Blackstone and Google have entered into a joint venture to launch a new U.S.-based compute-as-a-service enterprise, breaking Google’s exclusive native control over its proprietary AI hardware. Blackstone is making an initial commitment of $5 billion in equity capital from its managed funds to build and operate the infrastructure footprint. The independent entity will deploy Google’s custom Tensor Processing Units (TPUs) outside standard Google Cloud environments, giving enterprise customers and elite AI labs an alternative channel to lease custom silicon for training and model inference workloads.
The physical construction timeline aims to bring a baseline of 500 megawatts of specialized data center capacity online by 2027, with blueprints to scale the power footprint over time. Under the partnership terms, Google will supply the physical TPU hardware processors alongside native software and management services, while alternative asset giant Blackstone leverages its $1.3 trillion capital base to secure energy grids and digital real estate assets. To manage the operational rollout, Blackstone appointed Google infrastructure veteran Benjamin Treynor Sloss to serve as Chief Executive Officer of the spin-out.
The deal allows Google to distribute the heavy capital expenditures required to expand physical data centers while capturing massive commercial demand for accelerated computing assets. The specialized chips currently drive the computational workflows behind Google’s Gemini models and broader global software products. “We see a generational opportunity to invest capital at scale building AI infrastructure,” stated Jon Gray, President and Chief Operating Officer of Blackstone. Thomas Kurian, CEO of Google Cloud, confirmed the infrastructure expansion, noting the entity helps “meet growing demand for TPUs, which are optimized specifically for efficiency and performance in the AI era.”