A new Google and Blackstone joint venture will bring Google’s custom AI chips to outside companies for the first time at scale, backed by an initial $5 billion equity commitment as part of a plan that could grow to $25 billion total. The Google Blackstone joint venture will be based in the United States, offering compute-as-a-service that lets enterprises and AI developers rent access to Google’s Tensor Processing Units (TPUs) without routing through Google Cloud.
Benjamin Treynor Sloss, a veteran Google executive, will lead the company as CEO. Blackstone holds the majority stake. The venture targets 500 megawatts of data center capacity online by 2027.
What the Google Blackstone joint venture does
Google’s TPUs are custom silicon built for AI training and inference. Unlike Nvidia‘s GPUs, which dominate the AI accelerator market, TPUs use an architecture Google built internally over the past decade. Until now, Google kept TPU access largely within its own cloud platform and a handful of select partners. Anthropic, for example, already runs workloads on Google’s chips under that arrangement. The Google Blackstone joint venture takes that access to a broader market.
Customers will rent TPU-based compute capacity as a service, similar to how firms like CoreWeave sell access to Nvidia GPU clusters. Blackstone’s capital base enables the kind of infrastructure buildout that Google’s cloud division has not pursued on its own.
The Nvidia angle
The deal puts Google’s TPU architecture in direct commercial competition with Nvidia’s GPU ecosystem. Nvidia holds a dominant position in AI accelerator sales, and developers have built AI workflows around Nvidia’s CUDA software stack. Google’s TPUs require a separate software path, which has kept adoption outside Google’s own systems limited.
Google has not announced any reduction in Nvidia purchasing and remains one of Nvidia’s largest customers. The Google Blackstone joint venture opens a parallel commercial channel for TPUs, aimed at infrastructure demand that has grown faster than conventional cloud capacity can serve.
Major technology companies are on track to spend over $700 billion on AI infrastructure in 2026. Against that figure, a $5 billion initial equity commitment is an entry point. With leverage, the venture’s total capitalization could reach $25 billion.
What both sides said
Jon Gray, President and COO of Blackstone, described the partnership as a large-scale infrastructure bet: “We see a generational opportunity to invest capital at scale building AI infrastructure. This new company has enormous potential as it helps to meet the unprecedented demand for compute. We are incredibly proud to partner with Google, bringing together their world class TPUs and AI capabilities with Blackstone’s exceptional strength in energy and digital infrastructure.”
Thomas Kurian, CEO of Google Cloud, framed the venture around TPU access: “This joint venture with Blackstone helps meet growing demand for TPUs, which are optimized specifically for efficiency and performance in the AI era. Together, we’re providing more options for organizations to access accelerated compute capability.”
Blackstone manages over $1.3 trillion in assets and already invests in data centers and energy infrastructure. Its role in the Google Blackstone joint venture is to supply the capital and physical infrastructure that pure software companies cannot fund at this scale. The deal signals that private capital now treats AI chip infrastructure as a standalone investment category, separate from cloud software.