AI spending surge, contrarian take on tech stocks

0:09Hi, everyone. Welcome back to Investing Experts. Excited to have you on today. We're talking with Sara Awad, one of our favorites to talk to about tech from Tech Contrarians. Also wanted to highlight a webinar we're having, Top Income and AI Growth Stocks Worth Watching. It's with Stephen Kress, our VP of Quant Strategy, who's been on many times before. He'll be talking over a two-part session about six standout investing ideas, three with high-yield income, and three data-driven growth opportunities. I'm going to leave a link to register for that webinar in our show notes, so check that out if you're interested.

0:51And without further ado, please enjoy this episode where we get into many specific tech names, some of the bigger names you've come to love, and some of the smaller ones you may not be aware of yet. Here's our episode today with Tech Contrarians. Hope you enjoy it. Sara Awad from Tech Contrarians, always great to have you on Investing Experts and talk to you on Seeking Alpha. Welcome back to the show. Hey, thanks for having me. It's always great to have this conversation and touch base. Yes, and I would say there continues to be very relevant reasons to be talking tech that is truer today perhaps than ever, but certainly has continued to be true in terms of AI names driving this momentum, this stock market bullishness throughout this year, this high interest, what would you say you're most aware of and thinking about when it comes to the market right now and this tech sector that you're focused on?

1:44You've got it on the nail. It's really a market that's all about tech and it's all about AI. AI is really the main theme holding everything up. And I think that's why we're seeing so much panic about a potential AI bubble pop, because there's nothing to come in and really offset the momentum that AI has given tech in the broader market. I think what we're most aware of now is really the elephant in the room, which is, are we looking at, you know, big tech circulating a lot of money? Are we looking at open AI over commit in terms of capacity deals and AI infrastructure build out that they won't be able to actually pay for?

2:18So that's kind of what we're getting into in terms of the large question. And then under that, so many subsections fall in terms of how can we get a timeline on when we're going to see this bubble popping or potentially is this a deflation? And then in the midterm, we're going to see a bubble pop. And that's when the fundamentals really come into play. And they're very helpful because what we're seeing today is a very reactionary market. And the market's always reactionary, but we're seeing a hypersensitive market because of everything that's really hanging on the back of tech and hanging on the back of AI to be specific.

2:49So, you know, last time we spoke was at the end of September. It's, you know, not even a month and a half ago. And so much has been happening since. So what we're trying to do here is be aware of all the little moving parts, all the data points and circulate that into a thesis and see that play out and be proven or disproven by the fundamentals rather than market reaction to different headlines and AI bubble theorals or like what we saw last week with Michael Burry and the whole market melting down on that news. So if I may, I think it's best to lay the ground of all these moving parts and then I'd love to build up our logic from there in terms of how we're looking at the tech space and how we think the best way to navigate it is today.

3:30I think that's great. I was going to say, I think the big headline and one of the big fear concerns has been these valuation concerns, something you and I have talked about, we've talked about in the past on this podcast, has been this divorce between the fundamentals and what's happening in terms of market sentiment. I think to your point about Michael Burry, NVIDIA and Palantir has brought a lot of attention and questioning, even more so, I would say, because the questioning and intention were already there. So I think providing some context around that, what your thoughts are as tech contrarians, literally, I think we'd love to hear that.

4:05Yeah. So essentially, you know, what we've seen over the past couple of months is a massive build out of AI infrastructure in terms of commitments to spend. Right. So we have OpenAI committing north of one point four trillion over the past couple of months. You know, some of that is with NVIDIA, with Oracle, with AMD, with Amazon, with CoreWeave. Those are, you know, the biggest dollar value deals that we've really gotten. And all of that seems very off or out of place compared to the amount of revenue that we have speculation or idea about how much OpenAI generates. So the information reported, I think it was in September, that OpenAI for the first half of the year generated something around the lines of$4.3 billion.

4:46And that was up 16 % from the entire year of 2024. for. So that's not a huge number, especially not if you put that side by side by a 1.4 trillion commitment in spend. And at the same time, you know, OpenAI talked about this target of 13 billion for the full year. And since Sam Altman has come out and said, no, it's we're thinking about, you know, 20 billion in annualized revenue run rate for the current year. And they have plans to grow this to hundreds of billions by 2030. And, you know, if you take this, all the spending that OpenAI is doing in the sphere that people are spending, whether it's OpenAI or the tier one players, everyone's spending ahead of seeing a return on that investment.

