For most companies, a cash injection of $6.6 billion would be a game-changing event. For OpenAI, the proceeds from its latest funding round represent a mere drop in the ocean.
On Wednesday, the AI lab confirmed it had raised the money in a round led by investment funds Thrive Capital and MGX, with input from the likes of Nvidia and Microsoft, which sunk a reported $10 billion into the ChatGPT-maker less than two years ago.
The long-trailed funding round values OpenAI at $157 billion, meaning it is worth more than the likes of Goldman Sachs, Deutsche Telekom and Lockheed Martin.
Easy come, easy go
It is unlikely the $6.6 billion will be resting in OpenAI’s bank account for very long. As DCD reported in July, the company could spend up to $7 billion on training and inference for its AI models in 2024. Documents seen by The Information detailed how OpenAI had spent $4 billion in three months on the servers needed to run ChatGPT, and said the startup was on course to lose $5 billion in 2024. With costs only set to rise as OpenAI develops more powerful systems, CEO Sam Altman will likely be going to the well again in a few months.
The fact that investors are happy to throw cash at a company sustaining such heavy losses highlights the massive potential returns that many believe AI can bring. Whether OpenAI should be trusted to deliver these returns is another matter: a year ago I was sat meters from Altman when he told Marc Benioff at Salesforce’s Dreamforce conference that hallucinations - the euphemistic name given to errors made by systems like ChatGPT - are part of the “magic” of AI. Making a virtue of these mistakes is likely to be a hard sell with business executives staking the futures of their companies, not to mention millions of dollars, on the success of AI. A report from Goldman Sachs in July concluded that businesses so far have “little to show” for a collective $1 trillion investment in the systems needed to fuel AI development.
For OpenAI, the last 12 months have seen significant turbulence behind the scenes, with senior figures including influential CTO Mira Murati, departing in their droves, leaving Altman - who himself was fired and rehired over the course of a dramatic weekend in November 2023 - running the show as the company transitions away from its complex non-profit structure towards a future as a conventional, money-making business.
New products have been few and far between, with no release date yet for the company’s next-generation model, GPT-5, and no indication that it is closing in on its goal of developing AGI - an artificial intelligence capable of matching or surpassing human intelligence. Even the launch of its latest model, GPT-4o1, attracted some controversy, with Fortune reporting that it was deployed despite reservations from the company’s research and safety teams.
OpenAI loves data centers
Given that most of the $6.6 billion will be spent on compute costs, news of OpenAI’s latest funding round is, on the surface, a good thing for the data center industry.
Altman has also been talking up some potentially exciting data center-related developments, from $100 billion Stargate campus OpenAI is reportedly looking to build with Microsoft, to pitching the economic benefits of 5GW data centers at a meeting at the White House. Elsewhere, the OpenAI CEO is personally involved in two nuclear fusion startups, Oklo and Helion, both of which are working on technology that could deliver the vast quantities of renewable energy that digital infrastructure requires to meet the demands of AI systems
But these exciting future developments do little to mask the fact that data centers cannot wait forever for AI to fulfill its promise. Digital infrastructure providers have attracted billions of dollars in recent years to build out the server halls required to train and run AI systems, but until the technology proves that it can offer significant business benefits, the fears that AI may prove to be all hype and no substance remain.
Industry figures I’ve spoken to in recent weeks seem relatively unconcerned that this will happen, but as the most high-profile company in the space, any signs of stalling progress at OpenAI will lend credence to those that view AI as a bubble waiting to burst, akin to the Dot Com boom of the early 2000s. As long as investors continue to believe in Altman’s vision, it is likely OpenAI will not be short of money, and data centers will not be short of tenants. But the pressure is on for the AI lab to deliver.