Since the advent of the internet in the mid-1990s, investors have pretty much always had a next-big-thing innovation, technology, or trend to captivate their attention. However, no addressable market has seemingly loomed larger since the arrival of the internet than the current rise of artificial intelligence (AI).

What makes AI so special is the ability for software and systems to learn and evolve over time without human intervention. This means AI-driven software and systems can become more proficient at assigned tasks, as well as grow to learn new skills. On paper, this gives artificial intelligence utility in almost every sector and industry around the globe.

Image source: Getty Images.

The massive addressable market associated with AI — an estimated $15.7 trillion lift to the global economy by 2030, according to the analysts at PwC — isn’t lost on Wall Street’s top investment minds. However, quarterly Form 13F filings with the Securities and Exchange Commission show that billionaire money managers have mixed feelings about the most-popular AI stocks.

In the June-ended quarter, more than a half-dozen billionaire investors were active sellers of the “brains” behind AI-accelerated data centers. But interestingly enough, numerous billionaire asset managers were avid buyers of two other key AI players that have announced or completed a stock split this year.

Data-center hardware kingpins Nvidia and AMD were shown to the door

Although enterprise demand for the graphics processing units (GPUs) that power generative AI solutions, train large language models, and oversee split-second decision-making for AI-driven software and systems, has been off the charts, billionaires haven’t been afraid to ring the register and take profits on the two most-prominent AI-GPU developers: Nvidia (NVDA -9.53%) and Advanced Micro Devices (AMD -7.82%).

The June-ended quarter marked the third consecutive quarter that more than a half-dozen billionaires were sellers of Nvidia’s stock. The seven well-known sellers in the latest quarter include (total shares sold in parenthesis):

• Ken Griffin of Citadel Advisors (9,282,018 shares)
• David Tepper of Appaloosa (3,730,000 shares)
• Stanley Druckenmiller of Duquesne Family Office (1,545,370 shares)
• Cliff Asness of AQR Capital Management (1,360,215 shares)
• Israel Englander of Millennium Management (676,242 shares)
• Steven Cohen of Point72 Asset Management (409,042 shares)
• Philippe Laffont of Coatue Management (96,963 shares)

By some combination of fate and coincidence, AMD also had seven billionaire money managers that sent its shares to the chopping block in the second quarter (total shares sold in parenthesis):

• Ken Fisher of Fisher Asset Management (5,716,366 shares)
• Ole Andreas Halvorsen of Viking Global Investors (3,952,088 shares)
• Ken Griffin of Citadel Advisors (2,649,937 shares)
• Israel Englander of Millennium Management (977,904 shares)
• Jeff Yass of Susquehanna International (536,689 shares)
• Philippe Laffont of Coatue Management (502,688 shares)
• David Tepper of Appaloosa (260,000 shares)

While profit-taking is certainly one reason we’ve likely witnessed notable selling in Nvidia and AMD from billionaires, historic precedent might be an even bigger catalyst.

For 30 years, numerous can’t-miss technologies, innovations, and trends have come and gone, and not a one has avoided an early innings bubble-bursting event. Investors have a tendency to become wide-eyed with the potential for game-changing technologies, and almost always overlook that it takes time for new innovations to mature and be adopted by consumers and/or businesses.

The simple fact that most companies lack a well-defined game plan for their AI investments is a clear signal that AI euphoria has likely overshot the technology’s current utility.

Competitive pressures are another reason billionaire investors might be headed for the exit with Nvidia and AMD. Nvidia can’t keep up with enterprise demand for its H100 GPU and is liable to lose valuable data center “real estate” with its top four customers (all members of the “Magnificent Seven”) developing in-house AI-GPUs. While AMD can chip away at Nvidia’s monopoly like AI-GPU market share with its substantially cheaper MI300X AI-GPU, the end result of both companies ramping production is less AI-GPU scarcity and, over time, weaker pricing power.

