Cohen's Point72 Asset Management swapped out a potentially troubled AI stock for the presumed cream of the crop.
In November, Wall Street and investors were privy to a flood of important data releases. Election Day, monthly economic data reports, and earnings season -- the six-week period each quarter where a majority of S&P 500 companies announce their operating results -- make it easy for a meaningful announcement to go unnoticed.
For example, investors may have been so swamped by other news events that they completely missed the Nov. 14 deadline to file Form 13F with the Securities and Exchange Commission. A 13F is a required filing for institutional investors with at least $100 million in assets under management (AUM) that provides a concise snapshot of the stocks Wall Street's most prominent money managers are buying and selling.
As you may have guessed, no 13F is more anticipated than that of Warren Buffett at Berkshire Hathaway. When you crush the benchmark S&P 500 like Buffett has consistently done over six decades, you're going to draw quite the following.
However, Berkshire's "Oracle of Omaha" is far from the only billionaire money manager that investors pay close attention to. For instance, investors also closely follow the trades made by billionaire Steven Cohen of Point72 Asset Management.
Cohen's fund closed out the September-ended quarter with more than $39 billion in AUM, which includes various put and call options, as well as common-stock positions. But what really stands out about Point72's trading activity during the third quarter is what Cohen and his team were up to within the artificial intelligence (AI) arena.
Cohen's Point72 bids adieu to potentially troubled AI stock Supermicro
In Sizing the Prize, the analysts at PwC forecast a $15.7 trillion increase in global gross domestic product by 2030, due to the rise of AI. But history also tells us that not every company tethered to a game-changing trend is necessarily going to be a winner.
During the September-ended quarter, Cohen's Point72 Asset Management dumped its entire position in customizable rack server and storage solutions specialist Super Micro Computer (SMCI -1.60%), which amounted to 45,066 shares, as of June 30. This means Cohen's fund exited prior to Supermicro completing its first-ever stock split of 10-for-1 following the close of trading on Sept. 30.
On paper, a lot has gone right for Supermicro. Businesses wanting to take advantage of the AI revolution are aggressively spending on data center infrastructure, with the hope of gaining/maintaining first-mover advantages. Supermicro's customizable rack servers have been a top choice by businesses running AI-accelerated data centers.
To add to the above, Super Micro Computer has been incorporating Nvidia's (NVDA -0.05%) high-powered graphics processing units (GPUs) into its rack servers. Nvidia's hardware has proven superior to the competition on a computing basis, which has further increased demand for Supermicro's data center infrastructure.
According to the company, sales in fiscal 2024 (ended June 30) surged 110% to just shy of $15 billion. Meanwhile, Wall Street's consensus estimate calls for scorching-hot revenue growth of 67% in the current fiscal year to roughly $25 billion.
But there were also a couple of clear-cut reasons for Point72's brightest investment minds, including Cohen, to ring the register and head for the exit.
In late August, short-seller Hindenburg Research published a report that accused Supermicro of "accounting manipulation, sibling self-dealing, and sanctions evasion." While the company was quick to refute Hindenburg's allegations, it nevertheless delayed the filing of its annual report and, per The Wall Street Journal, is facing an early-stage probe of its accounting practices from federal regulators.
To make matters worse, Super Micro Computer's auditor, Ernst & Young, which had previously raised concerns about the company's internal controls, resigned in late October. Even though Supermicro announced earlier this week that a review by an independent special committee expected no restatement of the company's financials, there are simply no certainties until its new auditor signs off on its financial statements and the company files its annual report.
Wall Street and billionaire investors loathe uncertainty, which is likely what sent this hypergrowth stock to the chopping block during the third quarter.
Cohen can't stop buying the hardware backbone of the AI revolution
While Steven Cohen was showing Super Micro Computer to the door, he was stuffing Point72's proverbial pockets with shares of Wall Street's most cutting-edge AI stock, Nvidia.
Cohen's fund purchased 1,574,796 shares during the third quarter, which increased its stake by a cool 75% in three months. It should be noted that Point72 also holds call options in Nvidia, which were reduced by 89% during the September-ended quarter. In other words, some of this increase may be the result of Cohen and his team exercising these call options and increasing the number of common shares owned.
The most logical reason to buy shares of Nvidia, which I alluded to earlier, is that its hardware is in high demand and superior from a computing standpoint. Orders for the company's flagship H100 GPU (commonly called the "Hopper") and successor Blackwell GPU architecture are backlogged. It's easy to understand why Nvidia's share of the AI-GPU market has been monopoly-like to date.
There's little question that Nvidia has been able to use AI-GPU scarcity to its advantage. With demand for the company's hardware handily outstripping supply, it's been able to command $30,000 to $40,000 for each Hopper chip. For some context, this is double to quadruple the price point of Advanced Micro Devices Insight MI300X GPU. A significant price premium has lifted Nvidia's gross margin to the mid-70% range and sent revenue through the roof.
Credit should also be given to Nvidia's CUDA platform. CUDA is the software toolkit developers use to build large language models and maximize the computing potential of their Nvidia GPUs. It's effectively been an umbrella that's kept customers contained within Nvidia's ecosystem of products and services.
But even Nvidia has its flaws and may not be the slam-dunk investment Wall Street and billionaire Steven Cohen believe it'll be. For instance, Nvidia is likely to lose its otherworldly pricing power and GPU scarcity advantages over the next year. In addition to AMD rapidly increasingly its production, many of Nvidia's largest customers by net sales, which are members of the "Magnificent Seven," are internally developing AI-GPUs of their own.
Even though these chips won't have the same computing potential as Nvidia's hardware, they're going to be notably cheaper and more easily accessible. In other words, it creates a situation where Nvidia can lose valuable data center real estate in the coming quarters.
The other serious issue for Nvidia is that no game-changing technology or innovation for at least 30 years has avoided an early stage bubble. Investors have persistently overestimated how quickly a new technology would gain utility and be adopted. The lesson is that all technologies take time to mature, and artificial intelligence is unlikely to be an exception. If the euphoria surrounding AI fades, Nvidia and its shareholders would, presumably, feel the pinch.