5:28And because AI is so new and because AI is so hot, no one can afford to really pull back on spending because you don't want to not build out the capacity needed and then be caught dumbfounded next year when everyone else has actually done the spend and the demand is flowing through, but you don't have the supply to kind of see a higher return on your investment. So everyone's outspending each other. It's kind of like you have to keep the momentum going because everyone is spending, so you don't want to risk falling behind. So that's kind of the atmosphere in general. And while OpenAI is one side of that coin, you also have the fact that, you know, over the past month, we had all the tier one players report earnings, and we saw three out of four of the tier one players raise CapEx.

6:06So, you know, if we look at the numbers, Meta started the year with a CapEx expectation of 60 to 65 billion. And now they have a CapEx expectation of 70 to 72 billion. For Google, that's 75 at the beginning of the year. And now it's 91 to 93 billion. For Amazon, it was 100 billion. And now it's 125 billion. And so this is a huge move in terms of increased CapEx. And the logical assumption is that this is because they're seeing more demand and they need to increase CapEx spend, specifically when it comes to AI infrastructure. But another takeaway from the current market and the current backdrop is that Markets aren't treating all raised capex the same, right?

6:46And that's, I think, a direct impact of this OpenAI bender that we saw OpenAI go on with deals that amount to that$1.4 trillion. So, for example, we saw Meta take a huge hit post-earnings. And I think that has a lot to do with investors becoming more sensitive to the management's commentary when it comes to capex spend. You know, Zuckerberg's commentary was more about meta spending for the best case scenario. And I think that's what investors are now worried about. They're worried about all of these companies spending for a best case scenario when that might not be the case. And investors are looking for a plan B, like a contingency plan to make sure that everything will work out.

7:24And there will be some return on all this money spent if we see hiccups along the way. Because the way the market's trading now, no one can afford any hiccups. And that's why we see such a big bloodbath last week. And we're seeing that sell-off continue into this week. So that's kind of a bit of the lay of the ground. And if we take this from OpenAI, we take this from the tier one players in terms of CapEx spend, and then we look at the nit-gritty of earnings as well, we kind of get a better understanding for why people are freaking out so much. So if we look more closely at NVIDIA and the AI supply chain that is really centered around NVIDIA, We also see some red flags that I think the market and investors are interpreting as signs of a bubble pop.

8:07So we had Foxconn, also known as Honhai, deliver its October revenue earlier this month. And they actually saw a 7 % sequential growth. And while that's impressive, it's down from a 33.7 % sequential growth in September. And so that's a bit of an important data point because it means that there's something that's changed between September and October. And then if we take that with Supermicro's results and, you know, them lowering their forecast for Q1 results and then also missing on estimates at that Q1 print and seeing the stock sell off since that earnings report last Tuesday until now. And then you have CoreWeave also cutting guidance.

8:48And it's worth noting that CoreWeave's cut to guidance was actually the first time we see, sorry, not guidance, CapEx. CoreWeave cutting CapEx is actually the first time we see CapEx get cut since February of 2024. So it is a very big deal. So all of these kind of data points, if we circulate them and we put them side by side with OpenAI over committing to spend while not having necessarily the funds to back that up. Plus, you have tier one CapEx going higher and still questions about return on investment. And then you have all these smaller data points. it really does create a big picture of panic and specifically panic around this AI bubble.

9:26Where we come in is we like to look at things from a fundamental perspective to sift through what's really going to stick around and what's going to be temporary hiccup and it's going to pass. So when it comes to open AI, we think we have a big problem. Is this problem going to be addressed? Are we going to see a bubble pop because of it now? We don't think this is a conversation to have today, although the anxiety is beginning today. And the reason why we don't think it's a conversation to have today is because we're still in the very early innings of AI. And I think we have a little bit more leeway before OpenAI is going to have to, you know, be put under, you know, gun to its head, back to against the wall and need to actually figure this out.

10:03But for now, we're still seeing markets react positively when, you know, OpenAI closed a deal with Amazon post tier one player earnings. So we still see more room for this OpenAI kind of commitment and deal frenzy to stretch out. Sam Altman's played it very well so far. What we do think is more of the immediate concern is really those data points from CoreWeave and Supermicro and the supply chain guys like Foxconn, because those guys signal something that's very fundamentally important about AI server demand and a potential lull when it comes to NVIDIA now ramping its B300 and GB300 products. So what we think is that this ramp could actually have a bit more hiccups than markets had anticipated.

