Prof G Markets - Google Goes All-In on the AI Arms Race

Episode Date: February 11, 2026

Ed Elson discusses Google’s massive bond sale with Gil Luria, Head of Technology Research at D.A. Davidson. They also break down why D.A. Davidson upgraded Oracle’s stock. Then, Ed is joined by Doug O’Laughlin, President of Semianalysis, to examine why memory stocks are doing so well right now. Finally, Ed gives his take on why people have started to turn on AI.  Check out our latest Prof G Markets newsletter Follow Prof G Markets on Instagram Follow Ed on Instagram, X and Substack Follow Scott on Instagram Send us your questions or comments by emailing [e-mail] Learn more about your ad choices. Visit podcastchoices.com/adchoices

Support for the show comes from Funrise. For the past seven years, there's been a room in finance most people couldn't enter, a room where you could have invested in some of the biggest names in tech companies like Airbnb and Uber before their multi-billion dollar IPOs. I'm talking about venture capital. Fundrise recently took a sledgehammer to those closed doors by launching a venture capital product that's available to anyone. Their mission is to give everyone the chance to invest in the best tech and AI companies

before they go public. You can visit Funrise.com slash ProfG to check out Funrise's venture portfolio and get in early today. All investments involve risk, including the potential loss of principle, past performance is not indicative of future results. This is a paid advertisement. Oh, hey. Sorry, love to chat, but I'm busy shopping all the rollbacks and more at Walmart. Grab a what? Cancel that.

I got to grab these big savings on the Walmart app online and in store like right now. See who? Oh, no, I'm available. The only thing I want to see are the prices just lowered on Tech Home and all my must-haps. Wait, you want to shop Walmart with me? All righty. I think I can beat you in. Today's number 47.6. That's how many hours of sleep the average American gets each week.

Today's other number is zero. That's how many hours of sleep Michael Saylor got last week. Money markets matter. If money is evil, then that building is hell. The show goes up. Watch it. Sell, sell. Welcome to Profty Markets.

I'm Ed Elson. It is February 11th. Let's check in on yesterday's Market Vitals. The S&P 500 and the NASDAQ declined on weaker than expected retail sales data for December. Financial stocks also dropped after Altruist released an AI tax planning tool. Charles Schwab fell 8% and radio. and James fell 9%. Meanwhile, the Dow notched its third record close.

Spotify soared 15% after reporting record user growth and tripling profits from a year ago. And finally, Paramount sweetened its hostile offer for Warner Brothers Discovery. It offered to pay the $2.8 billion termination fee that WBD will owe Netflix if the deal falls apart. WBD stock rose more than 2% on that news. Okay, what else is happening?

Google just executed one of the biggest corporate debt offerings in history. On Monday, the company priced its largest ever U.S. dollar bond sale, raising $20 billion across seven different maturities. That deal drew more than $100 billion in investor orders, one of the most heavily subscribed order books ever for a corporate bond sale. The next day, Google raised another $11.5 billion in sterling and Swiss francs. The sterling deal included an ultra-reliven. where $1.4 billion a hundred-year bond,

which attracted close to 10 times the amount offered in invested demand. All then, Google raised nearly $32 billion in debt in less than 24 hours. The borrowing spree came days after the company announced plans to roughly double its Cappex for 2026, and the spending surge extends across big tech, Amazon, Google, Microsoft, and Meta plan to spend a combined $660 billion in AI infrastructure in 2026. That is up 60% from 2025. Okay, here to help us break down this debt offering from Google. We're speaking with Gil Luria, Head of Technology Research at D.A. Davidson.

Gil, good to see you. Good to see you. So Google just raised nearly $32 billion worth of debt in less than 24 hours. One of the largest debt offerings in a really long time. Take us through this. Why does this matter? What does this mean for Google? There's tactical and strategic aspects to this.

