Loading summary
A
Hello out there, everyone, and welcome to this week's edition of Macro Mondays. My name is Mikael Rosenwald and as usually, I'm joined by my co host and partner in crime, Andre Stino. What a Monday. Andreas, how's your day been?
B
You know, when I woke up this morning, I was about to hang myself. Honestly. No, it was such an open to basically the NASDAQ futures market. And you know, a couple of hours later, I'm, I'm, I'm starting to sort of grasp what this story is all about and I'm a little bit more comfortable now. Overall, this Deep Seek Nvidia's story is incredibly interesting also from a macro perspective. But what's a little bit amusing is that right about everything that we know right now, we also knew on Friday. So, you know, something happened over this weekend. Maybe people actually got the time to read through the research paper on Deep Seek, you know, because we had internal discussions on the topic in our fund on Friday and it was a big thing, but we were like, okay, we were like, let's address this when it becomes relevant. And oh boy, it's relevant now. So, yeah, it's been a crazy day.
A
You can't ignore that. We're looking at a route in tech stocks, obviously, the Nasdaq down over 3%, especially both Nvidia and Meta Alphabet, all of them taking quite the beating. Sandra, you just recorded a flash update with our great colleague Alex. It should be out in a few hours on the Pro macro side. So for a much deeper dive also into some of the specifics of Deep seq, you can go have a look at that. I'm really looking forward to that, but we'll try and cover some of the macro aspects of this and yeah, what to make going forward here. So, Andreas, what. Just to sum up, for those who've been living under a rock so far today, or maybe just getting up a little late here, what's the story?
B
So Deep Seek, a Chinese AI, yeah, released recently, I think the research paper was out Wednesday last week, all of a sudden pulled the rock from under the technology space, especially the semiconductor space in the US and in Taiwan this morning. And I guess the short version of a much longer story is that Deep Seek is incredibly more efficient than OpenAI, for example. Right. So look at it this way, it delivers, say, 80, 90% of what ChatGPT does at a much, much lower level of computational power. And is that a big thing? Well, you know, initially, yes, because obviously from a hardware perspective, right. If you need fewer GPUs to more models. That could be a short term issue for Nvidia. Right. And we obviously know that it is priced to some sort of perfection. Now, whatever you call it, we, we have expectations of a doubling of the top line this year and probably around 80% on top of that next year. At least we had that those expectations before today. And for a stock that is so dependent on its growth trajectory, even one or two quarters with disappointments will probably be enough to alter the trend substantially. Having said that, I guess the more medium term story is very compelling here as well. Because if Deepseek and the technology behind it is indeed a game changer and most people I talk to think that it is, it is also a game changer for medium term inflation. It is a game changer for medium term demand for semiconductors, it's a game changer for the entry barriers for large language models. You know, it's a game changer for right about everything. And it all points in the direction of us reaching some sort of singularity.
A
Earlier than expected and possibly also electricity demand.
B
Yeah, yeah, yeah, it's a good point, Miguel, because if you look at nuclear stocks today and utility, well, they're not doing well because this obvious link between, okay, we need to use much more power, we need to have small modular reactors close to data centers, all that story is also sidelined to some extent by this story today. But I'd like to show you a meme that I found early this morning.
A
I love these build shot beams.
B
Yeah, yeah, yeah, yeah. You know, I love them as well. And I, I think this take is Right. Yeah. So, you know, and as per usual, these Bell charts, right, they're, they're, they're taking it to sort of the edge of what's fair. Right?
A
Yeah.
B
But the point here is a lot of people talk about, okay, this is a problem for Nvidia this quarter next, because the big clients, Microsoft, Alphabet, etc. Will not buy as many GPUs as they can see that they can basically thrive with fewer GPUs, et cetera. Right. I think there's some merit to that story. But the bigger story is that entry barriers will be lowered substantially. And this is by the way, a story that we always see with software. Right? And now China is involved. Involving China in right about anything means that it's a race to zero right. In costs. And I think this is a classic example of the same. Right. And if you lower the entry barriers, you lower the costs of producing basically by increasing the efficiency, the demand for the hardware Underlying it will actually increase, not decrease.
A
Exactly.
B
That's the end game in my opinion. What happens in between that end game and now is probably that Nvidia will take a drop and then reaccelerate timing wise. I think it's tricky to find that bottom right now, even though it seems like we've, you know, at least judging from the price action, it seems like at least the first wave is over here. But let's see.
