Everyone said crossover funds like Situational Awareness had the AI edge. Now they're telling you to buy.
By far, the best tweet I saw on the Situational Awareness blow up was this meme
I already put up some (very) fast thoughts on Situational Awareness’s blow up last Thursday… but it’s been pretty much the story for finance over the past few days. Combine that buzziness with a really interesting angle / potential alpha opportunity I see in that tweet, and I wanted to put up a follow-up post.
What is that angle / alpha opp?
The biggest wins in the “AI trade” on the public markets have generally belonged to crossover funds that invested in both public and private equities. So, for the past two years, you’ve heard a lot of people complaining that a lot of those funds’ alpha has been driven by them getting access to monthly numbers from AI startups, seeing usage and demand go through the roof in real time, and then using that info to go buy all of the public AI beneficiaries. You can get as deep as you want with this theory; you can say they were just getting monthly numbers and using those to buy beneficiaries, or you can go as far as to say they were discussing component pricing, shortages, user usage, and potential deals in real time and using those to buy beneficiaries.
No fund better illustrates this than Situational Awareness. Leopold was not only a former researcher at OpenAI, but he’s engaged / just married the chief of staff at Anthropic. Hard to imagine someone better positioned to trade on AI news flow than that; all it takes is his fiancé coming home and saying, “Man, business is booming. These new coding tools are exploding and we can’t bring on enough compute to keep up with demand” and off to the races!
Personally, I think that thought process is a little too conspiratorial. While I’m sure that seeing Anthropic’s monthly numbers helped build conviction as the positions started skyrocketing (it’s easy to sell a stock up 100% and miss the next 500%; it might be easier to hold a stock that’s gone up 100% when you see demand has gone up 500%!), I think all of these funds had on all of this exposure (and invested in Anthropic and other AI start ups directly!) because they were true believers in the AI trade, not because they were swing trading quarters on the Anthropic data!
But let’s put my personal beliefs to the side for a second. Pretend you do believe that these crossover funds are trading on “insider” info on the pipelines and outlook from the large private labs, and that a bunch of the crossover funds got margin called because they failed to properly risk manage or something.
That makes what a lot of them are saying right now very interesting. Consider one of the closing lines in Situational’s letter to LPs:
On the portfolio itself: we are very optimistic about the current investment opportunity set. The underlying fundamentals are accelerating at the very same time that prices have declined significantly.”
Situational is not alone in that call out. Across the board when I talk to investors who are doing privates and publics, they are saying that demand continues to explode.
I am aware that asking crossover investors who are massively long privates exposed to AI and AI bottleneck trades if the outlook remains strong despite a 30-50% drawdown over the past month is somewhat akin to asking a barber if you need a haircut…. but it’s not just them saying demand is booming! Go read the Q2 calls for Amazon / Microsoft / Google1 / META2; all of them are telling you that demand is outstripping supply, that the ROI on buildouts and implementing AI is awesome, and that they’re only scratching the surface of demand. For example, here are three quotes from AMZN that make exactly those points:
“I'll start with AWS, which is booming right now, and I'll share the numbers what we think is going on and why we're enthusiastic about the ROIC equation,”
“the demand we already have for 2028 is striking. And remember, enterprises are still very early in using inference at scale in their current production applications.”
“Brian, we have so much demand right now. Apart from what we've talked about in '26, a lion's share of capacity in '27, we're adding a lot of capacity, as I mentioned just a few minutes ago, is largely reserved. And we have quite a bit of capacity that's already been reserved for '28.”
What I’m driving to is this: if you have been looking at everyone trading AI stocks through crossover funds over the last 18 months and jealously saying “they’re just using private company data to generate a once in a lifetime trading edge,” then right now you’ve got a unique alpha opportunity. All of those same funds are saying the fundamentals only continue to get better at the same time that the stock prices of their holdings are imploding. Maybe that implosion is being driven by the market’s skepticism of the future, maybe the implosion is forced selling from degrossing, maybe it’s a combination of the two…. but, whatever it is, if you believe the crossover funds, then this sell-off is a unique opportunity to generate a heck of a lot of alpha as the companies report continued record earnings and the degrossing fades away.
So that’s the setup / opportunity. Do I believe in it? Honestly, I’m not sure! Despite the recent pullback, a lot of these companies continue to trade for massive valuations. That said, there do seem to be some pockets of value. Remember that most of these companies are minting money in the short term, so, even if some of them are up quite a bit, a good chunk of that move could be attributed just to the near term cash generation of the businesses. And many of these companies have taken advantage of the supply crunch to sign long term deals that should materially derisk them and support longer term earnings.
So, yes, the stocks are up a lot despite the pullback…. but the fundamentals support a lot of those moves, and I do want to keep an eye out for value despite big moves. Just to tip my hand a slight bit, the place I’m personally most interested right now are power shells. Generally, these are former bitcoin miners that have converted to data center plays. Many of them have signed long term deals with hyperscalers for their power, and the recent pullback has a lot of them priced at valuations that simply credit their current signed contracts and give them no real value for future deals or the terminal value of the assets after the hyperscaler contract rolls off.
PS- I am working on a post or three diving into the power shells a little more; don’t forget to subscribe if you want to receive those!
I covered a lot of the Google bullishness in my “What is Google seeing in AI?” series; Part 1 (history's biggest, weirdest equity raise), Part 2 (Google's telling you the returns are real), and Part 3 (the real bear case).
Disclosure: Long META, largely because of their YOLO options package