May has been a data-packed month for investors. Between earnings season, a steady flow of economic data releases from the government, and the Federal Open Market Committee's federal funds rate decision, there's been a lot to unpack.
But arguably the most important data release of the quarter occurred one week ago, on May 15. This was the deadline for institutional investors with at least $100 million in assets under management (AUM) to file Form 13F with the Securities and Exchange Commission. A 13F provides investors with a way to track which stocks and exchange-traded funds (ETFs) Wall Street's most prominent money managers have been buying and selling.
Though Berkshire Hathaway's Warren Buffett is the most followed of all asset managers, he's far from the only billionaire investor known to deliver outsized returns and move markets. For instance, billionaire fund manager Philippe Laffont of Coatue Management, who's overseeing $22.7 billion in AUM, has a rich track record of outperformance.
Image source: Getty Images.
Laffont is also known for his love of high-growth stocks -- especially those in the tech sector.
Based on Coatue's first-quarter 13F, its billionaire chief continued to be a seller of the world's leading artificial intelligence (AI) stock, Nvidia (NVDA -1.82%), but absolutely piled into Wall Street's hottest AI-initial public offering (IPO) of the year.
Philippe Laffont has been a persistent seller of Nvidia stock for two years
Taking into account that Nvidia completed a 10-for-1 forward split in June 2024, Coatue's position in Wall Street's AI darling peaked at 49,802,020 shares in the March-ended quarter of 2023. Over the last two years, Laffont has been paring down this position with regularity.
During the first quarter of 2025, Laffont's fund dumped 1,460,653 shares of Nvidia stock, which represents a sequential quarterly decline of about 15%. But over the last eight quarters, Coatue's billionaire boss has overseen the sale of 41,256,185 cumulative shares of Nvidia, representing 83% of the fund's original stake.
To be objective, Nvidia has done a lot of things right to get to where it is now. Its Hopper (H100) graphics processing unit (GPU) and successor Blackwell GPU architecture have run circles around the competition, in terms of compute ability. Nvidia's hardware maintains a near-monopoly-like share in enterprise AI data centers.
Overwhelming demand for Nvidia's GPUs also boosted its pricing power. With the Hopper and Blackwell commanding a premium over all other GPUs, it's no surprise that Nvidia's gross margin surpassed 70%.
But not everything is perfect for Nvidia -- and Laffont's trading activity suggests it.
Despite Nvidia having superior hardware, competitive pressures are beginning to weigh on its margins. In addition to direct external competitors ramping up production of their AI-GPUs, many of Nvidia's top customers by net sales are internally developing chips they'll use in their own data centers. The cost and accessibility advantage of relying on internally produced AI solutions could realistically result in Nvidia losing out on valuable future data center real estate.
The presence of new external and internal competition is also working to minimize the effect of AI-GPU scarcity. This has been Nvidia's primary competitive edge for two years, and it's the core reason its gross margin surged to as high as 78.4% one year ago. With its gross margin expected to decline, yet again, in the fiscal first quarter, it's clear that Nvidia's biggest advantage is withering.
The other big-time concern for Nvidia shareholders is the likelihood of an AI bubble forming and bursting. Including the proliferation of the internet in the mid-1990s, there hasn't been a game-changing innovation in more than three decades that's avoided a bubble-bursting event early in its expansion. The fact that most businesses haven't optimized their AI solutions, and in many instances aren't generating a positive return on their AI investments, strongly signals that investors have (again) overestimated the early innings utility and adoption rate of a next-big-thing trend.
With more than 90% of Nvidia's net sales coming from its data center segment in the fiscal fourth quarter of 2025 (ended Jan. 26, 2025), a bursting of the AI bubble would be disastrous for its stock.
Image source: Getty Images.
The new AI apple of billionaire Philippe Laffont's eye is a scorching-hot IPO
Although Laffont was a seller of a lot of high-growth tech stocks during the March-ended quarter, there was one artificial intelligence company that caught his attention in a big way -- and it only debuted as a public company days before the end of the first quarter!
Arguably no stock was purchased more aggressively in the opening frame of 2025 by Coatue's billionaire chief than Nvidia-backed AI-data center infrastructure company CoreWeave (CRWV 18.79%). In its two business days as a publicly traded company in the first quarter (the company's IPO was Friday, March 28), Laffont scooped up 14,402,999 shares, which vaulted it to Coatue's 16th-largest holding by market value.
The allure of CoreWeave for Laffont almost certainly has to do with the insatiable enterprise demand for AI computing resources. CoreWeave has purchased 250,000 Hopper chips from Nvidia, which is no small investment. In return, the company can lease out its AI infrastructure and services to clients, with the amount it generates in sales all dependent on things like demand, the services rendered, and the GPUs needed to complete a task.
Coatue's billionaire money manager is likely also impressed with CoreWeave's expected growth ramp. Keeping in mind that consensus growth estimates for relatively early stage businesses are often fluid, CoreWeave's sales are projected to catapult from a reported $1.92 billion in 2024 to an estimated $19.66 billion come 2028. The company also announced a strategic deal with OpenAI that tacks on $11.2 billion in its revenue backlog.
The numbers on paper absolutely paint an exciting picture for CoreWeave. But the real world doesn't always pan out as things do on paper.
To begin with, CoreWeave's net losses are accelerating at the same staggering rate as its sales. As an early stage business, the company had to rely on debt financing to fund its GPU purchases. Last year, CoreWeave had nearly $361 million in net interest expenses. Its annual run in 2025 for net interest expense, based on its recently reported first quarter, is almost $1.06 billion! Investors should expect steep losses as CoreWeave's revenue ramp-up continues.
Another sizable concern for CoreWeave, which might actually trump its rapidly widening net loss, is Nvidia's accelerated innovation cycle. Nvidia plans to bring a new high-powered AI chip to market roughly once per year. This means that CoreWeave's predominantly Hopper GPU-powered data centers could quickly become obsolete -- or at the very least, it could substantially weaken the company's pricing power for its services.
Lastly, CoreWeave would almost certainly be adversely affected by an AI bubble forming and bursting. Until artificial intelligence matures as a technology, the threat of businesses paring back their AI infrastructure spending remains a tangible concern.
