Google, Blackstone Eye $5 Bn AI Venture to Take on Nvidia

Google and Blackstone plan to launch an artificial-intelligence cloud company using Google’s specialised chips known as Tensor Processing Units (TPUs), the Wall Street Journal reported on May 18.

How Will the Venture Work

According to reports, Blackstone will invest $5bn initially, with an aim to bring 500 megawatts of data centre capacity online by 2027, with plans to expand further over time. Google will supply hardware, including TPUs, as well as software and services to the new company so it can rapidly accelerate to meet the growing demand for accelerated computing, leveraging the benefit of Google’s technical and domain expertise.

The company reportedly stated that the new US-based firm will offer “compute-as-a-service”, allowing businesses to rent AI computing power using Google’s TPU chips instead of building their own expensive infrastructure.

Blackstone is one of the world’s biggest alternative asset manager, with over $1.3trn in assets under management, and the largest global provider of data centers. The joint venture between Blackstone and Google is intended to give customers even more choice and flexibility for running their AI workloads on TPUs.

What Is TPU Cloud?

TPUs are AI-focused chips developed by Google for training and running advanced AI models. They support Google AI products including its AI-assistant, Gemini. These could act as an alternative to Nvidia’s GPU that dominates the AI chip market at present.

 Google has previously offered TPUs mainly through Google Cloud, but the new venture will make them available through a separate AI cloud platform as well.

“The joint venture between Blackstone and Google is intended to give customers even more choice and flexibility for running their AI workloads on TPUs,” Blackstone said in a statement.

AI’s Growing Power Crunch

Meanwhile, AI-driven data centres are causing rapid growth in the global demand for electricity.

 According to a report from S&P Global, the US data centre annual power demand growth rose to 19% in 2024, from 8% in 2022. The report also estimated that the cost of building planned AI data centre capacity could reach nearly $200bn per year globally.

The Google-Blackstone venture is a sign of how quickly AI demand is driving data center expansion, led by big tech and large investors despite ongoing power and infrastructure constraints. Its long-term success will depend on a balancing act between aggressive build-out, efficiency gains and disciplined investment to avoid overcapacity as AI-driven energy needs continue to rise.