Why AI Is NOT the Next Dot-Com Bubble with Andreas Steno and Mikkel Rosenvold | Macro Mondays

Andreas Steno and Mikkel Rosenvold are back to challenge one of the biggest narratives in markets: that the AI boom is simply another dot-com bubble waiting to burst. They break down why today's AI investment cycle may be fundamentally different from the late 1990s, and what that means for tech stocks and the broader market outlook. They also unpack the latest political surprise involving Brazil's Lula and the implications for emerging markets, before turning to renewed EU–China trade tensions and what another escalation could mean for global trade, European industry, and investors. 🏆 The Real Vision Trading is now open! The best trade ideas on the platform compete for $80,000 in cash. 📈 Trading League: https://realvision.com/your-future?utm_medium=social&utm_source=youtube&utm_campaign=2026_uyf&utm_content=macro-mondays Timestamps: 00:00 - Anthropic IPO Filing: What It Reveals About the AI Buildout 02:41 - AI vs Dot-Com: Why the Comparison Still Doesn’t Hold 05:54 - Why Higher Rates Are Forcing AI to Prove ROI Now 08:15 - Big Tech Earnings: Can Alphabet, Amazon, and Microsoft Calm Credit Markets? 09:32 - GPU Rental Prices Are Rising — and That Changes the AI Bear Case 12:33 - Is Anthropic Running Out of Compute? 15:22 - Anthropic IPO: Could It Pull Capital Away From the Mag 7? 16:42 - Where to Position If the Fed Finally Acknowledges Tight Financial Conditions 18:33 - Has the Market Already Priced in the End of the AI Boom? 19:19 - Brazil Election Shock: Lula Wins and Markets Get Hit 22:08 - Israel Election: Why Policy Toward Iran May Not Change Much 23:50 - EU vs China: Europe Escalates the Trade War Learn more about your ad choices. Visit podcastchoices.com/adchoices

• 0:06Speaker 00

  Hello out there, welcome to Real Vision, welcome to Macro Mondays. My name is Mikael Rosenvold, I'm joined by Andreas Deneau. And today we have a great show for you, we're going to be covering the AI slaughter, the election shocker in Brazil, and maybe a little bit about the EU trade war with China. And oh, wow, how about that, Anthropic IPO, Andreas. We got the initial, what's it called, the S1 filing today, so not a full prospectus. What's your take on that?

• 0:35Speaker 01

  So I don't think, I don't think we got a lot of news today that wasn't already covered in the Wall Street Journal article last week, but, you know, from a top-down perspective, I have a couple of conclusions looking at that prospectus. First of all, Anthropic is very explicit when it comes to its compute capacity going forward. They need more, and they are very well aware that they don't have enough. That's, you know, a story I've been banging the drum on for at least a quarter, since we've seen Anthropic losing momentum since June, mostly a capacity issue, if you ask me. That's part number one. The second part is that we got some details on where they're going to buy this compute, and they have forward deals in place with both Alphabet and Amazon.

• 1:18Speaker 01

  So I think it will be very interesting to see how the MAC7 executives, both Alphabet and Amazon, report this week, how they address this whole compute capacity question, because, you know, currently, there is clearly a lack of compute, but the market is not willing to finance that compute capacity. So how do we solve that equation? That's going to be the key question for this week.

• 1:40Speaker 00

  Yeah, absolutely, Andreas. I have this feeling that markets are looking for someone to make money on AI. Do you think Anthropic will go into an IPO and announce that they are now cash flow positive? We heard some rumors that they are already in Q3.

• 1:55Speaker 01

  They're very close to, at least, you know, from what we could gather from the filing, we have the data of it, obviously, up until the end of the first half of the year. And they, they've basically gone from a margin of minus 255% a year ago to a margin of minus 5, something like that. So, you know, if you extrapolate that into the third quarter, we're very, very close to cash flow neutral, or at least, you know, around the neck of the woods of a cash flow positive reading. So I, it wouldn't, I wouldn't be overly surprised if they turn cash flow positive this year.

