AI Arms Race Intensifies Again. Here Are the 2 Big Takeaways For Investors.

OpenAi and Oracle have a joint data center. (Courtesy Oracle)

This article is from the weekly Barron’s Tech email newsletter. Sign up here to get it delivered directly to your inbox.

Infrastructure Wars. Hi everyone. Just when you thought the artificial intelligence infrastructure theme was due for a breather, the fundamental backdrop rises to another level.

This past week, several major developments suggest large technology companies are witnessing the level of investing in AI and see it as an existential threat to their businesses if they don’t also invest accordingly. It means a slowdown for AI infrastructure isn’t in sight yet, which bodes well for the AI trade.

It began on Thursday when Microsoft CEO Satya Nadella announced the company’s Fairwater AI data center in Wisconsin, which he claimed is the “world’s most powerful” AI data center filled with hundreds of thousands of Nvidia GB200 GPUs. Remember, it was only a little over one year ago that a leading AI model, Llama 3, was trained on 16,000 GPUs.

“We’re scaling our GPU fleet faster than anyone else,” Nadella said in a social media post, adding his company is building multiple identical Fairwater datacenters in other locations across the US.

The same day, Mark Zuckerberg confirmed he sees AI infrastructure spending as critical to Meta’s future. While much of the media focused on his remark admitting there is a possibility of “misspending a couple hundred billion dollars,” which referred to his pledge at the White House earlier this month that Meta will invest at least $600 billion in the U.S. through 2028, the media is missing his main message. He emphasized that Meta can’t afford be left behind and have a rival reach advanced AI technology first.

“The risk is higher on the other side [not spending enough] if you build too slowly,” he said on the Access podcast. “You are just out of position on what I think is going to be the important technology that enables the most new products, innovation, and value creation in history.”

The executive’s conviction on Meta’s AI infrastructure strategy is firm, and any speculation that he is wavering on its AI spending plans is incorrect.

Then came OpenAI’s flurry of announcements this week. On Monday, OpenAI and Nvidia announced a new partnership where the leading maker of AI chips plans to invest as much as $100 billion in OpenAI to support the buildout of 10 gigawatts of AI data center capacity. For context, one nuclear reactor can typically generate one gigawatt of electricity.

A day later, OpenAI announced five new data center sites are under development with partners SoftBank and Oracle . The new sites bring OpenAI’s Stargate infrastructure project to nearly seven gigawatts of capacity and more than $400 billion in investment over the next three years.

There are a couple of big takeaways from this past week. First, the size and scale of the AI infrastructure buildout is only getting larger. While it’s easy to point to the large numbers and wonder if this is sustainable, as of now the spending seems to be happening.

Investors should not underestimate the importance of the computing platform shift driving the demand for AI. Corporations are forced to use the technology to distill the best intelligent insights from vast pools of internal data, while being able to iterate and test all potential permutations for new product designs and innovations using the parallel processing capability of GPUs.

If they don’t use the power of AI models and GPUs, they will lag behind their competition that does. Think of AI as a productivity-enhancing tool such as a calculator or spreadsheet that enables workers to make faster, informed decisions versus those that calculate scenarios by hand or abacus.

The second takeaway is the news shows the primacy of Nvidia as the key supplier for AI computing hardware and solutions. It isn’t an accident that Microsoft mentioned the company’s AI servers and OpenAI CEO Sam Altman said only Nvidia had the scale and speed necessary to support their large buildout. Meta also uses Nvidia as its main AI chip vendor.

The latest developments show that the AI infrastructure growth story and Nvidia’s leading position as the preferred arms race supplier remain stronger than ever.

This Week in Barron’s Tech

• OpenAI, Oracle, and SoftBank Announce 5 New Stargate Data Centers
• Micron Stock Jumps After Earnings Beat. Why Investors Expect More Good News.
• Alibaba Stock Soars on AI Spending Hike. How Its Hot Streak Could Keep Going.
• CoreWeave Stock Gets Two Upgrades. Investors Can Thank Nvidia.
• Disney Is Raising Prices for Streaming. It Isn’t Alone.

Write to Tae Kim at tae.kim@barrons.com or follow him on X at @firstadopter.