According to the Wall Street Journal, NVIDIA has officially invested in Safe Superintelligence (SSI), a highly confidential AI laboratory founded by Ilya Sutskever, co-founder of OpenAI and former chief scientist. The investment is described as "substantial," but the specific financial terms were not disclosed.

NVIDIA

The core of this deal is computing power. According to the agreement, SSI will be allowed to use a large amount of NVIDIA's flagship GPU hardware, which would increase its computing power by "an order of magnitude"—previously, the company mainly relied on Google-made TPU chips. NVIDIA decided to invest after getting a glimpse of SSI's research progress, showing its recognition of Sutskever's technical approach.

A Secret Giant Comes to Light: Valuation of 30 Billion Dollars

SSI was founded by Sutskever in 2024. The company's positioning is extremely clear—to achieve "safe superintelligence" in one step, that is, an AI capable of performing any task humans can do without threatening humanity.

With Sutskever's strong reputation in academia, SSI quickly attracted about $2 billion in funding from top venture capital firms such as a16z and Sequoia Capital. Its valuation reached approximately $30 billion (about 203.3 billion yuan) last year, while its research details remain highly confidential to this day.

This is not the first time NVIDIA has bet on key personnel who left OpenAI. In March this year, NVIDIA had already reached a similar investment agreement with Thinking Machines Lab, founded by former OpenAI Chief Technology Officer Miriam Prabhu, whose first AI model released this month was trained using NVIDIA hardware.

From Prabhu to Sutskever, NVIDIA is systematically binding the top research forces in the AI field, building a deeply integrated computing power ecosystem on top of GPU sales—those who can train the next generation of models must enter NVIDIA's orbit.