The big AI labs’ circular dealmaking, in which corporate investors invest capital in exchange for commitments of future contracts for computing resources, may be less failsafe than it seems.
OpenAI is now testing the underlying assumption that the AI labs will undoubtably grow large enough to meet those obligations.
The company has reportedly missed internal growth targets, and CFO Sarah Friar has warned that if revenue doesn’t accelerate, OpenAI could fail to pay for future computing contracts, according to The Wall Street Journal.
If that happens, it could significantly impact a number of companies that have been banking on the AI giant’s growth continuing.
Sign up for The Daily Pitch newsletter Subscribe
“The revenue miss matters more than the market is pricing in—OpenAI lost ground to Anthropic and Google in coding and enterprise earlier this year, and its infrastructure obligations now exceed $1.15 trillion across Oracle, Microsoft [and] Amazon,” said PitchBook senior analyst Harrison Rolfes, who focuses on the big three large language model providers. “If revenue growth doesn’t reaccelerate, those contracts become the most expensive fixed-cost bet in technology history.”
OpenAI disputed the reporting. “We are totally aligned on buying as much compute as we can and working hard on it together every day,” Friar and CEO Sam Altman told the WSJ in a joint statement. The company added that “the business is firing on all cylinders, and the mood internally is incredibly positive.”
In any case, OpenAI has continued to pursue more massive computing partnerships. As recently as February, the company expanded its $38 billion multi-year commitment to Amazon’s cloud computing unit by $100 billion after the Seattle-based tech giant agreed to invest $50 billion in it.
In an October research note, Morgan Stanley accounting analyst Todd Castagno warned that even legally non-cancellable AI infrastructure contracts can get renegotiated when major customers run into financial trouble, comparing the dynamic to how retailers reworked their supposedly binding commercial leases during the pandemic.
That could particularly hurt the GPU maker Nvidia.
In September, the company announced it intends to invest up to $100 billion into OpenAI to support the deployment of 10 gigawatts of Nvidia hardware, although CEO Jensen Huang told reporters in Taipei this past January that the figure was nonbinding.
Nvidia has also been actively investing in AI neocloud startups, including Lambda Labs, Nscale and CoreWeave, also effectively acting as an underwriter and supplier for many of these companies.
