Google (Alphabet) and Blackstone Group unveiled plans to form a United States-based joint venture focused on artificial intelligence cloud computing. The new company will be majority-owned by Blackstone, which is investing $5 billion in equity capital [1, 2, 3, 4, 5, 6, 7].
Blackstone, the world's largest private data center owner with over $1.3 trillion in assets under management, has been expanding its AI infrastructure investments. Jon Gray, Blackstone’s president and COO, said, "This new company has enormous potential as it helps to meet the unprecedented demand for compute" [2, 3, 4, 6].
Google will contribute specialized hardware, including its Tensor Processing Units (TPUs), along with software and services to the venture. These TPUs are custom AI chips designed to compete with Nvidia GPUs. Thomas Kurian, CEO of Google Cloud, noted the partnership "will help meet rising demand for TPUs and provide enterprises more AI acceleration computing options" [1, 8, 3, 4, 5, 6, 7].
Benjamin Treynor Sloss, a senior Google executive known for leading Google’s infrastructure efforts, will serve as the CEO of the new company [1, 2, 3, 4, 5, 6, 7].
The joint venture plans to bring 500 megawatts of AI computing capacity online by 2027 with ambitions to significantly expand beyond that initial target. The total investment, including leverage, could reach $25 billion [2, 3, 9, 4, 10, 6, 7].
This effort positions the new company to compete with cloud AI infrastructure providers like CoreWeave and Nebius Group, which currently depend primarily on Nvidia GPUs [8, 2, 3, 5, 6, 7].
Alphabet plans to spend up to $190 billion on AI infrastructure capital expenditures in 2026, accelerating its buildout of data centers and AI chip capabilities. CEO Sundar Pichai said, "Supporting our products at global scale and serving worldwide businesses and developers requires large-scale infrastructure investment, and custom chips are a key part of that" [11].
The joint venture was first reported by the Wall Street Journal on May 18 and officially announced the next day [1, 12, 2, 4, 5, 6]. The company aims to have a substantial AI compute footprint operating by 2027.