A
Yeah, I always try to, to move to the left of these bill charts. That's the. Keep to the simple analysis. Don't fuck this up. I think that's also what rooted today. So, so extremely interesting address here. I just want to touch on the fact that Deep Seq is coming out of China because I think that's very, very relevant as well. This is in many media, many analysts are pointing to this as a Sputnik moment. Let's just remember the Sputnik was the first satellite to orbit the Earth. Earth was launched by the Russians, by the Soviet Union and it caused the US to invest massively in the space race. If this is viewed as a Sputnik moment, then this should provoke even larger investments into AI from within the US from the US Government as well. And I think that might very well be the case because remember this, AI tech is obviously it's important for the tech sector, but it's equally important for the defense industries of both China and the U.S. so if this is viewed as wow, the Chinese are leapfrogging us on AI, then we simply have to invest more into this. So that's one angle of it. So, Andres.
B
I wanted to add one thing to this. It's not only Chinese, it's a spin out of a Chinese hedge fund. And I've had long discussions with hedge fund counterparts today whether to actually trust this story fully because let's assume that hi fly as the hedge fund behind this ran some sort of negative or bearish position on Nvidia before throwing this news out there. I mean, it's a very, very weird incentive structure to have an AI company and a hedge fund at the same time. At least you need to be aware of that incentive structure when you read this. So I just wanted to get that out there because, you know, it's not irrelevant to know. At least I don't, I cannot prove anything. I don't have any information on it. Just want to stress that I just look at the incentive structure and I find it fishy. Let me put it like that.
A
I agree completely. There are some other points Too obviously, we've all seen the tweets of Deep Seq refusing to answer questions both related to whether Xi Jinping is related to Winnie the Pooh and all sorts of sensitive Chinese questions. Of course it refuses to answer these. There's also a point to this being sort of. Read that about this being sort of a data grab. I think that's an interesting angle because at the end of the day, one thing is hardware, one thing is software is the data that you need to continue to work on these models. And if you have the best model, that's where people are going to go with all the data, with all their prompts, and that is going to be a competitive factor as well. So to put out a model that people perceive as being better than OpenAI's model or Claude or whatever you have might be very, very powerful as well. Let's try and understand some of the market logics here because it shouldn't be a surprise that we're seeing innovation in AI space. Why such a heavy drop today?
B
I think, you know, looking at Nvidia first, right. I think it's, it's clear that it is a. It's an equity case that needs growth quarter by quarter. And even though the medium term prospects are pretty solid on the back of this, it's an issue for a growth case to have one or two disappointing quarters in a row. So should we get that get out? Right. That's kind of the short version of why it matters so much. The market is incredibly bad at looking beyond what's in the next quarter or two. And even though we can sit here and agree that the medium term is probably even better than a week ago, that will be tricky to navigate until the growth story is back quarter to quarter. So that's why it matters a whole lot. Then looking at the spillovers to US fixed income to the US dollar, I find the ramifications maybe even more interesting from a macro perspective because all trend lines are broken now in the strong dollar. And it's a very easy sell to say that this is a dollar negative story because flow wise the dollar has been thriving on the back of equity flows pointing only in one direction and that's basically towards Silicon Valley. And if we get some sort of pause in that story, it will allow the positioning to unwind quickly in the dollar. I think dollar yen looks incredibly compelling as a macro case. The Japanese yen obviously being a risk off currency as well. But also given that dollar yen trades maybe 10 index points or sorry, 10 big figures above its fair value when you look at interest rate spreads between the US And Japan. So I, I think that's a compelling story. I think we kind of needed a trigger to really sell the dollar in big size and maybe we've gotten that exact trigger today. Obviously also related to lower bond yields in the U.S. very interesting.
A
I was a bit surprised, just to round off this topic, Andreas, that it didn't take more than a storyline like this to rattle all this tech investment. Is this a case of all these tech stocks being priced in at such multiples at which. Which as you mentioned, the pricing in almost exponential growth that it takes so little to throw off the curve that that makes investors more, more nervous? Is that what we're looking at or am I, am I over reading this?
B
No, no, I think that's exactly what we're looking at. It was the same thing that happened to Tesla back in 2223. You simply need to deliver on the priced in growth also near term to justify the multiples. And that's why we'll have a rocky road ahead. You know, we've been banging the drum on buying equal weight S and P in this Trump era rather than Nasdaq, and so far we've been 100% right on that. It is incredibly difficult to time the price action in Nasdaq both on, on the up and on the down. I'm not even sure that I really want to try because, you know, the market has been incredibly schizophrenic since the December presser from Powell. You know, we've had these nasty drawdowns in nasdaq, then all of a sudden we're back. I like not to be involved when the market is behaving like that. I consider it a better choice to be in the equal weight S and P. And it's been outperforming Nasdaq for a couple of months in a row now.