Though billionaires were decisive sellers of the brains behind high-compute data centers, they were clear-cut buyers of two other AI stock-split stocks.

Image source: Getty Images.

More than a half-dozen billionaires piled into Broadcom

While prominent billionaires were busy dumping shares of Nvidia and AMD during the second quarter, there were piling into AI networking solutions specialist Broadcom (AVGO -6.16%), which completed a 10-for-1 forward-stock split in July. To somewhat keep with the theme, a total of seven billionaire investors were buyers of Broadcom stock, including (total shares purchased in parenthesis):

• Ole Andreas Halvorsen of Viking Global Investors (2,930,970 shares)
• Jeff Yass of Susquehanna International (2,347,500 shares)
• Israel Englander of Millennium Management (2,096,440 shares)
• Ken Griffin of Citadel Advisors (1,880,740 shares)
• David Siegel and John Overdeck of Two Sigma Investments (1,332,230 shares)
• Ken Fisher of Fisher Asset Management (865,090 shares)

Broadcom made its grand entrance into the AI space with the release of the Jericho3-AI fabric in April 2023. Jericho3 is capable of connecting up to 32,000 GPUs, with a purpose of reducing tail latency and helping businesses maximize the computing capacity of their GPUs.

But in spite of the clear boost Broadcom has received from artificial intelligence, billionaires might be even more impressed with its operating diversity. For instance, it’s one of the leading providers of wireless chips and accessories used in next-generation smartphones. Telecom companies continuing to expand the reach of 5G have lifted demand for Broadcom’s smartphone solutions.

Broadcom has also leaned on acquisitions as a way to expand its ecosystem of products and services and promote cross-selling opportunities. This includes buying information technology management software and solutions provider CA Technologies, gobbling up cybersecurity company Symantec, and paying $69 billion for cloud-based virtualization software provider VMware. The latter will help Broadcom meet the private- and hybrid-cloud needs of businesses.

Having true revenue diversity means that Broadcom would almost certainly fare better than Nvidia and AMD if the AI bubble were to burst, as history suggests it will.

Six billionaire money managers scooped up shares of Super Micro Computer

The other high-flying AI stock-split stock that had billionaires mashing the buy button in the June-ended quarter is rack server and storage solutions provider Super Micro Computer (SMCI 0.93%). Six billionaires opened a position or added to their existing stakes in Super Micro, including (total shares purchased in parenthesis):

• Israel Englander of Millennium Management (553,323 shares)
• Jeff Yass of Susquehanna International (508,814 shares)
• Ken Griffin of Citadel Advisors (98,752 shares)
• Steven Cohen of Point72 Asset Management (45,066 shares)
• Ray Dalio of Bridgewater Associates (15,777 shares)
• Cliff Asness of AQR Capital Management (1,040 shares)

On Aug. 6, when Super Micro lifted the hood on its fiscal fourth-quarter operating results, its board announced the approval of a 10-for-1 stock split, which will take effect after the close of trading on Sept. 30.

While investors have been placing a lot of emphasis on the GPUs that make AI-driven software and systems tick, Super Micro Computer is reminding Wall Street just how important AI infrastructure is for enterprise data centers. The company’s customizable rack servers, which incorporate Nvidia’s leading H100 GPU, have been in high demand, as evidenced by the 110% sales growth Super Micro recorded in fiscal 2024 (ended June 30). The midpoint of Super Micro’s sales guidance — $28 billion — for fiscal 2025 implies nearly triple-digit year-over-year sales growth.

The downside for Super Micro Computer is that it’s heavily tied to Nvidia’s fortune. Supply constraints have made it impossible for Nvidia to meet all of its customer’s demands, which means Super Micro likely isn’t reaching its potential.

Moreover, we’ve been here before with Super Micro. This is a company that soared during the initial enterprise cloud-computing boom, only to have orders fail to meet Wall Street’s lofty expectations not long thereafter. If AI needs time to mature as a technology, Super Micro’s stock could give back most of its gains.