• 2:35Speaker 01

  And, you know, it just reminds me, that I need to say this out loud again, maybe we can show the table on page four that I've posted on all social media outlets today, that the comparison between the AI buildout and the dot com is very, very weak on a range of parameters, but first and foremost, on everything from utilization of the stuff that we're building, to the price of the stuff that we're building, to the revenue of those building it. I mean, on all three parameters, the AI buildout this year, you know, we basically have full capacity, or a lack of capacity relative to demand.

• 3:16Speaker 01

  The price of that capacity is going up, which is in sharp contrast to what happened with the price of bandwidth back in dot com. And Anthropic and OpenAI actually generate quite substantial revenues. You could argue that it's not enough yet, of course, but in contrast to pets.com and what have you from the dot com era, it's a whole different story. So I think it's a very, very weak analogy, to be honest, but the credit market is currently stuck in a very dot com-like vibe. And I think it's up to the executives of Alphabet and Amazon and some of the other big players to change that. So they have some very, very important earnings calls coming up this week.

• 3:59Speaker 00

  Absolutely. Before we get back to these earnings calls, Andreas, remember to all of you listening here that the Real Vision Trading League is now open. You can sign up for free, you can post your top three trade ideas, and you enter a competition for prices upwards of $100,000. So head on over to realvision.com/your-future to find out more. And as I said, you can sign up for free, you got loads of great content, including our very new trade agent AI service, that's really, really neat if you're looking to get an edge on the competitors there. If you do choose to sign up, you should tune into your show, Andreas, Macro Meets Monday on, sorry, Macro Meets Micro on Thursday at 3:00 PM Eastern.

• 4:44Speaker 00

  It's for Real Vision Pro members, and it's, it's your show this Thursday, Andreas. What can people expect there?

• 4:52Speaker 01

  Well, they can expect 60 slides on AI versus dot com. You know, I'm going to spend-

• 4:58Speaker 00

  In an hour at most.

• 4:59Speaker 01

  Yeah, yeah, I've spent two or three days comparing the two scenarios, and I, yeah, you kind of got the preview there that I don't buy it at all. I don't buy it at all, and I think the market is wrong.

• 5:14Speaker 00

  Absolutely, Andreas.

• 5:16Speaker 00

  Let's just stick a little bit with the, or return to the story, Andreas. It seems to me that we're at still probably an early stage in this AI buildout, not at the absolute beginning, I get that, but we're at a relatively early stage. And suddenly, during 2026, perhaps due to the rate hike, investors are now asking the AI buildout to deliver results now, rather than in three or four years. Is that a fair assessment? And how do you even break that cycle? Because it seems to be, to be a very nasty cycle that we're in with these expectations of profit now on something that's supposed to be a decade-long buildout.

• 5:54Speaker 01

  I think it's very natural that there's an increased focus on getting return on investment here and now, when bond yields are going up. It's typically what we see in a hiking cycle, that there's a lot of focus on the quality of earnings, and there is a lack of patience for projects with, you know, whatever forecasts of revenue in five, 10 years' time. And let me say this, the bond market is always right in the sense that it's not because the bond market is smart, but you cannot really argue with the bond market, can you? I mean, if the bond market finds AI to be a credit risk, then AI is a credit risk until you convince the bond market that it isn't.

• 6:40Speaker 01

  Because if you need capital to build out AI, you cannot just stand there and argue, say, okay, hey, you guys are wrong, then you won't get your capital, right? So in that sense, it is a serious risk to the AI buildout that the bond market is currently not trusting it. And that's why I continue to say that the executives of the MAC7s have a very, very big task ahead of them this week, because they need to convince the bond market, because they also need capital for the buildout. Remember that when Anthropic tells the market that they are searching for more compute capacity, it basically means that Amazon and Alphabet needs to deliver it.