So part of this is treasury management. Google has plenty of cash. They probably have 80 billion of net cash. They can cover all of their CAPEX needs with the cash flow they have from their traditional advertising business. But they're choosing to borrow money to create more capacity to, this is very low-cost borrowing for them. And it aligns them with where they need.

the cash. Sometimes you have cash in one country and you actually need another. And so a lot of this is tactical, treasury management. Some of this is strategic, though. But let's not forget that Microsoft, Amazon, Google, Open AI, Anthropic, Meta, Elon are all in this big competition to be the biggest winners in AI, which is going to be a very expensive competition. Then many of them believe is going to be winner take all or at least winner take most. They believe that because their existing markets are like that. So what they're trying to do, and Google is doing this by issuing a hundred-year bond, is to say, we're in it for the long haul.

We are going to spend as much as it takes to win. And they're not the first ones to do this. Mr. Zuckerberg, at Meta, signaled the same thing by increasing his capex continuously and by paying tens of billion dollars for talent. I'd forget. He paid $14 billion to hire Alexander Wang. So we're having this kind of signaling in the market by these really large players that are saying,

you know what? We're going to outlast everybody. You should blink before we do. Is this of concern at all the fact that, you know, we know that they're spending all of this money on CAPEX that had just exploded. We just learned in their previous earnings. They're going to spend $660 billion on AI infrastructure in 2026.

Something we've been saying a long time is, you know, that's a lot of money, but this is money that they have. Now we're seeing these gigantic debt offerings, which is basically them saying, no, now we need to borrow money to do this. Is that a concern or is this kosher? I would say that it's a concern for whoever loses. Whoever wins, all this cap-ex will have been a great investment. If there's any losers in that group that I mentioned, they're going to be stuck with a lot of infrastructure that they're going to have to sell at a discount. So that's the concern. But in terms of borrowing, especially if we're talking about Microsoft, Amazon, Google, and to the extent Apple ever gets into the game, they have so much cash flow,

so much cash on hand, that them borrowing is more a flex than anything else, right? The banks always prefer lending money, the companies that don't need to borrow. And this is a great example of that. Google doesn't need to borrow, which is why it's so easy for them to borrow a lot at a very inexpensive price. And again, signal to the market that they have a lot more capacity than just their cash flow, which, by the way, is huge anyway. The market's reaction was, you know, tepid. The stock was down a little bit after the debt offering. Very different from what we saw last week when Oracle made the same move. and you saw just a giant crash in the stock down around 9%. Is that market's telling us we think Google does have the capacity to borrow right now,

and we don't think Oracle does? Yeah, Google has the cash and the cash flow to support paying back that interest for a very long time. By the way, Google has been down the last couple of days. There's starting to be a reversion to the mean, right? Google is multiple is now in the 30s, Microsoft and Nvidia in the low 20s, in spite of the fact that the growth rates are similar, and I would argue Microsoft and Nvidia is just as well-positioned

in the AI race. So we're starting to see a version to the mean. Oracle is very different. Oracle really painted themselves into a corner. They committed to an infrastructure build-out that required them to raise a lot of capital in order to execute. They don't have excess cash flow.

They're really stretched right now. But the good news is it looks like they will be successful in the raise, which means they'll be able to deliver. And if Open AI can raise their own $100 billion by the end of the quarter, which is the most important thing in AI right now, Open AI will be able to pay for that Oracle capacity, and we can all breathe a sigh of relief for Oracle, which, again, really wasn't a little bit of a bind.

Yeah, we should mention I found this fascinating. You guys over at DA Davidson, you upgraded, Oracle to a buy, which is interesting because you're one of the people who's been kind of sounding the alarm on the position that Oracle has found themselves in over the past few months. Take us through the upgrade on the stock. Why have you regraded to buy? Yeah, so we sounded the alarm when the stock was $345, trading 45 times earnings, and Open AI looked to be losing momentum. Yeah.

What's happened since then is, first of all, Oracle stock went down to $1.45,000. 18 times earnings. And then Open AI, very importantly, a few things happened there. One is Open AI started to focus. Instead of being spread too thin, making promises they can't keep, and trying a lot of different things, they're refocused on their frontier model and on Chad GPT, which you can tell by the fact they're starting to advertise.

Second thing that happened is because Google made such a splash with Gemini and just reported really good earnings this week, that really scared Microsoft, Amazon, and Nvidia, which means they're very likely now to give Open AI that $100 billion. And then finally, Open AI is probably pretty close to introducing another model that will be another lead forward should become the state of the art again. So if open eye has a state-of-the-art model is focused on the business and can raise the capital, then it will be able to pay all those bills that we were worried about.

