AI’s biggest bottleneck is starting to look less like software and more like real estate with a power bill.
Google and Blackstone are creating a US-based TPU cloud company that will let enterprises rent specialized compute for training and running advanced AI models. The venture is backed by an initial $5 billion equity commitment from Blackstone and is expected to bring 500 MW of data center capacity online in 2027.
The deal shows how the AI infrastructure race is moving beyond traditional cloud services. For companies building larger AI workloads, access to chips now depends on whether enough power, cooling, networking, and data center space can arrive at the same time.
A new route to Google’s AI chips
Blackstone highlighted that the company will offer data center capacity, operations, networking, and Google Cloud TPUs through a compute-as-a-service model. Google will supply the hardware, including TPUs, along with software and services for the venture.
According to Google, Blackstone would create a new TPU cloud in partnership with Google, giving customers more choice and flexibility in how they access TPUs in the cloud.
“This joint venture with Blackstone helps meet growing demand for TPUs, which are optimized specifically for efficiency and performance in the AI era,” Thomas Kurian, CEO of Google Cloud, said in Blackstone’s official announcement.
“Together, we’re accelerating AI transformation and providing more options for organizations to access accelerated compute capability,” Kurian noted.
Blackstone named Benjamin Treynor Sloss, a Google executive with more than two decades of experience building and operating Google’s global infrastructure, as CEO of the new company.
AI compute demand keeps reshaping the cloud
The partnership comes as enterprises, AI labs, financial firms, and high-performance computing users compete for access to chips, data center capacity, power, cooling, and networking.
Google’s TPUs are custom-built chips built for AI training and inference. Blackstone noted that chips have been developed and deployed in production for more than a decade and power Gemini and other AI-driven Google products used by billions of people.
Reuters reported that the venture is designed to capitalize on strong demand for AI computing services. Blackstone has also stepped up investments in AI-related infrastructure, including data centers, power generation, and transmission assets.
“This isn’t the biggest headline number we’ve seen. But it’s a high-quality bet on sustainable growth in AI infrastructure,” Brittain Ladd, an AI and supply chain consultant at Chang Robotics, told Reuters.
For enterprise customers, the venture adds another way to access Google’s AI chips as demand for the computing power behind larger AI workloads grows.
Infrastructure becomes an AI strategy issue
The deal between Google and Blackstone also highlights how AI infrastructure is becoming a capital-intensive business problem, not just a cloud architecture decision. Companies running large AI workloads need compute, power, cooling, and operational expertise to align simultaneously.
Blackstone brings financing and data center infrastructure expertise, while Google offers TPUs and the cloud software stack around them. If the venture delivers capacity as planned in 2027, it could help organizations that want access to Google’s AI chips but need more deployment options than the standard public cloud route.
Read more about what Google is expected to reveal at I/O and how the reported Gemini update could shape the next stage of the AI model race.