Palantir Billionaire Peter Thiel Sells Nvidia and Buys an AI Stock Up 476,900% Since Its IPO

view original post

Hedge fund billionaire Peter Thiel sold his entire stake in Nvidia and started a position in Microsoft during the third quarter.

Billionaire Peter Thiel is best known for his role in cofounding Palantir Technologies. But he also runs hedge fund Thiel Macro, which sold its entire stake in Nvidia (NVDA +1.42%) during the third quarter and reinvested the proceeds into Microsoft (MSFT +1.78%), a stock that has advanced 476,900% since its IPO in March 1986.

I think Thiel’s hedge fund abandoned Nvidia too early, but Microsoft is well positioned to benefit from the artificial intelligence (AI) revolution, even after achieving phenomenal gains over the last four decades. Here’s what investors should know.

Image source: Getty Images.

Nvidia: The stock Peter Thiel sold

Nvidia’s graphics processing units (GPUs) are the most coveted chips on the planet. They are used to accelerate demanding data center workloads, particularly those involving AI. Nvidia has an over 80% revenue share in AI accelerators, but increasingly stiff competition may have persuaded Peter Thiel to exit his position.

While Advanced Micro Devices trails Nvidia in terms of GPU performance, its MI350 chips achieved reasonably good results at the latest MLPerf benchmark event, standardized tests that measure the performance of AI systems. AMD could narrow the performance gap next year with the launch of its MI400 GPUs. Indeed, OpenAI plans to deploy MI450 chips in late 2026.

The market is particularly concerned about custom AI accelerators. Several hyperscalers (i.e., companies with massive data center footprints) have already designed and deployed custom chips to mitigate their dependence on Nvidia. That includes Alphabet‘s Google, Amazon, Microsoft, Meta Platforms, and OpenAI. But custom chips have a major drawback in that they lack prebuilt software tools.

Nvidia has spent nearly two decades building its CUDA platform, an unparalleled ecosystem of pretrained models, application frameworks, and code libraries that support developers. Nothing of the sort exists for custom chips, which means developers must build the necessary tools from scratch. When those adjacent costs are included, custom chips are often more expensive than Nvidia GPUs.

As a result, numerous analysts think Nvidia will maintain 70% to 90% revenue share in AI accelerators, a market forecast to expand at 29% annually through 2033. In turn, the Wall Street consensus says Nvidia’s earnings will increase at 37% annually over the next three years, which makes the current valuation of 44 times earnings look relatively cheap.

Today’s Change

(1.42%) $2.52

Current Price

$180.34

Microsoft: The stock Peter Thiel bought

Microsoft is the largest enterprise software company and the second-largest public cloud, and the company is exploiting its strong presence in those markets to monetize artificial intelligence. Its strategy includes integrating generative AI copilots into popular software products like Microsoft 365, and providing a broad range of cloud AI services

“Customers continue to adopt Microsoft 365 Copilot at a faster rate than any other new Microsoft 365 suite,” CEO Satya Nadella told analysts on the recent earnings call. He also said 90% of Fortune 500 companies use the AI assistant. In cloud computing, sales growth decelerated to 28%, and Microsoft’s market share remained unchanged. But the company was capacity constrained, so it could gain share as its data center footprint doubles in the next two years.

With that in mind, Wall Street expects Microsoft’s earnings to increase at 14% annually during the next three years. That consensus estimate may be conservative, given that enterprise software and cloud spending are projected to increase at 12% and 20% annually, respectively, through 2030. However, assuming Wall Street is correct, the current valuation of 34 times earnings is tolerable.

To be clear, those numbers give Microsoft a price-to-earnings-to-growth (PEG) ratio of 2.4, with values above 2 generally considered expensive. But that multiple is lower than the three-year average of 2.6 and the five-year average of 2.5, so I think it creates a reasonable entry point. Investors will still pay a premium to buy the stock, but the premium is a little bit smaller than usual right now.