• 7:20Speaker 01

  And they are currently closing in on a cash flow neutral state, right? They, they struggle to finance everything themselves. So we've seen Amazon issuing bonds, we've also seen Amazon tapping the, sorry, Alphabet tapping the capital markets over the course of the past few quarters. So they cannot just use their free cash flow anymore, they need to convince the market that this is a good idea. The best way to convince a market is to present some very convincing returns on investment of the already invested CapEx dollars, right? And I think they'll deliver very strong results. So as I've said, it's not a major issue to have credit stress as long as earnings deliver,

• 8:04Speaker 01

  but if earnings do not deliver, then these, pardon my French, glass half-empty types in the bond market, they, they will get even more concerned.

• 8:15Speaker 00

  So, so that's already relevant this week, Andreas, because we have, I believe it's Microsoft, Alphabet and Amazon with earnings reports this week, right? I'm a little feared that what happened to Meta will happen to them as well, you know? From what I read, whatever I could count, a decent report by Meta, but the markets just hates CapEx at the moment.

• 8:38Speaker 01

  Yeah, and they, you know, Meta basically decided to come up with a new equity category to fund the CapEx. So it was some sort of light dilution of existing equity holders. So in that sense, I get why the stock took a beating. I would be surprised to see a similar move from Alphabet and Amazon and Microsoft. The big question is whether, you know, because last quarter, they basically said that they, most of them said that they expected to use more third-party data providers, so some of the new clouds, for example, but they have an even bigger issue in terms of accessing capital right now.

• 9:23Speaker 01

  So I think the best thing they can do is to be very specific on the return on investment, not vague promises, very specific, how much do we get in return for the CapEx dollars that we've already spent? And as far as I can judge, I mean, Mikael, maybe we can bring up the chart on the rental price of various GPUs. I think it's on page six.

• 9:46Speaker 00

  Yeah.

• 9:47Speaker 01

  Because this is probably the most bizarre chart that I can think of right now, because it is so at odds with the consensus. If you look at the spreadsheets from all of the investment banks and from all of the credit managers around the globe, I have yet to see a spreadsheet that counts in rising GPU rental costs or rental prices.

• 10:13Speaker 01

  Everyone and their mother assumes that GPUs will drop in price, which is a natural conclusion, given that we get new generations of GPUs, better equipment, basically, right? But look at, for example, the H100, it's, it's an old series of GPUs from Nvidia, mostly installed back in '22 and '23. They're up 30% in rental prices this year. Look at the P200, the first Blackwell series, right? It's, it's more than doubled this year.

• 10:47Speaker 01

  That is so at odds with all of the assumptions made by the credit market, by the equity investors, that I struggle to put it in words, because everyone assumes the exact opposite trajectory.

• 10:59Speaker 01

  How on earth can this happen? Well, this is basically the symptom of a market that is short on compute capacity, because otherwise we wouldn't see this, right? So remember this, all of the CapEx into GPUs and memory chips, et cetera, is designed with a roughly five-year depreciation profile. And when the rental price of an H200 or an B200 increases, it basically means that it outpaces the depreciation profile by miles. And that is now the case on every single GPU installed since 2020.

• 11:41Speaker 01

  I don't think you can find a GPU anywhere on earth that has not beaten its own depreciation profile by miles.

• 11:49Speaker 01

  That's bizarre, because it's the exact opposite of what everyone expects. So if this continues, the return on investment is absolutely fine.

• 11:58Speaker 00

  This is incredible, Andreas. This also shows, I know you cannot translate this to PNL, but it shows you that investments in these GPUs pay out over at least six years from the 2020. I know it's not certain what will happen in the coming six years. My question to you, Andreas, these prices, and we believe these are going to continue. We know they're still massive bottlenecks.

• 12:24Speaker 00

  Will they hamper the growth of Anthropic and OpenAI, simply due to a lack of supply of compute?

