A consortium of 10 banks is arranging $22 billion in debt financing for Crux AI, a new cloud venture involving Blackstone and Alphabet, as the global AI investment cycle moves deeper into infrastructure. The financing demonstrates how the rapid expansion of AI computing demand is creating new funding structures around specialized chips, data centers and long-term customer contracts.

A $22 Billion Financing Package for AI Chips

The financing includes major lenders such as Goldman Sachs, Sumitomo Mitsui Banking Corp., Barclays, BNP Paribas and Bank of Nova Scotia. The proceeds will be used to purchase Google’s custom Tensor Processing Units, or TPUs, which are designed to handle AI workloads. The debt is expected to be secured by the value of the chips as well as Crux AI’s customer contracts, creating a financing structure directly linked to the venture’s underlying computing assets and contracted demand.

The banking group is also seeking additional lenders through syndication, allowing the participating institutions to distribute exposure to the transaction. A separate $1 billion revolving credit facility is also being arranged for the venture.

Blackstone and Alphabet Build a New AI Cloud Platform

Crux AI was created to address growing demand for dedicated AI computing capacity from AI laboratories, technology companies, enterprises and governments. Blackstone has committed an initial $5 billion in equity, while Alphabet is contributing its TPUs, software and services.

The venture plans to bring 500 megawatts of data-center capacity online in 2027, with further expansion expected. The scale of the planned infrastructure illustrates the amount of capital required to convert rising AI demand into physical computing capacity, particularly as companies compete for advanced processors, electricity and data-center space.

AI Infrastructure Moves Deeper Into Credit Markets

The transaction also reflects a broader shift in how the AI buildout is being financed. Earlier funding rounds were heavily concentrated in corporate capital expenditure and equity investment, but debt is increasingly being used to finance chips and servers alongside the data centers that house them.

The possibility of refinancing the Crux AI debt through longer-term institutional financing could further expand this model. If similar transactions become more common, AI hardware could increasingly become an identifiable financing asset, while customer contracts provide lenders with additional visibility into future cash flows. At the same time, lenders face risks tied to chip depreciation, technological changes, customer concentration and the pace at which AI demand translates into sustainable revenue.

The next stage will be the deployment of Crux AI’s computing capacity and the development of its customer base. Investors will also be watching whether the financing structure is replicated across the rapidly expanding AI infrastructure market, particularly as technology companies and infrastructure investors seek increasingly large pools of debt capital to fund the next generation of computing capacity.