So that's what's changed. Oracle's stock price went down a lot. People were expecting Open AI to not be able to deliver. And because Open AI has gotten their act together, now it looks like Open AI will be able to pay those Oracle bills. And again, bail Oracle out of a very tough situation. The other thing, just before we let you go, that I want to hear from you on,

the SaaS is dead, software is dead thesis, which really played out last week, all the software stocks got absolutely crushed. We've seen a little bit of a rebound this week, not fully or not totally equally distributed across the software stocks, but as a whole, tech is rebounding, software's rebounding a little bit. Where do you stand on this debate at this point? People said last week, software is dead, AI killed it.

Where do you stand today? AI is a major disruptive force. in all of technology and the whole economy and specifically for software. What happens when you have major disruptive forces is that excellent companies and good businesses execute well and win, and companies that are not very good

and not very good markets end up falling by the wayside. That's true in software as well. But what happened the last couple of weeks is that all software stocks sold, regardless if they're in the first or the second category. And we love that because that's what creates opportunities. You know, I've been on a staff for a long time, I've always resisted putting revenue multiples on any company, including

software companies. We don't have to anymore. These companies are now trading on free cash flow, on their actual profit, and you had an opportunity last week to buy unbelievable companies like Snowflake and Datadog at 35 times cash flow, Microsoft at 20 times earnings because of this onslaught on all of software, then again ignores the fact that Datadog, Snowflake, massively big winners in software. Microsoft massively big winner in AI, but the sentiment against software was so negative that really good companies that are executing well

are well positioned for AI were selling alongside companies where the disruption may actually hurt them, even though they have some years to execute. I understand that part. But again, I love the opportunity to buy companies growing 15, 20, 30 percent on a multiple of cash flow. We've never had that, and that's a great opportunity for investors.

All right, Gil Lurier, head of technology research at D.A. Davidson. Thank you, Gil. Thank you, Ed. After the break, why memory chip prices are soaring. And for even more markets insights, you can subscribe to my weekly newsletter simply put at edwardelson. com. We're back with Profty Markets. Memory chip stocks are on a relentless tear right now.

Shares of the three industry leaders have soared in the past year. Samsung is up 200%. Micron is up 300%. S.K. Heinix is up 340%. All of the memory chip stocks are soaring, and the reason is because AI data centers are devouring memory chips leading to shortages across the rest of the tech industry. Meanwhile, shares in Qualcomm and Arm declined

after they warned that these memory constraints could cap smartphone production. Apple's quarterly earnings were clouded by the same concern. So, what are these memory chips exactly, and why are they so precious right now? Well, to answer these questions, we're speaking with Doug Loflin, president of semi-analysis. Doug, thank you for joining us on Prof2 Markets. Thanks for having me. More than happy to answer, whatever questions do you guys have?

We have plenty. So chip stocks are soaring this year, memory chip stocks, that is, Can we just start with the basics here? Like, what do memory chips actually do? So, like the word implies memory also store something for recalling later, kind of like how our memory works. Now, there's two key styles of memory, which is DRAM, which is non-persistent, meaning it works while it's plugged in.

And there's something called NAND or actually there's also hard drives as well, but for most people really care about DRAM and NAND. Now, NAND is non-persistent, meaning that you can unplug it and it will still hold your information. So when you turn off your computer, all your files are still there. That's on NAND. But when you turn on your computer, most of the time it's using faster memory, which is called DRAM. So those are the big categories of memory, and there's actually kind of this newer category called high-hbm called high bandwidth memory, but don't want to get it to in the weeds too quickly.

So why are these memory stocks tearing right now? Yeah, so I think there's a little bit of history that helps set the context. So memory is notoriously cyclical. Pretty much every year there's new memory demand comes online for more data. But when new memory supply comes online, often it comes on in chunks of 40 to 50%. So what happens is, you know, supply overshoot demand, boom, price goes down. Demand overshoot supply, boom, price goes off. We are in what is probably the most historic memory cycle of all time.

demand specifically from AI has skyrocketed and effectively, we just came out of the single worst memory cycle ever. And when you have a bad memory cycle, pretty much your desire to spend on new capital equipment or new fab, like clean room space is very low. And so pretty much no one wanted to spend because they're burning all this money.

