In January 2025, just one day after Donald Trump’s second inauguration, AI tech leaders convened in the Oval Office as part of the announcement of a flashy $500 billion AI infrastructure deal, dubbed “Stargate.”
OpenAI CEO Sam Altman gushed over his newfound adoration for Trump, telling him during the event that “for AGI to get built here, we wouldn’t be able to do this without you, Mr. President.”
The company said it was committing $100 billion immediately, sparking a heated debate over whether it had secured the necessary financing with Elon Musk, founder of competitor xAI, who was notably absent from the proceedings.
Over a year later, OpenAI has dramatically reined in its ambitions as reality continues to settle in. For one, its astronomical commitments to spending $1.4 trillion before the end of the decade on AI infrastructure was recently more than halved to a still-hefty $600 billion. Meanwhile, panicked executives are looking to cut out any distracting “side quests” to double down on enterprise and coding, two areas potentially capable of generating some desperately-needed revenue.
The company’s remaining ambitions to secure computing power have landed it between a rock and a hard place — without more compute, the company could risk falling further behind the competition. On the other hand, attempts to secure more of the resource could balloon its spending even further, potentially scaring investors.
Ahead of its rumored IPO, pressure continues to mount over the company’s calamitous finances. As a result, as CNBC reports, OpenAI has started to sing a dramatically different tune compared to early 2025.
“OpenAI has come to the realization that the market doesn’t necessarily appreciate the reckless approach to growth and spending,” Futurum Group CEO Daniel Newman told CNBC. “The market wants to see OpenAI’s revenues rolling at a pace in which the spending can be justified. The pivot has been to try to show a little bit more fiscal responsibility.”
As inside sources told CNBC, the company still doesn’t own any data centers outright, instead relying on purchasing cloud capacity from other companies such as Oracle, Microsoft, and Amazon.
Meanwhile, its own ambitions to build data centers have remained on shaky ground. AI chipmaker Nvidia’s $100 billion investment in OpenAI, a plan to have the latter deploy ten gigawatts of computing power using the former’s chips, may encounter major headwinds.
That is, if it ever turns into a viable project. Nvidia CEO Jensen Huang effectively admitted earlier this month that the $100 billion figure was largely plucked out of thin air, saying that his company’s recent $30 billion investment in OpenAI “might be the last” before OpenAI goes public.
Even with a massively scaled down plan, actually bringing data centers online could prove extremely challenging.
Virginia Tech engineering professor Walid Saad told CNBC that it could take three to ten years to build a one-gigawatt data center. OpenAI claims such an inaugural data center will be ready to be deployed before the end of this year.
“There’s regulations, there are permits, different locations have different processes,” he said. “There are processes they cannot control. You never know what pops up.”
Meanwhile, investors will be looking extremely closely at any of OpenAI’s upcoming moves as the company continues to light billions of dollars on fire.
“To their credit, they built an incredible growth story,” Newman told CNBC. “It’s just — the rest of the ride won’t be a free one.”
“And because their cost structure is so high, their route to profitability will be scrutinized every step of the way,” he added.
More on OpenAI and data centers: Panicked OpenAI Execs Cutting Projects as Walls Close In
The post OpenAI’s Obsession With Data Centers Is Running Into Trouble appeared first on Futurism.