• 12:33Speaker 01

  So yeah, you're right, that you could argue that this is an issue for OpenAI and Anthropic, right? Because this is basically their cost base. So they need to be able to pass this on to you and I and other users, which they've to some extent struggled with. But let me remind you again about the numbers we got from Anthropic in the prospectus or in the filing. I think revenue has 15xed in a year, while compute costs have 4xed. So it actually tells you that they're able to pass on the costs and more to the average client. They haven't been able to pass on that cost to all of the API clients,

• 13:14Speaker 01

  but they probably manage overall to pass on the costs in a sufficient way. Otherwise, you couldn't go up 15x with compute only going up 4x.

• 13:26Speaker 01

  The issue for Anthropic is that they're running out of compute now. You know, I think it's been visible, and I think it's the reason why they've lost momentum from an enterprise perspective.

• 13:36Speaker 01

  And therefore, maybe that's a symptom of, you know, prices being so high now that it's difficult to pass on the costs. But then you'll get an adjustment in this chart as well, right? So I, by the end of the day, right now, I'd argue that in all parts of the supply chain, you know, prices are up on chips, the rental price on chips are up, and Anthropic, to some extent, OpenAI, mostly managed to pass that on,

• 14:04Speaker 01

  which is amazing. It tells you that the market is not full yet, if you know what I mean.

• 14:10Speaker 00

  No, no, that's the point. And even if we are upon whether Anthropic are close to being cash flow, I mean, if compute prices drop, that's obviously an issue for their providers, but that would only trigger the Gevon paradox that we talked a lot about last year. If the cost of compute falls, that could even speed up the AI adoption. So, I mean,

• 14:32Speaker 01

  At one point, if, I'd argue that the price of compute is already massively down. If you look at how a consumer price index or an inflation basket is typically calculated, one thing is the nominal price you pay for an iPhone or pay for a token, but it will always be adjusted by the power of that iPhone and the power of that token. And the token is so much more valuable today than it was 12 months ago, because the model is so much better. So you're paying in turn, more or less the same for more. So if you look at it from the standard of which the Bureau of Labor Statistics and other statistical bureaus measure inflation,

• 15:17Speaker 01

  I'd argue that the price of a token is down 95% or so over the past two years.

• 15:21Speaker 02

  So a quick break in your regular programming. If you're serious about your future, grab my free report called Prepare for 2030. I think you've got five years to make as much money as possible, and this guide will help you navigate what's coming. The link is in the description. Download it now.

• 15:38Speaker 00

  Fair enough, which is leading to the expansion. So still a lot of fundamentals pointing in absurdly the right direction, but markets are not following up. Where do you position? There's also taking into consideration the upcoming Anthropic IPO, which even though it's going to be a low flow, maybe 5%, still going to, you know, suck in a lot of capital from index funds, et cetera. Where do you position in this AI race? I know you'll dive much more into this on Thursday, but

• 16:09Speaker 01

  Yeah, remember, we had some volatility around the SpaceX IPO as well, right? Because you see these moves in advance of the actual IPO date. A lot of hedge funds try to sort of arbitrage that away. So I think it's natural to expect some some light repositioning in the Max 7s as a consequence of this. Max 7s down to, you know, adjust for the Anthropic inclusion over time.

• 16:37Speaker 01

  It won't be as bad as with SpaceX, because as far as I'm concerned, they're not fast-tracked into the Nasdaq as SpaceX was.

• 16:45Speaker 01

  I'll have to check up on that, but it's not my impression. So in that sense, sure, I mean, you could argue that it's a slightly negative story for the Max 7s and some of the big

• 16:57Speaker 01

  players in the Nasdaq and S and P 500 indices. But other than that, I'd like to show page 10, Michael, to give you an idea of

• 17:07Speaker 01

  where to position. I mean, in the current environment, right? Which is slightly at odds with what we've seen over the past month. You basically only have investments into data centers and the power grid and everything power-related, right? Everything else suffers basically from a lack of capital, but also from the current level of bond yields. So when when the Federal Reserve is not willing to say that interest rates are restrictive, I think they're kind of blindfolded by the whole AI build-out, because everything else than AI is down in terms of construction activity, which is typically a gauge of

• 17:50Speaker 01

  sort of the interest rate level and how it impacts the economy, right? So

• 17:56Speaker 01

  if, and I say if the Fed starts to acknowledge these tighter financial conditions, which is also something we'll get a first view on this week, I actually think that the broad equity market will rebound like crazy, because, you know, beneath the hood, I know we've seen a rotation towards the middle and the end of October, but beneath the hood, a lot of equity suffer. It's not only AI, because the economy cannot cope with interest rates at these levels. So I think we're basically one message away from a short squeeze in the market now.

