Google, Blackstone Launch $5B AI Cloud Joint Venture for TPU Access

Google is partnering with asset manager Blackstone to create an artificial intelligence cloud company focused on delivering specialized computing infrastructure. The venture will provide data center capacity, operations, networking, and access to Google Cloud's tensor processing units as a compute-as-a-service offering.

Blackstone is committing $5 billion initially to the joint venture, with the new company targeting to bring its first 500 megawatts of capacity online in 2027. Google will supply hardware including TPUs, along with software and services, enabling the venture to scale rapidly and meet growing demand for accelerated computing.

Customers will have the option to access cloud TPUs through this new venture in addition to using them directly through Google Cloud. The move addresses rising demand for Google's tensor processing units, which are custom chips optimized specifically for training and inference in advanced AI models.

This joint venture with Blackstone helps meet growing demand for TPUs, which are optimized specifically for efficiency and performance in the AI era, said Thomas Kurian, Google Cloud's CEO. > Together, we're accelerating AI transformation and providing more options for organizations to access accelerated compute capability.

Google's TPUs power both external customers such as AI labs and other firms, as well as the company's internal AI offerings like Gemini. Industry observers have wondered for years whether Google would commercialize its TPUs for widespread usage beyond existing agreements with companies like Anthropic and Meta.

Google currently has agreements with two major outside companies for TPU access. One partnership with AI startup Anthropic provides the company access to approximately one million of Google's chips. The tech giant also has a separate arrangement with Meta for TPU usage.

While discussions of AI technology tend to focus on questions of supply and capability, the harder challenge remains demand — whether businesses can actually integrate these tools into their operations. Recent moves like Anthropic's small- to medium-sized business plugin for platforms including PayPal, Intuit, Canva, and Docusign suggest the next stage of AI adoption depends less on what technology companies can build and more on whether ordinary companies can sustain usage within everyday workflows.

The market question is no longer simply whether frontier models can perform remarkable tasks, but whether accountants, nurses, insurance adjusters, teachers, procurement managers, financial analysts and their peers across Main Street can integrate AI into the highly specialized workflows they understand better than any Silicon Valley engineer.