And then boom, a giant demand vector hit the industry. And so now everyone's kind of racing to invest in supply. but there's literally no supply for the next two years. And so you have that perfect supply demand mismatch, and that's the reason why memory crisis have gone up 100%. And it's going to hurt most basic consumer products, like your phone, for example,

the price in your phone of the DRAM, even the NAND, the DRAM specifically is going to go up like 100%. And so that's kind of this crazy dynamic that's happening because of AI. AI is just demanding so much memory that it's kind of, you know, completely throwing the supply demand and balance off. In an interesting way, it seems to mirror what we've been seeing with energy prices, where you have this massive AI build-out, all of these data centers that consume huge amounts of energy, and one of the side effects of that is that energy costs go up.

What you're saying is these data centers also consume huge amounts of memory chips. We don't have enough memory chips, and therefore that might also increase the price of consumer goods. Is that right? Yeah, for sure. It will definitely increase the price with consumer goods. honestly, we think the it's kind of crazy because, you know, we need some here are 70,

I actually do like a lot of work. Maybe this is a little bit of a rabbit hole. We do a lot of work on the energy pricing side of things too. Ironically, mostly energy pricing that you're seeing in the grids, like essentially what is recognized the consumer is actually the past few years of investment. Honestly, given how much demand there is,

I think most of them energy prices so far have very little AI to do with it, but the energy prices down the pipeline. Now that's all. AI. So it's kind of interesting. You're already seeing these price increases, and this is even before the true demand vector came in. That's a complete side, kind of like a side tangent. But yeah, that's really what happened. There was essentially no supply. And I think the place that we're seeing, like our business is very focused on, like kind of seeing where AI impacts chips and semiconductors,

hence semi-analysis, right? And we think the biggest bottleneck in the market right now today is memory. And that's just because there's so much demand and there's so low. supply. And the supply response takes often 18, 24 months, and I think we're six to 12 months into that. So we have at least 12 months before more supply comes online. So some of the memory names we're talking about here, Western Digital Micron, Seagate, Sandith. Sandth is up 1,500% in the past year. Kind of unbelievable. Do you expect that this is going to continue over the next 12 months? it sounds like you think that the chips themselves will keep going up in price. Would you expect that the demand for the stocks that make the chips will also continue to go up?

Well, this is where memory stocks are kind of like a whole, you know, a whole like 40 chess altogether. Because what happens is stocks are very, very forward-looking, right? And the thing that really matters is right now, every day, the stocks have been ripping on the fact that spot or contract price effectively has been going up. Now, what's going to happen is that that's going to happen is that's going to, going to continue to happen until there is real fungible, like a real supply response. And so we just don't see it until the first half of 27. We think that memory prices will continue to go up quite meaningfully into the rest of this year.

And then in the first half of 27, we expect a meaningful amount of supply to come online. And we just don't think that cross is going to happen. Historically, what happens is when the first supply starts to come online in the stocks take. There's no other way to put it, right? going up this amount of this like you know precipitously is often not sustainable and i think everyone in the industry would tell you every everyone understands this is not a sustainable price and this this just takes capitalism to fix it right supply will react to demand but at this exact moment we're at this like crazy parabolic thing um and i think right now we see no reason that

memory prices won't continue to rip for the rest of the year and so that's kind of like the right that's how you think about it is i would expect them to take it to to do well, but maybe not the past rate they've done well, especially out of the worst cycle of all time into the best cycle of all time, that kind of inflection is usually where the stocks go crazy. How I think about it, actually, if we're talking about like from pure stock perspective, if you think about like Nvidia's giant year two years ago, right, when the stock went up like, you know, I don't, I know the numbers off my head.

And then last year, honestly, Nvidia did outperform the market meaningfully, but we're talking 35, 40%, right? meaningfully above the market, but not quite as much, I think some of that inflated expectation is going to come out. Now, that obviously doesn't mean the stocks are going down. In fact, I'm very bullish, but I don't expect the same setup going forward. All right, fascinating stuff. Doug O'Loughlin, president of semi-analysis.

