Alphabet and Nvidia have joined top venture capital firms in backing Safe Superintelligence (SSI), a startup co-founded by former OpenAI chief scientist Ilya Sutskever. According to a source familiar with the matter, SSI has rapidly gained prominence and is now considered one of the most valuable AI startups just months after its inception.
The investment highlights a renewed wave of interest from major tech companies and infrastructure providers in supporting startups pushing the boundaries of AI development—work that demands immense computing power. Earlier this week, Alphabet announced a deal through its cloud division to provide SSI access to its proprietary tensor processing units (TPUs), advanced AI chips developed in-house.
SSI was recently valued at $32 billion in a funding round led by Greenoaks, making it one of the highest-profile players in the AI research space. The startup benefits from Sutskever’s strong reputation in identifying breakthrough trends in artificial intelligence.
Alphabet’s dual move—corporate investment and cloud partnership—with SSI and another leading AI lab, Anthropic, underscores the company’s evolving hardware strategy. While Google initially reserved TPUs for internal use, it now expands access to external AI companies. According to Darren Mowry, Managing Director of Startup Partnerships at Google, the decision to supply SSI with a significant volume of chips reflects the company’s growing push to scale its AI hardware business.
“With these foundational model builders, the gravity is increasing dramatically over to us,” he said.
While Nvidia’s graphics processing units (GPUs) remain the dominant choice in AI development—controlling over 80% of the AI chip market—Safe Superintelligence (SSI) has taken a different route. According to two sources, SSI primarily uses Google’s tensor processing units (TPUs) rather than Nvidia’s GPUs for AI research and development.
Google Cloud offers both Nvidia GPUs and its proprietary TPUs, which are engineered to excel in specific AI workloads. These chips are designed to be more efficient than general-purpose GPUs and have already powered the development of large-scale AI models by companies like Apple and Anthropic—an OpenAI rival backed heavily by both Google and Amazon.
Meanwhile, Amazon is also emerging as a significant player in the AI chip race with its processors, Trainium and Inferentia. As early as 2023, Amazon confirmed that Anthropic would develop its models on these custom chips. In December, Amazon announced that Anthropic would be the first to use its new AI supercomputer, built with hundreds of thousands of these in-house chips.
The trend reflects a broader industry shift: cloud providers are funding top AI startups and locking them in as major customers for their infrastructure. Amazon and Google have poured billions into Anthropic, while Microsoft has made massive investments in OpenAI. Nvidia, too, has supported OpenAI and Elon Musk’s AI venture, xAI.
The surge in investment from tech giants like Alphabet and Nvidia into startups such as Safe Superintelligence (SSI) and Anthropic underscores the increasingly strategic role AI plays in shaping the future of cloud infrastructure and chip innovation.
