Is the Risk-On Trade Breaking Down? | Macro Mondays: September 14, 2026

0:06Mikkel Rosenvold:Hello there, welcome to Real Vision, welcome to Macro Mondays. My name is Mikkel Rosenvold, I'm joined as usual by Andreas Steno. And I have to say Andreas, this has been quite a weekend and quite a Monday. I have a feeling everything is coming tumbling down right now. Oil is at 1.20, the AI train seems to be stopping we're entering hiking season, summer is ending, the weather is horrible Andreas, how fucked are we right now?

0:37Andreas Steno:Well, I actually turned pretty upbeat Friday afternoon when I saw a pretty muted reaction to that CPI print, which was admittedly a bit hotter than what we expected and then Dario fucked up my weekend, Dario Mure of of Anthropic to the extent where, you know, he basically wrote that the internet could end in six to 12 months from now. And I see that some of the, you know, usual suspects within the software industry, they're responding to this straight out of the gate because we've received a price increase notification from Bloomberg already today, you know, in box, you've seen that as well. So it's like they've already responded with a lot more confidence after Dario asked to slow down the train in the AI space over the weekend.

1:23Andreas Steno:But yeah, let's get back to it. But I don't think we're as fucked as you think.

1:28Mikkel Rosenvold:That's good, Jandras. We'll try and unpack as much as we can within the time space we have here. We're obviously unpacking everything in Real Vision, trying to cover all the angles, everything that's going wrong right now and what your opportunities are. This Wednesday, for instance, we're doing an Ask Me Anything, you and I, Jandras, right after the FMC meeting. So that's at 3.30 p.m. Eastern. That's for pro members only at Real Vision. So if you want the live reaction to that, you can tune into that. Otherwise, I suggest having a look at our State of the Union show that you did last week, Andres.

2:01Mikkel Rosenvold:Really great run through of what our models are telling us right now. We'll touch upon some of it, but for the deep dive, you should check out our pro content at Real Vision to get the full coverage and to keep you ahead of the curve. So, Andreas, where do you want to start? Dario, inflation, oil, which disaster to look at first?

2:25Andreas Steno:Dario. Dario. I mean, when he's telling us that the internet will end in six to 12 months, potentially mankind.

2:33Mikkel Rosenvold:That is probably the big story, Andreas. Okay, so we had this tweet from a disgruntled ex-anthropic employee last week that really made the rounds, we have to admit, probably setting him up for a great Substack channel or podcast or whatever you want to do, hedge fund maybe. And then it seemed like Dario Amadei, I don't know if he sort of jumped on this bandwagon, he called this, or he felt that he needed to cover this issue ahead of an upcoming IPO. We can try and guess at his motives here. In any case, he suggested slowing the pace, pacing the development of AI. He received backing from Sam Altman and Elon Musk, not from Donald Trump, we can get back to that, not at all.

3:13Mikkel Rosenvold:So how did you read this? And is the market reaction fair, in your opinion?

3:18Andreas Steno:I think the market reaction is pretty vanilla. So by telling the market that at least he's in favor of slowing things down, he's obviously also trying to tell the market that the hardware trade will slow down. And some of those companies that have been on the receiving end of all of the AI beatings, if you know what I mean, all the software as a service names and all of those names, they've recovered a little bit of ground today. So, I mean, it is a straightforward market response to this. The question is whether the timeline and the risk scenario that he puts forward and he lays out, whether it's something to really worry about.

4:03Andreas Steno:And I did this simple study since 24 of all of the big quotes and big timelines by the AI bosses, right? Sam Altman, Dario Amadei, Mark Zuckerberg, et cetera, right? And I actually don't think any of those big promises with very tight deadlines and very tight timelines have come into fruition. I mean, just take an example. In May 25, he said that we would have a white collar blot bath and half of entry level jobs would be gone in 12 months. and unemployment could go to 10 to 20%. I think he said within one to three years, maybe one to four years, but at least we're now one and a half years into it and nothing is happening, right?

4:51Andreas Steno:That's point number one. He also stated in March 25 that AI would write 95 % of all code within three to six months. We're getting there, but clearly later than what he said here. So sure. I mean, is there a risk that we could see rogue swarms hijacking the internet within 6 to 12 months? I personally don't care a lot about that timeline, given how he's guided us in other instances. In terms of his motives, a couple of things. I've seen a lot of theories floated on the internet. And I personally just asked Claude to respond to his obvious AI slot, because if you read his blog, it is very, very clearly written by AI.

