Hedge Fund CIO: “By The Time These Things Are Glaringly Obvious, It’s Too Late, Checkmate”

By Eric Peters, CIO of One River Asset Management

Colosseums

“Classic late cycle,” said Sparks. “Beginning of the end, that’s what’s unfolding,” said the investor, entrepreneur, iconoclast. “At the end, everything converges to financial flows.” In every huge cycle, the last stage is Ponzi-like. “Who’s going to buy the last CDO? Who’s going to buy the last CDO-squared? How do those securities even work? That’s what happens in these cycles. The last securities to be issued are the most complex and hardest to understand. They’re the riskiest,” he said. “The cost of capital starts going up. This is how it always ends.”

“You saw the Aussie data center IPO [here] that failed this week, right?” asked Sparks. “It’s their neocloud equivalent, like CoreWeave, it’s the type of sign we saw when the Bear Stearns hedge funds failed.” Summer of 2007. “Some of the marginal deals will not pencil anymore.” And that slows the whole machine down. “When that happens, investors go from the fear of missing out to the fear of holding the bag. That’s a process, it doesn’t happen overnight, but by the time these sorts of things are glaringly obvious, it’s too late, checkmate.”

“When the process starts in earnest, it becomes self-fulfilling, irreversible,” explained Sparks. “That’s really the point in investing, finding processes that once set in motion lead to a predictable outcome.” Even if the timing remains uncertain. “Someone buys a bond that funds a Meta data center. Now it’s trading at 88 cents. And then you read Anthropic just offered Blackstone a yield of over 10% to fund its datacenter.” Four months ago, Anthropic funded at 8.5%. “So, you take your loss and the bond volatility into account, and you’re not eager to buy.” 

“These dynamics slow the flywheel,” continued Sparks. “The greater fool theory starts to reverse and self-perpetuates.” I nodded. “Everyone wants to know exactly when it will fall,” he said. “On some level it doesn’t really matter, because once the process starts, you pretty much know stocks won’t massively blow through the highs.” The rough ceiling on the market is in place. “You don’t have to worry about the FOMO move that rips 30% higher and makes you look foolish. So, instead, you can build a portfolio for the sacking and looting ahead.” 

“Once the cost of capital rises for the weaker players, you crack the hidden pillars supporting the edifice,” said Sparks. “It’s like removing the little fasteners holding the structure together.” The Roman Colosseum flashed through my mind, its structure originally secured by 300 metric tons of iron clamps. But after Rome fell, thieves stole the metal, which in turn weakened the arena, and it succumbed to earthquakes. “That’s why a major top is a process, it takes time to undermine the structure. And it all ends when the Generals are killed.”

Anecdote 

“What do you want to accomplish in capitalism?” asked Sparks, not waiting for a response. “You want to spend as little money as possible to create the biggest monopoly you can. That’s literally all you want to do. Nothing more.”

Fair point. “You can talk about adoption and demand curves all you want, and it’s not that they’re unimportant, but they tell you less about whether you have a good business and more about the size of your net worth as a monopolist.”

Owning a big monopoly is better than a small one but having a baby monopoly is better than running a big business in a fiercely competitive industry with razor thin margins. “Everyone loves Amazon, but if you strip out its AWS and advertising businesses, it’s really just a retailer and logistics company, and those are generally low margin, capital intensive businesses,” he said.

“And yes, there are some elements of Amazon retail that are monopolistic but compare it to Google – that’s a real monopoly, with massive margins.” Amazon’s retail business is probably worth roughly the same as Walmart at $1trln. Google is worth $4trln. “So what will these AI companies look like on the other side of this rapid build out?” Amazon retail or Google?

“Even if you make the argument that everything in our economy will eventually be turned into AI tokens in one form or another, so that token demand will be explosive, why do you think that the frontier labs will be the real beneficiaries?” asked Sparks. “From the days when we first commercialized electricity to today, global usage has exploded. But owning utilities has not been the way to make your fortune.” Indeed.

“The nature of this technology and the transformation of electricity into tokens looks a lot more like the utilities than something that will manifest into durable monopolies. And I think that as this cycle matures, if we survive as a species, we’ll find that the real beneficiaries will be the users of tokens, not the AI companies that manufacture them.”