Blackstone is teaming up with Google to launch a new AI infrastructure company to challenge Nvidia.
The new company will provide cloud services, data center capacity, operations, networking and Google Cloud’s Tensor Processing Units (TPUs) as a compute-as-a-service offering.
In many respects, it will rival CoreWeave, the neocloud vendor that Nvidia has invested heavily in to build out infrastructure.
Blackstone, a private asset management firm that is also one of the world’s biggest data center owners, said in a statement on Tuesday that the joint venture aims to bring 500 megawatts of data capacity online next year and would back it with $5 billion. Beyond 2027, the asset management giant says there are “plans to scale significantly over time.” Blackstone will hold a majority stake.
Google’s TPUs are purpose-built chips that are optimized for training and inference of AI models. Although they have been around for a number of years, they have not been able to match the penetration of Nvidia’s GPUs, which have dominated the AI market.
However, the search giant has redoubled its efforts of late to expand the reach of TPUs and reduce the industry’s reliance on GPUs, and the new joint venture will give it further opportunity to do so, following recent deals with Anthropic, Broadcom and Meta.
Alongside the TPUs, Google will deliver software and services as demand for AI continues to surge.
While the new company still doesn’t have a name, it has been confirmed who will be in charge, with Benjamin Treynor Sloss, a former Google executive with extensive experience operating global infrastructure, appointed as CEO.
For Blackstone, the new company is yet another foray into AI after chief executive Stephen Schwarzman claimed in April that his firm was “the largest investor in AI-related infrastructure in the world”.
“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,” Blackstone COO Jon Gray said in the statement.