5:38Andreas Steno:And everyone does that. So it's not like I'm trying to paint another story. But having said that, Miguel, let me put it like this. So he's asking for a slower pace of the whole model build out for safety reasons. My first initial reaction was, okay, this is a hedge against the midterms. The Democrats have clearly been talking about this. so in in light of that i think it makes sense that both sam altman elon musk and dario amade of anthropic that they all hitch this risk uh in unison since it could be a real risk that the democrats would would actually like to see something like this if they gain back power right um not that it necessarily matters a whole lot if they win the midterms but you know it's It's probably a sign that we're getting there again, that the Democrats will, directionally speaking, get back in power.

6:45Andreas Steno:That could be one theory, which I personally have a lot of sympathy for. The other thing that I initially noted was that, okay, he's asking for a slow pace three, four weeks after his biggest competitor just outpaced him. with the release of the Astra model. At least on all benchmarks, Claude is now second in line. But that doesn't really answer the question why Sam Altman said that he did agree to it. So the third thing, which is more worrisome if you're into the long hardware trade, and we can maybe get the hardware versus software trade on the screens here, is that maybe they're a little scared of how costly this race is.

7:32Andreas Steno:And I think that would be the most bearish of all of these explanations. It just, you know, at least for now, I think the business momentum, the underlying business momentum in each of the three businesses that we've touched upon, SpaceX, Anthropik and OpenAI, I think they pretty much back up their pipeline of compute liabilities. you know if Anthropic can 3x their revenue from here remember they've gone from 0 to 70 billion dollars in annual recurring revenue if they can get to 200 within a few years I think they're decently ahead of their compute curve in terms of revenues versus costs same for open AI so I'm personally not really worried about it but it could be that they're trying to hedge that risk, right?

8:26Andreas Steno:But I personally hold most sympathy for this theory that they're trying to hedge their political risk here because they ultimately want the U.S. Treasury or the administration in the U.S. to underpin this case, right? To underpin or to hedge the left-tail risk of their business because they know that there is a left-tail risk, that everything falls apart and maybe the open bottles will gain too much ground and all of that. So I think that is what it's all about. They're trying to remove the left-tail risk. And if that is the case, that they're just trying to incentivize politicians to help them hedge against that left-tail risk, then this is actually bullish.

9:12Andreas Steno:But again, we're basically guessing here. Yeah.

9:18Mikkel Rosenvold:No, but I agree with you, Andreas, because look at Donald Trump's response very, very quickly. very, very falsely, completely denies getting involved in any of this. This might change when a new administration comes into place, but right now no appetite for AI regulations within the US government. I think Derry Amadei knows this or would have known this. You can see, as we all can, that this is a new nuclear race against China, so there's no stopping. But that also means if it becomes too expensive to borrow for CapEx, if the IPOs happen, okay, you get some money. But eventually, if AI and the hyperscalers are struggling to find cash for CapEx, does the government swoop in?

10:00Mikkel Rosenvold:That's the big question that you're also hinting a little bit at here because Donald Trump has essentially made this national security issue that he's not going to regulate this because they're afraid of falling behind China. That logically also means that if CapEx slows down because of a lack of cash, essentially, the government may step in to support this, which removes a big part of the left tail, I should say, for the trade here. Let's talk a little bit about the hardware case interest. Let's try and work a little bit about the logic here, because if we did get, because that must be what the market is assuming or trying to price in a risk of an actual slowing down of sort of the tip of the spear in AI development, why does that hit the hardware trade?

10:45Mikkel Rosenvold:I'm not sure I get that, because this doesn't affect regular companies adopting AI. This doesn't affect your CRM adoption. What might it be of AI? Probably it opens up more compute for businesses like ours, Andreas. We've been struggling with that, to be honest. So how does that affect the hardware and software traits, in your opinion?

11:07Andreas Steno:It depends a little bit on both the time horizon and how you look at it. So to begin with, I agree with you that slowing down the frontier does not necessarily mean that you slow down the model deployment. and as long as you deploy the model there is obviously a need for compute and the question is whether there whether there's a need for next gen GPUs including high bandwidth memory and all of that so all of the hyperscalers into next year and the year after buying Vera Rubin and the new generation of GPUs with loads and loads and loads of high bandwidth memory chips in combination with that, does it risk slowing that down?