10:49And that's showing up in the numbers of deliveries for October, because the GB300 and the B300 are much more complicated than the B200 and the GB200. And so the supply chain needs to actually figure out how to go about it in terms of getting the right balance for liquid cooling, making sure that there's no overheating issues before they can continue to ramp and deploy these AI servers. And so I think what we're seeing now is a bit of a hiccup when it comes to that ramp. And we've seen this before with NVIDIA, specifically when we first had Blackwell hit those shelves. So it's nothing new. It's nothing out of character.

11:21It's the growing pains that come with advancing and keeping up with Moore's Law at the end of the day. But we do think that given the big picture backdrop with open AI's overspend and this, Michael Burry's shorts, markets are interpreting this as a bubble pop. We don't think that's necessarily the case yet. We think this is a bit of a bubble deflating. And then we think that conversation about a bubble pop is actually pushed out for the midterm, potentially second half of 2026. But we don't see this being a conversation that's going to permanently weigh markets down in Q4, because part of this is a temporary lull because of this product ramp, in our opinion.

11:57And that creates risks for NVIDIA's January quarter outlook, which we'll know more about next Wednesday. By the same time, it creates opportunities because if we do see a softer January quarter outlook from NVIDIA because of potential ramping issues, then that's all the more opportunity to buy NVIDIA, right? We've seen NVIDIA across the 5 trillion mark since we last spoke. And it could be very well, in our opinion, headed to 6 trillion, especially if we see some progress on the China front. This notion of earnings and also short term versus near term versus long term and what gets pushed out and what is focused on now.

12:32I'm curious, we had David Alton-Clark in August, and in July, he had written an article on Seeking Alpha, alerting similar or highlighting similar points that Michael Burry is now highlighting in terms of what you've just been talking about, AI-driven CapEx, surging. And his point was that companies are not shortening their depreciation schedules, which would potentially inflate short-term earnings, which is masking the true economic costs and what that means kind of long-term. To your point and to his point, what would you say further about kind of three, six months down the line, 12 months down the line, what earnings may look like, what those names may look like, and what market sentiment, how that may evolve along with that story?

13:24So I think that's a million dollar question, to be honest. What I would say in terms of from current visibility is that the companies that exist within the semi-supply chain that have solid fundamental exposure within this AI server buildout, this AI infrastructure buildout, whether that's GPUs, AI GPUs or AI custom ASIC, these are the guys that I see doing well in 2026. I think these guys will outperform in 2026, regardless of the bigger concerns of an AI bubble. And the reason is because AI momentum will continue, although right now we're seeing spend happen preemptively. And that's very scary for investors because we don't know at the end of the day what this AI market looks like to be exact.

14:07This is the first time we're interacting with all of this. And so I think when it comes to what markets will look like in three or six months, I think what we're getting in Q4 is a lot of the negatives priced in. And that makes me feel a lot better about the first half of 2026, because that means that the first half of 2026, we're going to get a lot more clarity on, you know, wafer fabrication spend, which impacts the semi-cap peer group, which are usually the leading indicator for semis as a whole. and then by extension, hardware and software sectors. So once we get a clearer view of that wafer fabrication spend at the start of 2026, then things are going to be set in stone because that wafer fabrication spend is going to determine how markets are moving.

14:50So for example, for memory, which has very high content in AI servers and in ASIC, memory is going to have to move from DUV tools, ASML to EUV tools. And so that's going to require a higher spend because EUV tools come with a much higher ASP. So if we see, you know, in memory transition to EUV, then we're going to see a healthier wafer fabrication spend, even if smartphone and, you know, PC don't pick up. And to me, that means that we have tailwinds that are from a fundamental perspective on AI that's still in its early innings, but the other names that kind of have this attached or, you know, market impression of AI exposure, I see those guys falling out, the more AI matures.