A
Go to the left side of the bill chart. That's.
B
Yeah, maybe.
A
So dress. Finally a week where Trump is in the top topic here. Although I just want to touch upon the whole.
B
Maybe he'll end up being the center of attention in this AI case as well. Because we obviously know that the Stargate project was launched, was it last Wednesday as well? Right. It was more or less around the same timing of the research paper from Deepseek. It. It sounded like Trump was like personally vested in that Stargate story as well. Then we had Elon going out on X, like trying to question whether they had the financing and all that. It's obviously something that's on the radar in this administration. And I think it was as late as last week. We talked about the prospects of doing a tech deal between us and China. Not necessarily sure that that tech deal has become a whole lot easier after this. And maybe Trump is one or two hours away from tweeting something about this.
A
That would be a major game changer again.
B
Yeah, yeah, exactly. It's just he probably needs to comment on this. I would go as far as saying, yeah, absolutely.
A
And it seems like while it seems like Elon is sliding a little bit in the internal hierarchy of the Trump administration, why don't just buy this whole deep seek thing and get himself right back in the, in the, on the front seat of the Trump car. I don't know. Trump did have a major win over this weekend, the standoff with the, with Colombia. I think this was a demonstration to the, to the analysis that we've laid out over, over the past month in dress that, that of course Trump is serious about tariffs, but mostly for the point of being serious, like people need to take this serious as a threat. And the Colombian government absolute. So long story short, the Colombian government refused to accept or to allow a couple of planes full of illegal immigrants expelled from the US to land in Colombia. This infuriated Trump who then retaliated with, I think it was 25% tariffs that would rise to 50% in a week. And then within hours, the Colombian government folded very much demonstrating one, that this is a viable strategy for Trump to gain what he wants, at least short term. He's not building any long term friendships here. And also to demonstrating that these suggested tariffs, he has to do them very specifically on single countries, not entire regions or globally. That's where they're the most efficient. Because they hurt Colombia. They will hurt Colombia much more than they will hurt the US and yeah, great effect out of that. We might see that weapon used even further down the road. Perhaps not in China. It seems like China are slipping out of that one one. But a big win for Trump essentially, at least in the short term.
B
Yeah, but maybe Migo, you know, I don't know. On, on the question on tariffs in, in China, he said as said as much last week that he kind of preferred not to use them even on China. Right. So if they do deals, he will not use them. The question is just how they handle this Deep Seek stuff. Could, could be a game changer for those negotiations. But on the other hand, right. Is deep seek. Could that, could that become too powerful for Xi.
A
Right.
B
And you know that's another question that we cannot answer at this point. You know, he basically made a power grab on Alibaba in 21, I think it was. Right. So I mean could, could this be something that is too powerful for, for xi's taste as well? You don't really know that. Right. So lots, lots of questions right now. If again, and I want to stress that we had a great discussion with our CTO, Alex Miller on the technical ramifications of Deep Seq, why it matters for open AI, why it matters for Nvidia, etc. And you can, you can watch that in the Pro macro session on Real Vision. It requires you becoming a member.
A
Absolutely. You should do that. So Andreas, we try to sort of tune out of all the current noise or the Trump noise in this final part of the show. You had in your weekly state of signals post on Sunday, you had some very interesting indicators on, on the economic activity in the U.S. could you just explain a little bit this chart on withheld taxes, I think was, was very, very telling. Could you just, just give us some insight into that and what, what that means for, for positioning?
B
Yeah. So the reason why. And we basically intend on doing that every week in this show, Mick, like try to cut through all of the noise to find bets that have strong embedded characteristics in a very noisy macro and geopolitical world as we have right now. And you know, one of the things that I've noticed basically since as you can see on the chart November thereabout, is that the amount of taxes withheld on behalf of employees in the US is rising quite substantially. So why is that an important indicator for us live? Well, it's, it's one of the time series components of our now costing model in the US since it gives you a daily flavor of the labor market activity basically. Right. And you know, we had a stronger non farm payrolls report than most people anticipated. The first one that was released this year. And I think the answer is found right here. You know, you cannot expect withheld taxes to pick up unless there's a pickup in hiring. Right. Maybe if wages rise a lot. But we either need to see wage growth or hiring for these withheld taxes to increase. It's as simple as that. So I think it's a pretty strong sign that we've actually seen a hiring pickup since the election. And again, something that, you know, it speaks against having a negative view on the US from a like fundamental macro perspective. Right. We can have niche stories as this one on, on AI and, and semiconductors, but if you Buy the equal weight, S and P. Based on this, I think you're okay.