Thanks for joining us, Doug. Yes, thank you for having you. Well, all this talk of AI and chips and data centers is honestly getting a little bit Exhausting. If you watch the Super Bowl, you'll have noticed that one in four of the ads that were aired featured AI in some capacity. Meanwhile, AI is being mentioned in more earnings calls than ever before. Also, Google's search interest in AI is already twice as high as it was just 18 months ago. And throughout this dialogue, a lot of people are asking a lot of different questions, such as how are we going to use AI? How will it be generated? Who will be the winners? Who will be the losers? These are some of the most important questions. of our time right now, and that's why we are covering AI so aggressively. But there is one more question that is arguably more important than any of the others, and it's a question that investors

need to start taking seriously, and that is how many Americans actually want AI? That might seem like a stupid, maybe simple question until you dig into the numbers, and you start to realize that increasingly across America, people actually don't want AI. More than 80% of Americans say today that they are concerned about AI. More than 75% say that it could pose a threat to humanity. More than half say it's going to negatively impact our ability to do things on our own. And perhaps most importantly, less than half of Americans currently have a favorable view of AI right now. Put another way, AI is broadly unpopular in America.

Now, to be fair, this happens a lot with new technologies. People find it scary, they say it's too much, they push back, and then ultimately the technology ends up being so useful that we end up using it anyway. And in all likelihood, that's probably what will happen with AI. But we should also recognize it's getting to the point where AI is so disliked that now actual obstacles are being put in place to prevent this massive AI build-out

that we keep on talking about. Most notably, we're seeing this in politics. You might remember a few months ago, I said that I thought data centers were about to become the new political football. One thing about data centers, huge costs on energy and temporary job creation through the construction,

but once it is built, actually the local community doesn't really benefit from it. It's a robot that's building shareholder value for the people who own AI stocks. And so it's becoming a really interesting, and I think this is going to become more prominent in the political sphere, but it's this NIMBY versus Yimbi debate. Well, we are now seeing this play out just last week. Ronda Santis discussed a new proposal to prevent data center construction in Florida. These hyperscale data centers use as much power just to power the data centers as a city of

half a million people. You have this much power that you're capable of generating. You have this much demand, right? And that's kind of where we are. If you double the demand, unless you also double the supply, prices are going to go up. Now DeSantis's argument is that data centers, while they might seem good for a local economy. They're actually quite bad for a local economy. Why? Because they don't really employ people, and more importantly, they send electric costs through the roof. Both of these statements, by the way, are true. You look at Open AI's new Stargate data center, for example. That's going to employ about 100 people. That is roughly a third of the number of employees that work at a standard Walmart location.

Meanwhile, electric bills are indeed going up. In areas where data centers have been built, the price of electricity has risen roughly 250% over the past five years. And this is why these anti-Data Center movements are now piling up around the country. In Michigan, protesters have filed several lawsuits against one of those Stargate data centers.

Over in Arizona, the town of Morana has elected to block a data center as well. In Virginia, more than 50 data center regulation bills have been proposed this year. And in Georgia, more makers have suggested simply banning data centers across the entire state. All around America, people are slowly but surely deciding that they actually don't like AI. They don't like the CEOs. They don't like the companies. And they especially don't like the data centers. And we can have all of these businessy conversations about AI. We can talk about energy capacity and enterprise usage and memory storage, et cetera, et cetera.

But the biggest conversation we are not having is how many people actually want this. This is what investors should be tackling. And it's also what investors should be pricing. What does America disliking AI actually do to valuations? What is it due to earnings? What is it due to cash flows? These are real questions. But so far, we seem to be treating them as footnotes.

or background information. But ultimately, the answers to those questions will be what determines the future of AI. Like any other product, AI success will be a function of how many people like it. And increasingly, that number is going down. Okay, that's it for today. This episode was produced by Claire Miller and Alison Weiss,

edited by Joel Patterson and engineered by Benjamin Spencer. Our research team is Dan Shalan, Isabella, Hinsel, Chris Nodonoghue, and Mia Silverio. Thank you for listening to Profitri Markets from Profitri Media. If you liked what you heard, give us a follow. I'm Ed Elson. I will see you tomorrow.