11:59Andreas Steno:Do you really need the state-of-the-art equipment if you're no longer running as fast towards developing the frontier even further? That's questionable, in my opinion. You could argue that it's bad for NVIDIA and ultimately those supplying NVIDIA indirectly, for example, in the memory space. So I think it is fair that you at least need to reassess that outcome space for NVIDIA and the hardware names. But you're absolutely right that it doesn't overall lead to a slowdown of the compute need. It leads to a slowdown potentially, I'd say, in the state-of-the-art equipment that you're going to buy next year and next.

12:48Mikkel Rosenvold:Yeah, fair point, Andres. Let's look a little bit more on the IPO angle and maybe also on the strategic angle here because these companies that are now backing a call for reducing the pace of the frontier development here, they're also the companies who are perhaps trying to build moats as they enter IPO territory and as they're trying to keep out Chinese competitors, keep out potential competitors. We don't know if their mistrawl in Europe suddenly finds out something. But I mean, Andreas, Europe often gets a lot of heat for having too much regulation on businesses. A lot of that regulation comes from the businesses themselves.

13:30Mikkel Rosenvold:Is this another instance and perhaps the first steps in this? Because this smells to me a lot like European banks going to the EU asking for stricter controls in the financial sector, which essentially builds a mode for the established businesses because it's so costly to even enter the space.

13:48Andreas Steno:Yeah, I think this is what is called a barrier of entry in the Portis analysis of forces, right? Sure. And I think it kind of makes sense for them to try and navigate the market ahead that way. And therefore, again, as I stated initially, I think the motive is mainly political, both as a hedge against the Democrats, but also to try and get the Trump administration a little bit more involved in this because they're essentially a little bit hands-off while they're still trying to accelerate the race, if you know what I mean. And I think the assessment from David Sachs, Donald Trump, et cetera, is that by having your hands off, you're getting to full speed, right?

14:33Andreas Steno:Which has so far been a decent assumption, in my opinion. So sure, it kind of makes sense that this is all a political hedge by the end of the day by Dario. Otherwise, I don't see how Musk, Altman, and Amadei could agree on a message if it wasn't for a political reason. Because they typically just hated each other, right? So I actually think it's the first time they've ever agreed on it.

15:03Mikkel Rosenvold:More or less. It is just true, Andreas. 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. OK, Andreas, we could talk much more in depth about this, and I'm sure we will in the coming weeks and months. But we have to look a bit ahead for Wednesday as well, Andreas. markets are increasingly settling with the fact that we are probably going to get a hike on Wednesday.

15:40Mikkel Rosenvold:Let's just begin there. You expect that as well?

15:42Andreas Steno:Yeah, I think it's difficult not to. I mean, a lot of the fence sitters, as I typically like to label them within the FOMC, they've put out pretty clear reaction functions. So if core inflation printed above 0.2, maybe 0.25, then we would get a hike, and it did. So, you know, in all honesty, we're a little off on that.

16:07Andreas Steno:I'm still not worried about the core inflation outlook when you look at the observations that we're able to make. I mean, month on month, it's always difficult to get 100 % right. But sure, headline inflation is nasty. And, you know, it's nasty what we see in oil products right now. Now, just as we speak here, Michael, Trump is out saying on Truth Social that the diesel price is not caused by the Iran war. He's trying to put the focus on Russia versus Ukraine now. He's telling Russia and Ukraine to stop hitting each other's energy targets, right? Yeah, I mean, as far as I remember, Russia typically exports roughly 10 % of global diesel exports, right?

16:51Andreas Steno:So they've typically had a 10 % market share of all global diesel exports, and now they have a market share of 1 % or so because their refiner's capacity is very low, and they've stopped exporting as a consequence of it. So there is some merit to what he's saying here, but he's obviously also trying to get people to focus on something else because they're not really getting anywhere with all that stuff. And in my opinion, it's become even more messy this weekend, and I'll allow you to unpack that.

17:19Mikkel Rosenvold:Yeah, absolutely, Andres. Let's just look at what happened this weekend. So obviously, we're nowhere closer to a US-Iran deal. I don't think that's anywhere close. I am looking at some relief from a potential Iran-Qatar deal or Iran-Gulf state deal, which could be imminent. We don't know exactly what that's going to encompass. We hope it's some products. It's maybe too much to hope for jet fuel, but could be fertilizers, could be helium, other stuff that the Gulf states are then allowed to export, or that they are more broadly allowed to export, but have to pay dues to the Iranians. We can live with that.

18:01Mikkel Rosenvold:That would still be a huge relief probably for energy costs. if we could I mean getting these sound tolls or what you might call them set up would be a worst case scenario 12 months ago, now it's almost like a Narnia, we're dreaming of getting to a point where you can exit the straight and have to pay the Iranians for it but anyway Andreas that's one point of relief we're looking for that's also what caused me to be somewhat positive on oil, not too pessimistic about where we're going. What obviously tipped that analysis was the hit on the Yanbu pipeline, the East-West pipeline. And Andreas, I had a lot of questions on this.

