Skip to main content

OpenAI CEO Sam Altman is still chasing billions to build AI chips

OpenAI CEO Sam Altman is still chasing billions to build AI chips


Bloomberg reports Sam Altman is in talks to raise money for a ‘global’ network of fabricators building hardware for AI.

Share this story

If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.

Sam Altman onstage at OpenAI DevDay.
Image: OpenAI

A new report from Bloomberg says that once-again CEO of OpenAI Sam Altman’s efforts to raise billions for an AI chip venture are aimed at using that cash to develop a “network of factories” for fabrication that would stretch around the globe and involve working with unnamed “top chip manufacturers.”

A major cost and limitation for running AI models is having enough chips to handle the computations behind bots like ChatGPT or DALL-E that answer prompts and generate images. Nvidia’s value rose above $1 trillion for the first time last year, partly due to a virtual monopoly it has as GPT-4, Gemini, Llama 2, and other models depend heavily on its popular H100 GPUs.

Accordingly, the race to manufacture more high-powered chips to run complex AI systems has only intensified. The limited number of fabs capable of making high-end chips is driving Altman or anyone else to bid for capacity years before you need it in order to produce the new chips. And going against the likes of Apple requires deep-pocketed investors who will front costs that the nonprofit OpenAI still can’t afford. SoftBank Group and Abu Dhabi-based AI holding company G42 have reportedly been in talks about raising money for Altman’s project.

A person holding Microsoft’s Azure Maia 100 AI chip
Microsoft’s new Azure Maia 100 AI processor.
Image: Microsoft

Other companies developing AI models have also ventured into making their own chips. Microsoft, an investor in OpenAI, announced in November that it’s built its first custom AI chip to train models, closely followed by Amazon announcing a new version of its Trainium chip. Google’s chip design team is using its DeepMind AI running on Google Cloud servers to design AI processors like its Tensor Processing Units (TPU).

AWS, Azure, and Google use Nvidia’s H100 processors as well. This week, Meta CEO Mark Zuckerberg told The Verge reporter Alex Heath that “by the end of this year, Meta will own more than 340,000 of Nvidia’s H100 GPUs” as the company pursues the development of artificial general intelligence (AGI).

Picture of the GH200 AI processor platform, showing two dies on a circuit board
Nvidia GH200 “Grace Hopper Superchip”
Image: Nvidia

Nvidia has already announced its next-generation GH200 Grace Hopper chips to extend its dominance in the space, while competitors AMD, Qualcomm, and Intel have launched processors designed to power AI models running on laptops, phones, and other devices.