Billionaire David Tepper Just Made a Massive AI Chip Rotation — Will It Pay Off?

David Tepper of Appaloosa Management is considered one of the world’s top hedge fund managers, and through his investment prowess, he’s amassed an estimated net worth of $28.5 billion. The Carolina Panthers’ owner was reshuffling his artificial intelligence (AI) semiconductor positions in Q3, exiting positions in Advanced Micro Devices (NASDAQ:AMD) and Broadcom (NASDAQ:AVGO), while buying shares of Nvidia (NASDAQ:NVDA) and Micron (NASDAQ:MU). He also cut his stake in Alphabet (NASDAQ:GOOGL) (NASDAQ:GOOG) nearly in half.

While Tepper’s track record speaks for itself, I think he got it wrong in this case. Here’s why.

Missed AI’s “Act 1”? Act 2 Could Be 16x Bigger. Most investors think they missed the AI boat because they didn’t buy Nvidia in 2005. But according to our analysts, we’re only at the end of “Act 1″—the R&D phase. “Act 2” is the global rollout. Continue »

The great rotation

While Tepper was buying Nvidia shares before Jensen Huang stepped down as CEO, the company is facing potential issues beyond the loss of its CEO. Competitors have started to catch up, with Semianalysis recently reporting that Alphabet’s Tensor Processing Units (TPUs), which it co-developed with Broadcom, are delivering better performance per dollar. Meanwhile, all of Nvidia’s largest customers, outside of Space Exploration Technologies (NASDAQ:SPCX) are working on developing their own custom chips.

The AI application-specific integrated circuit (ASIC) market is growing much more quickly than the GPU market, with Counterpoint Research projecting that custom AI chip unit shipments will surpass data center GPU shipments in 2028. Two of the biggest winners from this trend are Alphabet and Broadcom.

Alphabet developed its TPUs more than a decade ago, and its comprehensive ecosystem is starting to rival Nvidia’s. Not only is TPU growth supported by Alphabet’s internal use and its cloud offering, but it is also beginning to sell its solutions to external data center operators, such as Anthropic and OpenAI.

Bloomberg Intelligence believes that TPU sales outside of Google Cloud could grow from just $12 billion this year to over $135 billion in 2028. This is a huge revenue driver for both Alphabet and Broadcom, which appear to be two of the biggest winners in this rotation.

Even bigger news for Nvidia competitors, though, is that Anthropic (NASDAQ:ANT) revealed that its Claude Mythos model demonstrated that it can get models up to speed running on AMD GPUs in just two to three days without any human involvement, dramatically cutting down the six to 12 months it used to take a dozen engineers to get models running on different chips. This greatly narrows Nvidia’s CUDA moat and opens the door for both AMD and ASICs to take market share.

Source link

Tagged:

Leave a Reply

Your email address will not be published. Required fields are marked *