OpenAI and compute partner Oracle have reportedly abandoned a planned expansion of their flagship Stargate datacenter, after negotiations were stalled by financing and Sam Altman's apparent fear of commitment.
This scuttlebutt comes courtesy of Bloomberg, which cites sources familiar with the matter who claim Oracle and OpenAI had planned to grow the Abilene, Texas, facility to as much as 2 gigawatts of capacity, up from the 1.2 gigawatt campus currently under development.
The cavitations responsible for derailing the plan were attributed to a combination of financing and OpenAI's inability to forecast demand effectively.
Now, Meta is reportedly in the running to lease the untapped and unbuilt capacity from the site's developer Crusoe after Nvidia stepped in to broker a deal.
According to Bloomberg, Nvidia put down a $150 million deposit on the future capacity before approaching Zuckercorp about maybe moving in together.
Naturally, El Reg reached out to all parties involved to find out what's really going on. Crusoe declined to comment; we'll let you know if we hear back from anyone else.
Since announcing the $500 billion Stargate initiative just over a year ago, OpenAI has tapped Oracle to deploy 4.5 gigawatts of compute capacity to fuel model development and services. That contract, which Oracle values at $300 billion over its lifetime, is apparently still on track.
Making good on that commitment will require taking on a lot of debt. This week Oracle announced plans to raise an additional 50 billion in debt and equity to finance its datacenter aspirations.
However, it seems the financing won't arrive in time to save Oracle and OpenAI's expansion plans.
Meta CEO Mark Zuckerberg meanwhile is happy to burn capital to secure as much datacenter capacity as he can lay his eyes on.
During the Social Network's Q4 earnings call in January, it announced its intention to plow up to $135 billion, roughly equivalent of Kenya's gross domestic product, into capital expenditures with an eye for GPU compute capacity.
Meta isn't alone. Combined, the eight largest hyperscalers that's - Google, Amazon, Meta, Microsoft, Oracle, Tencent, Alibaba, and Baidu - are expected to spend a collective $710 billion, much of it on bit barns and the GPUs and other trappings that go into them in 2026 alone.
While Oracle and OpenAI may not be moving forward with the Abilene expansion, Altman and crew are going to need to find somewhere to perk 5 gigawatts worth of GPUs to claim the $30 billion carrot Nvidia has offered it as part of the $110 billion funding round announced last week in collaboration with SoftBank and Amazon. ®
OPINION Passing the buck, and the blame, down the road shows lack of AI companies' maturity
Non profit loses several staffers including its executive director
Agent Memory stores AI chat scraps off to the side and recalls them when needed
The bar for creating visual assets has been lowered to the ability to converse with a model
Pause your Mythos panic because mainstream models anyone can use already pick holes in popular software