15:32And so I guess what I'm saying is the trajectory three or six or 12 months down the line, in my opinion, the biggest winners will still be the biggest winners that we have today. And those guys that come to mind are Broadcom, NVIDIA, Micron, Seagate, Western Digital, of course, all contingent on supply-demand dynamics, especially for the memory and storage peer group. But big picture, I still think that we'll have the same kind of winners because there are guys that dominate the tech narrative. And I do see them continuing to dominate it into 2026 on fundamental positioning. So that's where what we say in our investing group is that if you're in the right names with this exposure, red days are days to double down.

16:11They're not days to panic because you know that this company is going to come back up because of it's embedded in this AI supply chain and AI is not going anywhere, even if we see a bubble pop. And then just because you started mentioning specific names like Broadcom, which you've been on the past being bullish about. Curious your updated thoughts on some names you've brought up in the past, like Intel and Broadcom. And just if you would give a couple minutes on each of those names. Of course. So I think there's a couple of names to get to here. And the first one I'd want to start with is actually AMD, because AMD has been an interesting case for us.

16:47So we were actually bearish on AMD earlier this year, and then we upgraded the stock post Q2 earnings. I don't know if you remember, it took a massive nosedive on Q2 earnings because, you know, investors were looking for more from management in terms of AI growth tailwinds and in terms of specific AI GPU sale numbers, which we haven't ever really got from management at AMD. And so we upgraded AMD around 160 because we thought, you know, the negatives have been priced in. They have stellar CPU positioning. They're gaining share against Intel. And we should see market get excited once again about their AI positioning, because that's kind of the trend with AMD.

17:26Everyone gets very excited. The financials show up, people get more cautious, and then it's a rinse and repeat cycle. So we upgraded it then. And then we had the OpenAI and AMD deal come through for six gigawatts, which was a massive deal. And, you know, AMD soared to all-time highs. And so with that deal, we think that deal in specific is very interesting because OpenAI and AMD essentially agreed to deploy six gigawatts of power with an initial gigawatt to be deployed in the second half of 2026 on AMD's next generation MI450 series. And so you'll notice that in the press release, there's no actual dollar amount for that deal.

18:04There's a lot of speculation about the dollar amount, but there's no solid dollar amount. And that's because this isn't really your typical kind of deal. This is actually part of Lisa Su's genius because AMD has been falling behind when it comes to competition against NVIDIA on the AI front in terms of design wins. Because at the end of the day, no one from the tier one players to the NeoCloud players, no one wants to be solely reliant on NVIDIA. So everyone wants to diversify. And AMD has been trying really hard to become that second pick, that alternative player. But the reality is that when it comes to design wins, and this is something that we talked about back in 2023, actually, for AMD.

18:40When it comes to design wins, AMD is having a hard time keeping up. I mean, the MI350 is much better than the MI300, but it's still lagging behind NVIDIA's current generation of B200, GB200, and it's currently wrapping generation of B3 and GB300 offerings. So for AMD, essentially, they didn't really have a leverage to get positioned in a very strong way within the AI supply chain. And what Lisa Su did is essentially bake AMD into the future of AI by closing this deal with OpenAI. And the deal actually comes at very little cost. It benefits OpenAI much more than I think markets understand because there's no dollar amount on this.

19:26And essentially, with every gigawatt of deployment, OpenAI is going to be able to purchase stock from AMD. And so they're going to be making their money back on this investment, on this partnership. It's a very strategic partnership because it's a win-win. AMD gets baked into the future of AI and the stock as it is now trades on this future potential that it has within the supply chain, although its fundamentals continue to lag behind. And we saw that this quarter with data center sales, I mean, they grew 34 % sequentially. But on the Q &A session, we had analysts really push hard for a number in terms of a split for data center sales between AI GPU sales versus CPU sales.

20:08And we didn't get any split. We didn't get a split of AI GPU sales versus AI server CPU sales. And that was a big problem because it makes us think that our initial thesis that AMD's positioning in CPUs is much stronger than AI GPUs is correct, because management would tell us something good if they had some good number to share. So from a financial perspective, AMD is still lagging behind in terms of design wins. But because of this strategic nature of this deal, in which essentially OpenAI is going to be compensated with shares from AMD in return for choosing AMD to deploy six gigawatts of power over time, is actually a win-win because it puts AMD back into this strong position.