18:47Mikkel Rosenvold:We knew that something was happening. We could see the smoke cloud in the central Saudi Arabia. We could see that the pipeline had been hit. My initial reaction was to look at what happened And the last time this pipeline was hit, we had some comparisons back in April that took four days for the pipeline to be back up and running. So that was my best case. That was a pumping station getting hit as well. It seems like the damage is a little bit bigger this time, Andreas. They have told us that it's going to take a few weeks to get this done, that they're missing some spare parts. Admittedly, I didn't factor that in.

19:20Mikkel Rosenvold:I didn't know that. but that's obviously what's driving the oil price up today. My best guess would be that this few weeks of disturbances should be priced in at 110. The question is, can this pipeline be hit in other places? Could it take even longer? It's tough to say. In any case, Andreas, we have to say at this point, looking across the economy, the world, it's hard to point at any part of Trump's agenda that hasn't been completely derailed by the Iran war. We're talking China policy. We're talking trade policy. We're talking Europe, defense, oil, energy, cost of living. Every aspect of Trump's presidency and his political agenda is almost, aside from women in swimming pools or trans women in swimming pools, aside from that, every part of Trump's presidency is currently being holded.

20:13Mikkel Rosenvold:The progress in all other areas are being holded by the Iran war. So the pressure is very, very high on him to get dissolved. I don't see any solution anytime soon. They're also preparing the public for that. So, yeah, Andres, zooming in on oil prices, that's what people want to hear about. Anything can happen down there. We could have another strike on the pipeline that could drive oil prices up even further. Aside from that, I don't see why we should get a lot higher at this point, but let's see. Well, very, very old time times. Yeah, and Mikkel, allow me to say this.

20:50Andreas Steno:You can read more in detail around how I view the market response, both to the events this weekend and obviously also the inflation report on Friday. But look at, for example, Polymarket or other prediction markets. I don't think anyone expects a solution anytime soon. So, I mean, how much worse can it get from here? I mean, sure, it can get worse. But everyone's already in the camp that it cannot be solved. Exactly. At least this year, right? So in that sense, it's not a major negative surprise if it gets a little worse because we're already there, right? No one expects this to be solved. So in that sense, it reminds me a little, as I've said for roughly a week, it reminds me a little of the setup back in March and April where no one could see any light at the end of the tunnel.

21:42Andreas Steno:Oil prices were up. The market started positioning very bearishly. And right now, the market is long dollars. It's very short bonds starting to pile into energy longs at a fast pace. It's also, to some extent, betting against the equity markets, especially betting against some of the momentum names that we've seen in the AI build out. So you've got the Very solid cocktail for a complete wrong footing of everyone if something magically happens. That's just what I'm saying. Everyone is positioned the wrong way now if it happens.

22:18Mikkel Rosenvold:All it takes is one tweet. Remember that? Yes. And that tweet may not come this week, but all it takes is one tweet of opium to turn this around, Andreas. Okay. getting back to the FMC meeting on Wednesday, Andreas how much are you expecting in terms of forward guidance of what we're going to see over the next 3, 6, 9 months because it hasn't really been Kevin Walsh's game do you think he'll dig more into that or are we just going to have to rely on the data that we have?

22:51Andreas Steno:At least we've gotten some sort of a clear reaction function if they hike this week because then I think it's safe to assume as long as inflation is printing above 0.25 % in call terms every month, we'll probably get some more. And I guess that's been Walsh's target for a while to get us to a point where the market knows how to read the data instead of listening to him. So I would expect him to say nothing at all rather than stating, now you can see that we move if XYZ happens. And that is how you should expect us to act. So again, it will be more of a inflation now casting slash forecasting exercise than trying to second guess how they operate, which I think is a step forward.

23:41Andreas Steno:actually. I hold a lot of sympathy for his lack of guidance or his guidance of a lack of guidance. Having said that, apparently everyone, even though he spent like 30 minutes in Jackson Hole, trying to say, this is not forward guidance, this is not forward guidance, this is not forward guidance. Okay, you're hiking in September. That's sort of the conclusion, right? So it is so difficult not to forward guide at all. And the market is still caught in that exercise of trying. Okay, was that a little hint of that? Was that a little hint of that? I actually don't think he's hinting at anything. He's basically not trying to, at least.