A
Right.
B
Because it seems like there's, you know, momentum in, in labor markets and in the, you know, spending gauges and all that in the US Still.
A
Very interesting. I just wanted to touch upon this. You're obviously contributing greatly to this number of, number of flight flights, pointing to great consumption also.
B
Yeah. You know, in my opinion, flight data is the best gauge of discretionary consumer spending. Right. Because. And I think you'll agree, right, Miguel, if, if you struggle to pay your bills, you're not booking a flight to Japan as a tourist. Right. You know, it's, it's one of the easiest expenses to cut off your budget.
A
For businesses as well. Yeah, yeah.
B
Also for businesses. Right. It happens straight away in a downturn. Right. And we see this flight data picking up again. So I think that's another strong sign that discretionary spending is actually on the up. And last week we went to the Middle East. I'm going to US this week, Singapore soon. Right. So we're contributing. Yes. We'll have to ensure that our carbon emission is not too crazy. Right. For our ESG reporting in the company, Miguel. But anyway, that's done.
A
ESG is done.
B
At least in the US it seems to be done. It's certainly not done where we're from. Yeah. I had a laugh when I saw Christine Lagarde saying that maybe a lot of people will leave the US because of all of the turmoil created by Trump. And then I tweeted, well, at least a lot of ESG officials will be open to new work. Because it just seems like ESG is done, DEI is done. Because Trump has basically told both private and public companies to get over with it. Right. Stop it. But we are not there in Europe. Not even close. Let me put it like that.
A
No. One thing is Trump, another thing is simply sort of the dynamic, the societal development. I think it's a topic that's getting a little less relevant right now. It's going to find a natural level perhaps of importance. So we're seeing all the major businesses scaling it down, major politicians scaling it down. As I mentioned before, I think the moment where Alexandria Ocasio Cortez removed her pronouns from her X account, that was a, a very, very big moment for, for, for the anti work movement, obviously. Okay, Andreas, a crazy day here. Any, Any final thoughts for the show?
B
Well, you know, I think you need to be very agile if you're involved in the semiconductor slash stack trade. I think there's merit to the story that we have an issue for growth targets of hardware sales, say over the next 1 2/4. Based on this, let's assume that you entered the C level of either Microsoft, OpenAI Alphabet, etc. Having received the news that they've more or less proven, they've basically proven that they can run large language language models with less computational power in China, I think the first thing I would demand as a board member would be to say, okay, we need to do that before we buy more GPUs. Basically do that first and then we can buy some more GPUs. But by the end of the day it is disinflationary because as an end consumer of chat gbt, I would also demand the price to drop now because more alternatives, less entry barriers, all of that, it's disinflationary. It's incredibly bullish for the demand for GPUs over time and will reach singularity before forecasted one or two weeks ago. So, you know, I'm kind of torn between the very, very near term and the structural outlook here. So if you're one of those involved in markets with a long term horizon, I think you should be happy about what happened today.
A
So address. Finally, I just want to touch. We've had some audience questions here. As I was looking at things this morning, I was looking at our strategy portfolio, my own portfolio. I was thinking, oh well, at least crypto is safe from this. Well, not, not at all. Why does this affect crypto? How do you think this, this will affect crypto?
B
It's a sentiment thing. I don't think there is, you know, to my knowledge there is no like crystal clear and direct links between the two stories. So I think it's more of a sentiment thing. And we've also seen, yeah. Some sort of consolidation later in the day here. So you know, by the end of the day, not, not, not a biggie. You know I'll, I'll still be focused on on the next moves from the Trump admin on this rather than this story.
A
Great stuff, Andreas. I think we will round it off at that. Thank you very much all of you out there for tuning into our Macro Monday show. It's a real pleasure to hold hosties every week. Remember that you can check into an even deeper dive into the deep sea tech and or case and what this means for tech if you're a pro macro member. If you're not, you should try that out. Absolutely. It includes our shows with role as well and a lot of other great content. Thank you so much for joining Andreas. We'll be back next week. Thank you all for looking in out there. Have a good one.
