Alphabet Inc. and Nvidia Corp. have taken stakes in Safe Superintelligence Inc. (SSI), the artificial intelligence research firm co-founded by former OpenAI chief scientist Ilya Sutskever, a source confirmed to Reuters. Their participation anchors a substantial $2 billion funding round, which now values the young company at $32 billion.
The round, led by venture capital firm Greenoaks, also saw participation from prominent VCs including Andreessen Horowitz, Lightspeed Venture Partners, and DST Global. This major backing, paired with a concurrent deal for SSI to utilize Google Cloud’s specialized AI hardware, underscores the high value placed on ventures aiming to define the future of artificial intelligence.
This infusion of capital marks a notable shift for SSI. The company was established in June 2024 by Sutskever (Chief Scientist), Daniel Levy (Principal Scientist, previously OpenAI), and Daniel Gross (ex-Apple AI lead, focusing on compute and funding). Operating with dual headquarters in Palo Alto and Tel Aviv and a small team, SSI’s stated focus is purely on the challenge of developing safe superintelligence. Its initial $1 billion funding round announced in September 2024, which valued it near $5 billion, explicitly lacked Big Tech participation then. Despite the current $32 billion valuation, SSI has not yet released any public AI models or technical demonstrations.
Google Cloud Secures SSI as Key TPU Customer
In parallel with Alphabet’s investment, SSI has committed to using Google Cloud’s Tensor Processing Units (TPUs) for its AI research. Google confirmed the partnership during its Cloud Next ’25 event on April 9th, stating in a blog post, “Safe Superintelligence is partnering with Google Cloud to use TPUs to accelerate its research and development efforts toward building a safe, superintelligent AI.” This positions SSI as potentially Google’s most significant external TPU customer since the custom AI accelerators were offered commercially.
TPUs are Google’s hardware designed specifically for accelerating machine learning workloads, contrasting with Nvidia’s more general-purpose GPUs. While Google Cloud offers both, sources indicate SSI is primarily using TPUs for its current development, aligning with Google’s strategy to attract leading AI players. Google Cloud executive Darren Mowry told Reuters, “With these foundational model builders, the gravity is increasing dramatically over to us.”
Google recently detailed its seventh-generation ‘Ironwood’ TPU, focusing on inference performance. The choice of TPUs over GPUs often involves balancing specialized performance gains against broader applicability, a calculation central to large-scale AI research efforts like SSI’s.
Strategic Bets In A Competitive AI Landscape
The investments from Alphabet and Nvidia are viewed by observers as strategic placements within the fiercely competitive AI sector. For Alphabet, backing SSI provides insight into a research direction potentially complementary to its own DeepMind division. Nvidia’s participation ensures it remains connected to a potentially influential player, even if SSI currently favors Google’s TPUs, aligning with its broader strategy of investing across the AI ecosystem which includes OpenAI and xAI. This follows a pattern where major tech firms fund compute-intensive AI startups who often become significant customers, such as Microsoft’s relationship with OpenAI and Amazon’s backing of Anthropic.
SSI’s large funding round occurs amid considerable activity in the AI startup scene, particularly involving OpenAI alumni. Mira Murati, OpenAI’s former CTO, launched Thinking Machines Lab in February 2025, attracting top researchers like OpenAI co-founder John Schulman and pursuing substantial funding. This illustrates the trend of well-resourced, specialized AI labs emerging to pursue distinct research directions, making the ability of SSI to secure major backing from Alphabet and Nvidia a notable development.
Sutskever’s Focus Beyond Scaling
Ilya Sutskever, who departed OpenAI in May 2024, has publicly indicated a shift from the pure scaling approach that dominated the 2010s. He remarked to Reuters in November 2024, “The 2010s were the age of scaling, now we’re back in the age of wonder and discovery once again. Everyone is looking for the next thing,” adding, “Scaling the right thing matters more now than ever,” as covered previously.
Speaking at the NeurIPS 2024 conference in December, he reiterated the challenge of data limitations, stating, “We’ve reached peak data. There’s only one internet,” and highlighted the difficulty and importance of building AI capable of genuine reasoning, a focus distinct from current models’ pattern-matching strengths, as reported before. This philosophical direction likely drives SSI’s research agenda and its specific technology choices, such as the adoption of Google’s TPUs, as it pursues the complex goal of safe superintelligence.
