Google and Blackstone have announced the creation of a new U.S.-based AI infrastructure company designed to provide large-scale access to Google’s Tensor Processing Units, or TPUs, through a standalone compute-as-a-service platform.
The venture represents one of Google’s most aggressive moves yet to challenge Nvidia’s dominance in AI computing infrastructure while expanding external access to its custom AI chips beyond Google Cloud’s own ecosystem.
Under the partnership, Blackstone will commit an initial $5 billion in equity capital to help build the platform’s infrastructure footprint. The companies expect to bring the first 500 megawatts of data center capacity online in 2027, with plans to scale significantly over time.
The new company will offer customers cloud access to Google’s TPUs alongside networking, operations, and data center infrastructure optimized for training and running advanced AI models. Until now, most external TPU access primarily flowed through Google Cloud services.
Google said its TPUs have been deployed internally for more than a decade and currently power workloads for Gemini as well as AI systems used by major research labs, financial firms, and high-performance computing customers.
Blackstone, which manages more than $1.3 trillion in assets and is one of the world’s largest investors in digital infrastructure and data centers, will help finance and scale the operation. The partnership combines Google’s AI hardware and software stack with Blackstone’s experience in large-scale energy and infrastructure deployment.
The companies also named longtime Google infrastructure executive Benjamin Treynor Sloss as CEO of the new venture. Sloss spent more than two decades helping build and operate Google’s global infrastructure systems.
Google Pushes TPUs Beyond Its Own Cloud
The partnership signals a strategic shift in how Google positions its custom AI chips in the broader AI infrastructure market.
While Google has long used TPUs internally and offered limited cloud access through Google Cloud, the new company effectively creates an independent distribution channel for TPU-powered compute capacity. That could make Google’s hardware more directly competitive with Nvidia-powered infrastructure providers such as CoreWeave and major hyperscale cloud operators.
The move also reflects rising demand for alternatives to Nvidia’s GPUs, which currently dominate the AI training and inference market. AI companies, enterprises, and governments are increasingly searching for additional sources of high-performance compute as shortages, pricing pressure, and supply concentration around Nvidia continue to intensify.
Google said the standalone platform will give organizations more flexibility in accessing accelerated compute infrastructure optimized specifically for AI workloads.
The AI Infrastructure Race Intensifies
The launch comes during an unprecedented global expansion of AI infrastructure investment as technology companies, private equity firms, and cloud providers race to secure computing capacity for next-generation AI systems.
Blackstone described the project as a “generational opportunity” tied to accelerating demand for AI compute. The company has been expanding aggressively into digital infrastructure, power systems, and data center development as AI workloads drive soaring electricity and hardware requirements.
For Google, the deal provides a pathway to scale TPU adoption more aggressively outside its own products and cloud services while creating a stronger competitive position against Nvidia’s expanding ecosystem. The company has spent years developing TPUs internally, but the Blackstone partnership could transform them into a much broader commercial infrastructure platform for external AI customers.