This week’s Macro Mondays focuses on the seismic market reaction to DeepSeek, a Chinese AI model that claims radically higher efficiency than existing Western models like OpenAI’s ChatGPT. Andreas Steno Larsen and Mikael Rosenwald analyze the implications for tech stocks (especially Nvidia), global macro trends, U.S.-China political dynamics, and market sentiment. The discussion balances real-time investment insights with deeper macro themes, skepticism around the DeepSeek news, and implications for future growth, geopolitics, and asset allocation.
Key Points:
“If you need fewer GPUs to more models, that could be a short-term issue for Nvidia … even one or two quarters with disappointments will probably be enough to alter the trend substantially.”
Memorable Moment:
Quote (Mikael, 04:39):
“Earlier than expected and possibly also electricity demand.”
Key Points:
“…if this is viewed as a Sputnik moment, then this should provoke even larger investments into AI from within the U.S. …”
Notable Skepticism:
Questionable Incentives: DeepSeek is a spin out from a Chinese hedge fund (HiFly), raising questions about market manipulation.
Quote (Andreas, 07:56):
“…let’s assume that HiFly, as the hedge fund behind this, ran some sort of negative or bearish position on Nvidia before throwing this news out there … I find it fishy.”
Censorship & Data Angle: DeepSeek refuses to answer sensitive Chinese questions; possible “data grab” strategy at play.
Quote (Mikael, 08:47):
“There’s also a point to this being sort of a data grab … if you have the best model, that’s where people are going to go with all the data, with all their prompts…”
Key Points:
Nvidia and Exponential Growth Pressure: Market only cares about near-term growth—one “bad” quarter can devastate a growth stock.
Broader Ramifications: Could disrupt dollar strength via unwind of equity flows (Silicon Valley = strong dollar thesis).
Tech Stock Valuations: Current multiples leave stocks vulnerable to any disruption in growth trajectory. Quote (Andreas, 12:27):
“You simply need to deliver on the priced in growth also near term to justify the multiples. And that’s why we’ll have a rocky road ahead.”
Strategic Asset Allocation: Promotes S&P equal-weight over Nasdaq for stability. Quote (Andreas, 12:27):
“We’ve been banging the drum on buying equal weight S&P … It is incredibly difficult to time the price action in Nasdaq…”
Key Points:
Memorable Moment:
Quote (Andreas, 16:32):
“Could this be something that is too powerful for Xi’s taste as well … .You don’t really know that. Right. So lots, lots of questions right now.”
Key Points:
Withheld Taxes Indicator: Rising withheld taxes signal robust U.S. labor market and hiring. Quote (Andreas, 18:12):
“The amount of taxes withheld on behalf of employees in the US is rising quite substantially … one of the strongest daily flavors of labor market activity.”
Discretionary Spending: Flight data shows uptick in personal and business travel, indicating robust consumer health. Quote (Andreas, 20:18):
“Flight data is the best gauge of discretionary consumer spending … if you struggle to pay your bills, you’re not booking a flight to Japan.”
ESG/DEI Diminishing in U.S.: Societal focus shifting; U.S. companies and politicians are dialing back ESG and DEI priorities. Quote (Andreas, 21:16):
“ESG is done, DEI is done. Because Trump has basically told both private and public companies to get over with it. Right. Stop it.”
Key Points:
AI Efficiency = Disinflationary: Consumers should see prices drop and more alternatives emerge.
“By the end of the day it is disinflationary because … more alternatives, less entry barriers, all of that, it’s disinflationary. It’s incredibly bullish for the demand for GPUs over time and will reach singularity before forecasted…”
Crypto Market Impact: Current tech rout affecting crypto is mainly “sentiment-driven”; no direct link to DeepSeek. Quote (Andreas, 24:35):
“It’s a sentiment thing. I don’t think there is, you know, to my knowledge there is no like crystal clear and direct links between the two stories…”
The DeepSeek saga is shaking up market narratives, but Henrik and Mikael argue it’s not just a threat; it could also be a long-term positive for growth, efficiency, and the AI ecosystem. Investors should expect short-term volatility—especially in hardware—but recognize the structural, disinflationary, and potentially bullish effects as barriers fall and innovation accelerates.
Key Advice:
Stay agile, focus on diversified or equal-weight allocations, and watch the intersection of politics, AI, and macro signals to guide decisions in this rapidly evolving landscape.
For deeper dives on technical AI ramifications, check out Steno Research’s Pro Macro content.