20:49And it also lets AMD escape with not needing to show a lot of financial outperformance until that deal happens in the second half of 2026. It also makes AMD appear as a credible, legitimate alternative to NVIDIA because it's getting into bed with OpenAI, which is essentially the face of the entire AI boom when it first began with ChatGPT. So I'm not sure if I went through that too quickly, but I think this is kind of an aspect of that OpenAI AMD deal that a lot of people glossed over. And it doesn't really show how strategic, how genius this move is from Lisa Su because it really adds an edge to AMD that wasn't there before.

21:27And it also gives AMD the option to actually be competitive on its MI450 series, which is going to be the one OpenAI is deploying next year, because that'll be the first time that AMD is actually, you know, ramping a product that directly in real time competes with the product Verirubin that NVIDIA is ramping also in the second half of 2026. So I think that there is a brighter future for AMD because of this. But the reason why I get into the nits greedy is because it's not necessarily for the reason that a lot of people think today. It's not because at the moment AMD has stellar AI positioning.

22:03It's because in the future it will have because of the strategic nature of the steel. And for the meantime, it still has very good share gain when it comes to CPUs. So that's a bit about our take on AMD. We're still bullish on AMD. We see the stock outperforming in 2026, but it's a stock we always say we don't like to hold into earnings. That was actually the title of our last article on Seeking Alpha. And the reason is because financial performance is lagging behind what market is pricing AMD on, which is the future. I was going to say that that to me encapsulates everything about the tech sector right now.

22:39It's not about what's happening right now. It's about what's happening in the future. It's the promise of AI. I wouldn't look at valuation. That's a time to get concerned about being invested in the stock. Would you say the valuation is the sole concern about AMD? And I don't think that you went too much into the weeds there because I think it's very instructive for investors to hear all these nuanced details about the stock. Anything else you would add to that, I'd be happy to hear as well. For AMD, I think that's essential for anyone investing in AMD to know that it is trading on that future potential rather than the current value that the company has.

23:22and that's where valuation gets a lot trickier because how do you value something that's trading on the future right and how do you know if you're overpaying for this how do you know if there's going to be any hiccups along the way i once again the biggest element i would say that factors into all of this is that ai is all new and i think too often we forget about that even if you look at the nit gritty about you know how many gpus it would require for one gigawatt of power and i don't know if you've heard this number before but it's an insane number in my opinion and it would take half a million GPUs for one gigawatt of power.

23:53And that's a very high number. And we're talking about deals with six gigawatts of power over time. So it's a massive buildout. And I think that investors and the market as a whole needs to resituate kind of how we're looking at this AI narrative. I think we're jumping the gun when it comes to how much spend we're going to see and how much capacity we're going to see. And that's making everything a lot murkier. But the reality today, very clearly, in our opinion, is that demand continues to outpace supply when it comes to AI servers. There's still more demand than there is supply. The issue is just about the timing.

24:27So the timing of when you can fulfill the supply for said demand and when you're going to see return on investments for the hyper demand for AI that we're seeing now. So it's all a timing issue. And that's why we think it's too early to get concerned about a bubble pop today. I think that's something more of a midterm conversation. And even today, while valuations are crazy, you know, everything's trading at an AI premium and AMD is definitely no exception. I mean, the forward EV sales for AMD is 12.32 against, you know, like 250%. I'm looking now higher than the sector average. You know, NVIDIA is also trading at premium.

25:04So is Broadcom. So all these names do have premiums. In our opinion, it's about finding, again, the names with fundamental exposure to this AI build-out that would help justify that higher valuation. So you have guys, for example, like Morvell that are kind of exposed to this ASIC trend, but don't have, in our opinion, solid fundamental exposure, especially with a competitor like Broadcom in the field that has the economics of scale on its side and also has OpenAI as a customer. It also has Meta, Google, ByteDance. So Broadcom's positioning is so much better when it comes to custom ASIC and when it comes to its exposure to networking, when it comes to AI GPUs, that in contrast, Marvell doesn't have fundamental positioning that can actually support upside, in our opinion.

25:58But it's still trading at a forward EV sales of 9.91, 180 % higher than the sector average. So valuations are high, but it's about sifting through which one can be justified down the line. Last time you were on, you were discussing this notion about Marvell being better than AMD and you were also, or sorry, Broadcom being better than Marvell. And but you were you're just touting the reasons to be interested or optimistic about AMD. But last time you were on, you were talking about how NVIDIA, much like Broadcom is to Marvell, NVIDIA is to AMD. What would you say was the main turning point into that stock?