In the high-stakes world of artificial intelligence, competition has never been fiercer or more intimate. The race to dominate AI has turned Silicon Valley into a sprawling web of uneasy alliances, where trillion-dollar rivals are simultaneously fighting and funding each other. In this new landscape, enemies are partners, competitors are clients, and no one seems entirely sure who’s keeping score.
Over the past few months, a dizzying number of cross-company deals have redefined the AI battlefield. OpenAI, majority-backed by Microsoft, signed a $300 billion agreement with Oracle for cloud compute power. Meta turned to Google Cloud in a $10 billion partnership. Microsoft, despite being OpenAI’s biggest investor, is now giving customers access to Anthropic’s models, which run on Amazon and Google’s servers. Meanwhile, Nvidia announced a $100 billion investment in OpenAI to build 10 gigawatts of AI data centers the same OpenAI that’s working on its own chips to one day compete with Nvidia’s.
This isn’t just competition. It’s corporate entanglement on a scale the tech industry hasn’t seen since the dawn of the internet.
“The stakes are so high that you’re seeing behavior that in the past wouldn’t happen,” said Gil Luria, managing director at D.A. Davidson. “It’s a big chess match of how can I progress on my stuff, but also not miss the boat if someone else is winning.”
But as the alliances grow, so do the risks. These multi-billion-dollar partnerships often rely on heavy debt, vendor financing, and circular money flows that blur the line between real growth and artificial demand. Nvidia, for example, is investing in AI cloud companies like CoreWeave, which then buy Nvidia’s chips and rent them back to Nvidia or its partners. It’s a closed loop that echoes the telecom-financing schemes of the late 1990s and those didn’t end well.
In some ways, this is nothing new. Silicon Valley’s history is littered with stories of rivals secretly depending on each other. Google pays Apple tens of billions to be the default search engine on iPhones. Netflix runs its streaming service on Amazon’s cloud. Even Apple buys components from Samsung the same company it battles in phone sales. The AI boom is simply magnifying those contradictions to absurd proportions.
At Goldman Sachs’ Communacopia + Technology conference, executives from OpenAI and Meta openly acknowledged their reliance on Google Cloud, even as both companies compete with Google in the AI space. Apple reportedly trained its AI systems using Google’s Tensor Processing Units. Google, in turn, offers Nvidia GPUs to its own cloud clients even though Nvidia’s hardware directly competes with Google’s AI chips. Meta’s internal engineers, meanwhile, have been caught using rival models to do their work, frustrated by the limitations of Meta’s own Llama AI systems.
Analyst Rishi Jaluria of RBC believes this tangled ecosystem is driven by necessity, not strategy. “People recognize it’s hard to build large language models, and not only hard, it’s really expensive,” he said. “So how do you benefit from that without taking the financial burden on your balance sheet?”
That fear of missing out is palpable. “We saw what happened when people missed the internet,” Jaluria said. “Sears could have been Amazon. Blackberry could have dominated enterprise mobile. Nobody wants to miss this race.”
Yet this wave of AI partnerships could also plant the seeds of future rivalries. OpenAI’s massive infrastructure deals with Microsoft, Oracle, and Nvidia are teaching it how to design its own data centers knowledge that could eventually make it a threat to the very cloud giants enabling its growth. As one analyst put it, “It’s like David and Goliath, if Goliath was helping David build the slingshot.”
The structure of these agreements has also raised eyebrows on Wall Street. Many involve a form of “round-tripping,” where one company invests in another, which then spends that capital buying back services from its investor. Amazon’s $4 billion investment in Anthropic which selected Amazon Web Services as its “cloud provider of choice” is a textbook case. Google, which also invested in Anthropic, now indirectly benefits from the success of a direct rival. Nvidia’s own investments in AI startups that rent its chips back to itself are even more circular.
“CoreWeave is the most glaring example of bad behavior,” said Luria. “Nvidia seeded CoreWeave to create competition. Then they signed up as a customer.” When CoreWeave’s IPO fizzled earlier this year, Nvidia swooped in with a $6.3 billion agreement to buy back any unsold capacity. “It’s like taking a margin loan from your broker to buy treasuries,” Luria said.
Still, few believe the AI boom will end like the metaverse or crypto manias that came before it. “The demand for AI is very real,” Luria said. “There’s not going to be a moment where we all say, ‘Oh, this AI stuff is nonsense.’”
Patrick Moorhead of Moor Insights & Strategy agrees. “By no means am I saying there is no risk,” he told Newspapers. “But the downstream impact of AI for companies that are doing it right is very positive.”
Even Nvidia CEO Jensen Huang, who many see as the central figure in this global AI arms race, has framed the chaos as necessary. Speaking to CNBC, Huang said he and his peers were “building a brand-new industry called AI infrastructure.”
The real question is whether demand can keep pace with the $1 trillion that’s already been poured into building this new world. Almost all of it from chips to data centers to compute access ultimately flows through OpenAI and its CEO, Sam Altman, the gravitational center of the AI universe.
“Sam Altman has the power to crash the global economy for a decade or take us all to the promised land,” Bernstein analyst Stacy Rasgon wrote this month. “And right now, we don’t know which is in the cards.”