• 18:36Speaker 01

  If the Fed provides just, you know, a tiny bit of hope that maybe this hiking cycle will not go on for long, then the market is ripe for a correction upwards.

• 18:49Speaker 01

  On top of that, if you look at, you know, I have various gauges of this, and I'll go into detail on Thursday. I'd argue that the market has basically priced now that the AI build-out is over.

• 19:01Speaker 01

  So if that's not true,

• 19:04Speaker 01

  then everything AI-related needs to rally like crazy. So it's basically either or now. We're at crossroads, right? Either the AI build-out ends or it doesn't. And if it doesn't, the market rally that we'll see in such an environment is absolutely bizarre, because the assumption now is that chips growth will basically slow to zero next year, probably also negative into the second quarter, which I find highly, highly, highly debatable.

• 19:35Speaker 00

  Interesting, Andreas. Let's just go a little bit beyond the US economy here and talk about what happened with a couple of the ongoing elections. We'll touch more upon the midterms next week, but the US midterms. But we had a shocker last night in Brazil, Andreas. I'm doing other podcasts and I've been tempted to, when Flavio Bolsonaro won the first round in Brazil, and he was very much the odds-on favorite, to win the second round. I was very, very tempted to tell people to go long Brazil in anticipation of Bolsonaro winning. But again, I think, ah, it's probably priced in and you have that slight 5% risk that everything falls apart and Lula wins. And then, lo and behold, what happened? Lula, after losing the first

• 20:19Speaker 00

  round, or not winning the first round, he bought himself a smartphone, he started using social media, and boom, he won the second round of the presidency. At least for now. So Brazilian equities are throwing up and everything looks horribly. Do you just stay out of Brazil at this point, or what's your take, Andreas?

• 20:40Speaker 01

  Well, I guess it's too late not to have that view. It's been very, very nasty in Brazil from the morning in US hours. I think the real is down 10% of the currency and the equity market is down. Basically, it came to a halt because it was down so much from the get-go of the trading. So I think a lot of people, thankfully not you included, Mikel, told people to buy Brazilian stocks ahead of this. I'm curious to hear your take on this. I don't think this is over. Especially since, you know, Bolsonaro is not willing

• 21:23Speaker 01

  to accept the result, first of all. That's part number one. Part number two, you know, this, you could call it a wave towards the right-hand side of the political spectrum in South America, has been very much cheered upon in the White House. So could Trump be tempted to get involved here? Or could he decide to throw extra tariffs on Brazil or whatever? I mean, I think we're all kind of waiting for a response from Trump to this. And he's got plenty of things on his plate right now, but I wouldn't rule out that he gets involved here.

• 22:03Speaker 00

  No, no. I think, no, no, Lula is a very good boogeyman, just like Vladimir Zelensky is. So it might even be useful in the run-up to elections, especially if this fight over the election result takes to the street in Brazil, which is the worst case. So yeah, I'm staying out. We have a perhaps slightly less volatile setup in Israel. I just wanted to get across to that, because we're getting a lot of questions on that. We are expecting a hung parliament. Benjamin Netanyahu will have a hard time creating a majority, but so will Isenkot, Gadi Isenkot, the former general and his party will have a hard time

• 22:45Speaker 00

  as well. So the base case here is a hung parliament, where the result would either be a coalition government between Netanyahu and Isenkot. That's looking really tricky. There's been some very, very harsh words in the run-up to the elections. Or that Gadi Isenkot will need to reach out to the Arab parties, very, very interestingly, in the Israeli parliament. So potentially a hung setup. I don't expect, and that's what I think a lot of people get wrong, even if Isenkot is able to form a government, we shouldn't expect a flip-flop or change of course in the regional policies. We might

