OpenAI began decade ago as nonprofit lab. Musk and Altman now rivals

On Dec. 11, 2015, OpenAI launched as a nonprofit research lab after Elon Musk and a group of prominent techies, including Peter Thiel and Reid Hoffman, pledged $1 billion to develop artificial intelligence for the benefit of humanity. The idea was for the project to be be free of commercial pressures and the pursuit of money.

A decade later, that founding mission is all but forgotten.

Musk, now the world’s richest person, is long gone, having created rival startup xAI. And he’s been engaged in a heated legal and public relations fight with OpenAI CEO and co-founder Sam Altman.

Far from the nonprofit realm, OpenAI has emerged as one of the fastest-growing commercial entities on the planet, zooming to a $500 billion private market valuation, with almost all of that value accruing since the company’s launch of ChatGPT three years ago. More than 800 million people now use the chatbot every week.

Musk’s xAI, meanwhile, is expected to close a $15 billion round at a $230 billion pre-money valuation this month, sources familiar with the matter told CNBC’s David Faber in late November.

OpenAI and xAI are two of the main companies, along with Google, Anthropic and Meta, pouring money into AI models, as the market rapidly evolves from text-based chatbots to AI-generated videos and more advanced compute-intensive forms of content, as well as into agentic AI, with large enterprises customizing tools to enhance productivity.

For OpenAI, the price tag is almost incomprehensible: $1.4 trillion and growing. That’s primarily for the mammoth data centers and high-powered chips required to meet what the company sees as insatiable demand for its technology. For now, OpenAI is a cash-burning machine going up against tech’s megacaps and their chip suppliers, drawing comparisons to earlier waves of high-growth tech firms that spent heavily for years to challenge behemoth incumbents, but to mixed results.

“OpenAI has a very big role in the in the history of the development of artificial intelligence, and will forever have that role,” said Gil Luria, an equity analyst at D.A. Davidson, in an interview. “Now, will that role be Netscape, or will it be Google? We’ve yet to find out.”

It’s a position that would’ve been hard to imagine in 2016, when Nvidia CEO Jensen Huang hauled a black DGX-1 supercomputer up to OpenAI’s offices in San Francisco’s Mission District. The $300,000 machine had cost Nvidia “a few billion dollars” to develop, and there were no other buyers, Huang recalled recently on Joe Rogan’s podcast.

Musk, at OpenAI, was the only one who wanted it.

When Musk told him it was for “a nonprofit company,” Huang said all the blood drained from his face at the thought of parking such a costly box inside an organization that wasn’t meant to make money.

Behind the scenes, though, the nonprofit ideal was already under intense strain, and Musk didn’t like what he saw.

“Guys, I’ve had enough. This is the final straw,” Musk wrote in an email to his co-founders in 2017. He warned that he would “no longer fund OpenAI” if it turned into a tech startup instead of a nonprofit. Altman wrote back the next morning: “i remain enthusiastic about the non-profit structure!”

In February of the following year, Musk left the OpenAI board, and said at the time the move was to avoid a potential conflict of interest as his car company, Tesla, dove deeper into AI.

The story was more complicated.

Musk sued OpenAI and Altman in early 2024, alleging they abandoned the company’s founding mission to develop AI “for the benefit of humanity broadly,” and he’s regularly criticized OpenAI’s close ties to Microsoft, its principal backer. He also went to court to try and keep OpenAI from converting into a for-profit entity and, earlier this year, went so far as to try and acquire the AI lab for $97.4 billion.

In October, OpenAI announced it had completed a recapitalization, cementing its structure as a nonprofit with a controlling stake in its for-profit business, which is now a public benefit corporation called OpenAI Group PBC.

Musk isn’t the only early OpenAI team member who’s turned into a bitter rival. Siblings Dario and Daniela Amodei left OpenAI in late 2020 to form Anthropic, which said last month that Microsoft and Nvidia would invest in the company. The valuation from the funding round could reach as high as $350 billion.

Anthropic’s Claude family of large language models is one of the biggest competitors to OpenAI’s GPT models.

Altman is wagering that he can win the race by outspending the competition. While his company has sketched out plans for a trillion-dollar-plus AI infrastructure outlay, Anthropic has made roughly $100 billion in recent compute commitments, spaced out at various intervals over the next few years.

It all amounts to a giant bet that demand for AI services will continue apace.

“We’ve got all the various AI vendors making these huge capital investments,” said David Menninger, executive director of software research at ISG. “There’s a question as to how long those capital investments continue and whether or not they all pan out.”

Luria says Anthropic and others are making reasonable commitments based on their current growth trajectory and the funding they’ve already secured. But he said OpenAI’s approach has been based on a “fantastical set of commitments” with a “faint belief that those numbers are even possible.”

Altman told CNBC in an interview on Thursday that OpenAI is already seeing enough demand to justify its spending plans, which “makes us confident that we will be able to significantly ramp revenue.”

“It’s obviously unusual to be growing this fast at this kind of scale, but it is what we see in our current data,” Altman said, adding that “the demand in the market is pretty extreme.”

Altman said last month that he expects annualized revenue to hit $20 billion by the end of this year and to reach hundreds of billions by 2030. Its historic pace of growth has been a big boon for major tech companies.

Oracle signed a roughly $500 billion deal to sell infrastructure services to OpenAI over five years. Chipmakers Advanced Micro Devices and Broadcom have woven OpenAI-linked demand into multi-year forecasts.

But Oracle’s shares plunged 11% on Thursday after the software vendor reported weaker-than-expected revenue, a miss that dragged down Nvidia, CoreWeave and other AI-related stocks. Despite a surge in long-term contract commitments from companies like OpenAI, Meta, and Nvidia, investors are growing concerned about Oracle’s debt load that’s fueling its buildout.

Still, venture capitalist Matt Murphy of Menlo Ventures, said that in his 25 years in the venture business, “this is the mother of all waves.”

Murphy, an early investor in Anthropic, said the combination of AI models, custom chips and hyperscale data centers adds up to the potential for trillion-dollar outcomes. That explains the eye-popping level of capital expenditures and the astronomical valuations, he said.

Altman recently declared a “code red” inside his company, and shuffled resources to focus on making ChatGPT faster, more reliable and more personal, while delaying work on ads, health and shopping agents and a personal assistant called Pulse. His declaration came after Google released its Gemini 3 model last month, further accelerating the search giant’s ascent in the market.

On Thursday, OpenAI unveiled ChatGPT-5.2, a faster, more capable reasoning model that the company says is its best system yet for everyday professional use. It also struck a three-year, $1 billion content and equity deal with Disney around the Sora AI video generator.

Altman downplayed the threat from Google, telling CNBC that Gemini had less of an impact on the company’s metrics than OpenAI initially feared.

“I believe that when a competitive threat happens, you want to focus on it, deal with it quickly,” Altman said.

He said he expects the company to exit code red by January.

— CNBC’s Kif Leswing contributed to this report.