gemini kind of create new revenues for them. So the last thing I’d say on Google stock here, separate from AI, but related to AI, you see on the chart you have on your screen there, kind of the earnings blow up that Google had. And that goes back to a concept that I referenced a little earlier.

What caused that blow up was the profitability performance, Google’s EBIT margins, which missed in the fourth quarter under a new CFO. And so I think where Google really outperforms as a stock, that move all the way from, go back two years and it was under 100 to more than doubled since then, where they really stand out and outperform as a stock is when they’re expanding margins. They expanded their EBIT margins about 400 basis points last year.

Extraordinary growth margin expansion for a company that’s pretty mature. So I think investors really want to see them showing discipline on the cost side. I’d say almost probably more so than any of the other mega tech companies.

And so that’s what I’m going to be watching near term, any sort of signaling from management or what are the margin results look like for Google over the next couple of quarters? How are you deploying capital into the tech space overall? I know we’ve talked about some specific equities and specific themes, but in terms of asset allocation, are you more focused on tech companies with a certain business model? Are you more focused on the large caps in the NASDAQ space? Are you more interested in the small caps perhaps? What’s your preference for 2025? There’s a lot of ways I could answer that question. I would say starting with on the AI side, the very high level framework I’m thinking through right now is we’re coming off this enormous multi-year period of investment in AI infrastructure. You’ve got the four big investors in AI infrastructure, Meta, Google, Amazon, Microsoft, collectively guiding to $325 billion in CapEx this year.

Just eye-popping numbers. That’s on top of huge CapEx numbers over the last two years. We’re in year three or four of this massive AI infrastructure buildout.

Yet to date, business models or end-user facing products and services created with AI are somewhat lacking, I think you could argue. There still hasn’t been a really clear, brand new, generative AI commercial use case or a killer app, as people say. I think as we look into this year and the coming years, I’m increasingly focused on identifying what are going to be the killer apps for AI.

I’m pretty optimistic there will be many, but I think there’s still a lot of money to be made in identifying the companies that can harness all the investment that’s been made and turn it all into something that’s useful that can really move the needle for their customers. A couple of use cases that seem to have a lot of excitement around them right now, the two big ones that stand out in my mind would be, one is agentic software. Basically, software that can automate mundane or routine workflows or tasks.

I think the clear early leader there would be ServiceNow. Then the second use case, I think, is autonomous driving. I think everything Tesla’s doing with autonomous driving and Tesla’s talking about this year, their full self-driving service reaching a point where it’s at parity with human driving on a safety basis using their core safety metric.

That’s a long-winded way of saying, I think if you start to see some traction and success in agentic software and you start to see autonomous driving really prove itself out in the real world, two examples of things that are going to make investors look at this and say, wow, these investments that are being poured into artificial intelligence really do have paybacks, really are real. That can, I think, fuel a further investment cycle. I’m still very focused on AI.

I guess the only other point I’d make is, when I think about these megacap tech companies, they’re very generally well-loved, pretty well-owned, obviously, with nuances and differences across them. There’s a lot of focus on consolidation or concentration in the stock market and the idea that the stock market has been propelled by such a small number of companies. I would just argue, when you look at the fundamentals here, in some ways, that’s not surprising.

These companies are delivering well over 20% earnings growth. They’ve got great margins. They’re growing top lines, double digits.

They’re clear AI arms dealers and beneficiaries. As a group, the Mag 7, I’m going to exclude Tesla from this number, but trading at something like 28 times forward earnings versus the broad S&P 500 at 24 times. Are they trading at a premium? Yes, but it doesn’t feel like it’s a dot-com bubble type setup here, where there’s no valuation support, no fundamentals, pure speculation on technology that’s unproven.

I think that’s the last thing. Some of these megacap tech companies that have been working for my portfolio for the last couple of years, obviously, case by case, every company is different, but I’m not necessarily scared of where we sit from a valuation standpoint today for those companies. Since you brought up Tesla, what are the major opportunities and risks facing Tesla today? Tesla has no shortage of long-term opportunities here.

I think what they’re building with autonomous driving, can they license that to third-party OEMs? It sounds like they’re having conversations about doing that. Everything they’re doing on the battery technology side, will they eventually one day become a battery supplier to third-party OEMs? I think that’s definitely a possibility. What they’ve built on the energy storage side, that business is crushing it right now.

They’re guiding over 50% revenue growth in that business, already a several billion dollar business. I think that is really well positioned in a world of decarbonization and electrification. Then on top of that, the more moonshot type bets of what they’re building with RoboTax is what they’re building with this humanoid robot Optimus.

It’s not hard to dream the dream when it comes to Tesla. Now, you asked about opportunities and risks. I think a lot of the risks exist more in the short term for their core auto OEM business.

I think you’ve seen sales really slow there. They actually didn’t grow units or deliveries last year. I think you’ve heard a lot of negative checks, even shorter term, in various regions around the world, whether it’s Europe, different countries in Europe about the association with Elon Musk dampening consumer demand.

Then you’ve got potential risks of losing some of the IRA benefits, the $7,500 tax credit going away, the EV tax credit going away. I think there are a lot of short term risks to their core auto OEM business. All that makes it a really tough stock because it doesn’t really trade on a clear fundamental framework.

It doesn’t mean it can’t be a good stock. It has been a great stock over the last, call it six months or so. But it makes it a tough stock because I think there’s always going to be a mix of investors who are in it for the long… If you have 10 years to play for some of these Tesla moonshot bets, then yeah, I think it’s much easier to advocate for it as a stock than if you have 10 weeks to try to play a short term fundamental inflection.

Excellent. Appreciate your thoughts, John. We’ll speak again next time.

Where can we follow you and your work in the meantime? Yeah. Thanks a lot, David. Again, I’m a portfolio manager on our growth team at Gabelli.

I have three funds, two mutual funds in an ETF, the mutual funds being Gabelli Growth Fund, GABGX and Gabelli Growth Fund, GICPX, and then the Gabelli Growth Innovators ETF, GGRW. All information for all those funds is available on our website, Gabelli.com. Okay. Excellent.

We’ll put those links down below, so make sure to follow Gabelli Funds and John there. Appreciate it, John. Take care and we’ll speak again soon.

Sounds good, David. Take care. Thank you for watching.

Don’t forget to like and subscribe.