Under the deal, the AI startup will get access to Nvidia’s newest AI infrastructure, increasing its compute capacity “by an order of magnitude.” Nvidia has also made a financial investment, which sources told Bloomberg is worth US$5 billion. Neither side disclosed the amount of capital or number of chips involved. (The Logic, Bloomberg)

Talking point: Safe Superintelligence—which Canadian computer scientist Ilya Sutskever founded in June 2024 after leaving OpenAI—plans to build smarter-than-human AI in a single shot, without releasing any interim products along the way. Precisely how it plans to do that remains a closely guarded secret, but it will require a lot of compute. Nvidia seems to have seen enough to spend big on the startup. The chip giant has taken stakes in most every AI lab developing frontier models, including Anthropic, Cohere, Mistral, OpenAI, and Thinking Machines.