How to understand the circular dealmaking fueling the AI boom

There’s a lot of money changing hands in the tech world these days. AI companies are racing to secure a steady supply of compute. Chipmakers are placing bets on who they expect to go the distance. And, occasionally, competitors are even investing in one another.

OpenAI, on Friday, announced a $110 billion funding round, with $50 billion coming from Amazon and $30 billion from Nvidia, along with other backers. AMD and Meta last week unveiled a partnership that will see the chipmaker deploy 6 gigawatts’ worth of graphics processing units to Meta’s AI data centers, while the social media/AI giant may take up to a 10% stake in AMD. That announcement came just a few months after OpenAI and AMD struck a virtually identical deal.

Nvidia, meanwhile, has become one of Intel’s largest shareholders, buying a 4% stake last September. And Amazon may buy up to 2.7% of semiconductor company STMicroelectronics over the next seven years.

It’s a dizzying pace of deals and investments. But are they strategic, or a way to inflate AI’s growth by blurring the lines between real demand and companies effectively buying from themselves? The answer is complicated.

The AI horse race

Deals like the Meta/AMD collaboration certainly carry strategic value. The chipmaker secures long-term demand for its products, while AI companies lock in a reliable supply of the chips they need to stay competitive. That creates a feedback loop of sorts: Spending on chips fuels growth at the supplier level, which leads to more advanced chips, which in turn justifies more spending. Wash. Rinse. Repeat.

“The AI industry represents a small core of companies that are building this ecosystem they believe will define the next era of technology,” says Jacob Bourne, an analyst with eMarketer. “Deals like this really underscore that belief.”

Part of what’s driving this dealmaking is the jockeying for position among the dominant players. Nvidia commands a formidable market share, but AMD is working to carve out a stronger foothold, emphasizing lower costs. By landing deals like the ones it has with OpenAI and Meta, AMD gains both visibility and credibility in the AI space—even as it continues to operate in Nvidia’s shadow.

However, says Bourne, while these deals have clear upsides for both sides, they also highlight how concentrated the current market is for the end product.

“It illustrates we’re not dealing with a very diversified industry here,” he says. “This is not a validation of organic demand for AI. They’re making these deals and it’s circular. It’s a blurring of the line between customer revenue and partner investments.”

A multibillion-dollar bet on the future

So, yes, the somewhat incestuous relationship between AI companies and AI chipmakers could be described as subsidized usage, but that usage is fueled by belief in the product. While there’s not a tremendous organic demand for AI right now, the companies behind the AMD (and other) deals are able to stockpile chips while cutting costs, says Bourne.

The circular spending is essentially a bet on the long-term future of AI. Both chipmakers and AI companies acknowledge there will be pain points in the near and mid term. Those could include hurdles in consumer and corporate adoption, or cash flow problems that prevent AI companies from meeting their spending commitments.

By diversifying in the way they currently are, the companies apparently believe they can protect themselves from those pain points.

“There’s fierce competition among this small cohort of companies, while they’re also reliant on each other,” says Bourne. “They’re trying to stake claims in this ecosystem, really betting on the power of broader AI demand in the marketplace in the future.”