Google Cloud has announced that Safe Superintelligence (SSI), an artificial intelligence startup founded by Ilya Sutskever, is leveraging its tensor processing units (TPUs) for advanced AI research.
This collaboration, revealed on April 9 at Google Cloud Next 2025, positions Google Cloud as SSI’s primary computing provider, enabling the startup to accelerate its efforts in developing safe superintelligent AI systems.
SSI, which emerged from stealth in June 2024 following Sutskever’s departure from OpenAI, secured $1 billion from prominent investors, including Andreessen Horowitz and Sequoia Capital. The funding will support the computational demands of training large-scale AI models and recruiting top researchers.
Sutskever, who co-founded OpenAI and previously worked at Google Brain, has emphasised SSI’s mission to focus exclusively on developing safe superintelligent AI. The company’s website underscores this as its sole objective. Aside from the partnership with Google Cloud, it remains unclear whether SSI has agreements with other computing providers as well.
Besides SSI, AI music startup Udio is using TPUs to help train its models for music generation and serve its rapidly growing customer base.
Meanwhile, Google unveiled Ironwood, its seventh-generation TPU, designed specifically for inference. It’s a key part of Google’s broader AI Hypercomputer architecture.
“Ironwood is our most powerful, capable and energy-efficient TPU yet. And it’s purpose-built to power thinking, inferential AI models at scale,” Google said. The tech giant said that today, we live in the “age of inference”, where AI agents actively search, interpret, and generate insights instead of just responding with raw data.
The company further said that Ironwood is built to manage the complex computation and communication demands of thinking models, such as large language models and mixture-of-experts systems. It added that with Ironwood, customers no longer have to choose between compute scale and performance.
Ironwood will be available to Google Cloud customers later this year, the tech giant said. It currently supports advanced models, including Gemini 2.5 Pro and AlphaFold.
The TPU supports two configurations, one with 256 chips and another with 9,216 chips. The full-scale version delivers 42.5 exaflops of compute, over 24 times the performance of the El Capitan supercomputer, which offers 1.7 exaflops per pod. Each Ironwood chip provides 4,614 TFLOPs of peak compute.
The post Ilya Sutskever’s Startup Uses Google’s TPUs to Build Safe, Superintelligent AI appeared first on Analytics India Magazine.
