Behind Blackstone's US$5bn Bet on a Google Cloud AI Company

Blackstone and Google have announced a partnership to build AI infrastructure in the US, marking a collaboration between one of the world's largest asset managers and a hyperscale cloud provider.

The partnership will create a standalone business focused on AI compute services built around Google Cloud's tensor processing units (TPUs).

Blackstone is committing US$5bn in initial equity through funds it manages. The first 500MW of capacity could come online in 2027.

Google will provide hardware, software and operational support for the platform.

Compute-as-a-service for AI workloads

The venture will offer infrastructure specifically designed for organisations deploying AI and high-performance computing applications, in a compute-as-a-service model. 

Customers will access Google Cloud TPUs outside the standard Google Cloud environment. This creates an alternative route to accelerated compute resources.

TPUs are custom chips designed by Google for both model training and inference workloads. Inference is the process of running trained models to generate outputs.

According to Google, TPUs already power workloads for AI laboratories, capital markets firms and high-performance computing users. The chips also support Gemini and other Google AI products.

Financial scale meets AI infrastructure

Blackstone manages more than US$1.3tn in assets and holds a strong position in digital infrastructure and data centres.

Jon Gray, President and COO of Blackstone, says: "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.

"We are incredibly proud to partner with Google – bringing together their world-class TPUs and AI capabilities with Blackstone's exceptional strength in energy and digital infrastructure."

The structure creates a separate company rather than a traditional cloud deployment model. Services will include data centre operations, networking and TPU access packaged together.

Blackstone says the arrangement could give enterprises more flexibility when deploying AI workloads that require accelerated computing.

Expanding access to TPUs

From Google's vantage point, the deal expands TPU availability at a time when competition for AI infrastructure intensifies across hyperscale providers.

Although TPUs have operated inside Google infrastructure for more than 10 years, this model creates another channel for enterprise access outside Google's own systems.

Thomas Kurian, CEO of Google Cloud, says: "This joint venture with Blackstone helps meet growing demand for TPUs, which are optimised specifically for efficiency and performance in the AI era.

"Together, we're accelerating AI transformation and providing more options for organisations to access accelerated compute capability."

The partnership blends Google's AI hardware and software with Blackstone's experience in financing and scaling large infrastructure assets.

This combination could matter as AI deployments move from experimental phases towards long-term operational use.

The structure also shows how financial firms continue to deepen involvement in digital infrastructure ownership.

As data centres become more focused on AI workloads, developers face larger upfront costs for land, power procurement, networking and cooling systems.

Jas Khaira, Head of Blackstone N1 (BXN1), says: "Capital alone doesn't build category-defining platforms – the right partner, the right structure and the conviction to underwrite singular opportunities do.

"Google's TPUs, a decade in the making and foundational to the AI economy, are exactly the kind of platform BXN1 was built to back."

Google veteran leads venture

Benjamin Treynor Sloss has been appointed CEO of the new company.

Benjamin spent more than 20 years building and operating Google's global infrastructure and operations.

His appointment could signal the operational scale expected for the platform as capacity increases over coming years.

Over time, the new business plans to scale beyond the initial 500MW deployment, although the partners have not yet disclosed total future capacity targets or specific data centre locations.

The venture demonstrates how AI infrastructure increasingly depends on partnerships between cloud technology providers, capital firms and data centre operators.

Access to power, advanced chips and large-scale facilities now sits at the centre of competition for enterprise AI workloads.

The project also shows the growing importance of dedicated AI data centre environments rather than multi-purpose colocation deployments.

As accelerated compute demand continues to rise, operators are facing increasing pressure to deliver facilities capable of supporting power-intensive AI systems, making this a welcome news for innovative, compute-intensive AI companies.