• 23:26Speaker 00

  see some changes in the approach to the West Bank and the Gaza Strip, but nothing major. This is not a complete shift in Israeli foreign policy, probably not a big shift in the policy towards Iran. Isenkot has very much been behind the strategy to hammer civilian infrastructure as well, as part of the wars going on down there. So smaller changes to be expected. This is not a case of Gadi Isenkot wins and then the Iran war is over. Far from it, to be brutally honest. And then, Andreas, just a final note, we got in earlier today, that the EU is now ramping up its trade war with China. And we've been discussing this a lot, especially within Europe, but it now seems that the EU is essentially copying Trump's

• 24:11Speaker 00

  playbook of putting import caps on chemicals, on machinery, on furniture even. So do you think this is a new cause for the EU? And how will this spill over into markets, you think?

• 24:30Speaker 01

  I've personally impatiently waited for a message like this from the European Commission. We've seen how the European market has been flooded with Chinese EVs, not least, especially over the past couple of years. And interestingly, in most countries, you've then also decided to make these EVs tax-free versus some of the locally-produced EVs. And it just felt like you gave China an unfair advantage in many ways. I still think it's the case, Mikel,

• 25:10Speaker 01

  that you can get subsidized airfreight from China to Denmark in case you buy something online, because China is defined as an emerging market in Danish airfreight law, which is absolutely bizarre, because then it's basically, it costs almost nothing to transport some weird stuff from China to Denmark. Now they're, you know, trying to set up some roadblocks, and then China has obviously moved some of their warehouses to within the union to try and avoid those. But in any case, it's about time that the EU enters this arena, because

• 25:51Speaker 01

  everyone's in the arena already. And the EU, in my opinion, has basically spent the last three or four years trying to act as a referee, telling everyone else, oh, you cannot do that, you cannot do that, you cannot do that, you cannot do that, while no one's been listening, right? So it was about time. That's the way I would put it. And I think it signals kind of a wave of change within the European Union as well. I think the European Union is slow due to its, you know, lack of uniformity, basically. But when it moves, it can move. And I actually think that the EU can agree on something here.

• 26:28Speaker 00

  Absolutely. Okay, Andreas, I just want to show the listeners on YouTube, at least, the viewers on YouTube, a final thing, because tonight, we're hosting a show where we go through the leaders of the current Real Vision Trading League. And I just wanted to show you the leaderboard as it looks right now, just to give people an idea of what's happening here. So we have all the contributors, the big community at Real Vision chipping in with trade ideas. And in the show later today, we're going to go through them. There are some very, very creative, ingenious trades in here, and we're going to go through what you and Jamie and the other contributors likes and dislikes. I just want to show Paul here, who's leading. It was a bet on Athena. I didn't have that one in my deck, Andreas.

• 27:13Speaker 00

  And we have a guy called Jamie here with a bet on Seacast and Cypherpunk. So a couple of really, really creative bets. This is Real Vision, so obviously a lot of great crypto bets, but also some very, very interesting stocks and other stuff. And we're going to be diving through that in our show tonight, which is out on all our channels, YouTube, I believe on X, and then obviously on realvision.com as well. So be sure to check in on that. If you want to join the trade league and join in the contest for upwards of 100K in prices, check out realvision.com/eurofuture and sign up for free to take part in this great little contest. That's all we had for you this week. Remember to tune in for Andreas' Macro Meets Micro on Thursday as well. And thanks to you for joining, Andreas.

• 27:59Speaker 01

  Thank you.

• 28:00Speaker 00

  Thank you. And thanks for joining out there. We'll be back next week.

• 28:03Speaker 02

  You obviously enjoyed the episode because you're here with me at the end. But listen, don't forget to go to realvision.com/join and grab a free membership. It's an incredible community packed with alpha, great investment ideas, and the research that you need to help you unfuck your future. So get started now. Go to realvision.com/join.