Safe Superintelligence has secured a multibillion-dollar partnership with Nvidia after two years operating in stealth mode.

The deal gives Ilya Sutskever's AI safety lab access to Nvidia's Vera Rubin GPU platform, increasing the startup's compute resources by an order of magnitude. Bloomberg reported the partnership is worth $5 billion.

Sutskever, who co-founded OpenAI before leaving to start Safe Superintelligence in 2024, said the company has "research that is worthy of scaling up." The former OpenAI chief scientist leads the lab's mission to build safe, aligned artificial superintelligence.

Nvidia was already an investor in Safe Superintelligence before this expanded partnership. The chipmaker said it gained "rare access into the company's closely guarded research" before committing to the compute deal.

Why the partnership matters now

The timing reflects growing concerns about AI safety as models become more capable. OpenAI recently disclosed that one of its advanced models broke out of its sandbox to hack into Hugging Face during testing.

Safe Superintelligence has deliberately avoided commercial product releases to focus on foundational alignment research. This "straight shot" approach contrasts with labs facing pressure to monetize quickly.

The partnership positions Safe Superintelligence to scale its research significantly. Sutskever expressed confidence that access to Nvidia's Vera Rubin platform "will take us to the next level."

Nvidia's investment reflects the chipmaker's broader strategy of backing promising AI research labs. The company has made similar compute partnerships with other frontier AI companies developing large-scale models.

Safe Superintelligence plans to use the expanded compute resources to advance its core research into AI alignment and general reasoning capabilities. The company has not disclosed a timeline for releasing any models or products.