Safe Superintelligence Nvidia partnership brings new AI compute
Ilya Sutskever’s AI lab will use Nvidia’s Vera Rubin platform under a long-term deal that includes an undisclosed investment.
By Theo Nakamura · Staff Writer
· 3 min read
Safe Superintelligence Nvidia plans now include a long-term strategic partnership aimed at giving Ilya Sutskever’s AI lab far more computing power. For investors watching the AI buildout, the deal is another sign that advanced model research remains tightly linked to Nvidia’s chips, cloud-scale systems and capital.
Nvidia said the agreement gives Safe Superintelligence, also known as SSI, access to its Vera Rubin GPU platform. GPUs, or graphics processing units, are the chips used to train and run large AI models because they can handle many calculations at the same time. Compute means the processing capacity those chips provide.
The partnership includes an investment, though Nvidia did not disclose the size in its announcement. A person familiar with the transaction told TechCrunch that Nvidia’s investment extends into multiple billions of dollars.
Nvidia said SSI’s compute resources are expected to rise “by an order of magnitude,” a phrase that generally means about a tenfold increase. The company also said SSI had reached significant research milestones, without giving specific details.
What does the Nvidia partnership give Safe Superintelligence?
The deal gives SSI access to Nvidia’s Vera Rubin GPU platform and a closer working relationship with Nvidia on current and future compute platforms, according to Nvidia. Nvidia said the companies will use SSI’s technology and its views on where AI is headed to inform that work.
Sutskever said in a statement that SSI has “research that is worthy of scaling up” and that access to a large Nvidia computer will help the company do that. He also said SSI is making a “big bet” on the Vera Rubin platform.
Nvidia said it entered the compute partnership after getting rare access to SSI’s closely held research. The chipmaker was already an investor in the company before the new agreement.
What is Safe Superintelligence?
Safe Superintelligence is an AI research company founded by Sutskever, a former OpenAI co-founder and alignment lead. Alignment means work intended to make AI systems behave in ways that match human goals and safety constraints.
SSI has kept a low profile since its founding two years ago. The company says it is focused on building safe, aligned artificial superintelligence through a direct research path, rather than releasing commercial products or chasing near-term revenue.
The company also partnered with Google Cloud last year to power its research, according to TechCrunch. The new Nvidia deal adds another major infrastructure partner at a time when cutting-edge AI labs need large clusters of specialized chips to train more capable models.
Sutskever is widely known in AI research. He co-authored and co-created AlexNet with Alex Krizhevsky and Geoffrey Hinton, work credited with showing that GPU scaling and deep neural networks could produce major advances in machine learning.
Before starting SSI, Sutskever led OpenAI’s Superalignment team, which has since been dissolved. He left OpenAI months after an unsuccessful effort to remove CEO Sam Altman, following what Sutskever called a “breakdown in communications,” according to TechCrunch.
SSI has raised $7 billion to date and carries a $32 billion post-money valuation, according to PitchBook data cited by TechCrunch. Its backers include Nvidia, Andreessen Horowitz, Alphabet, Lightspeed Venture Partners, GV, Sequoia Capital Partners and others.
This story draws on original reporting from TechCrunch.
