Date: August 4, 2026 Main source: Financial Times Link: https://www.ft.com/content/549f2e23-5aa2-49c7-9ea6-a9784ab7087c
Summary
Google has structured a financing deal of close to $200 billion to supply TPUs and data center capacity to Anthropic. The scheme combines chip purchases, private credit vehicles, lease contracts and energy projects backed by players such as Broadcom, Morgan Stanley, Apollo and Blackstone. Much of the risk stays off Google's direct balance sheet, although the structure depends on Anthropic being able to keep up its long-term payment commitments. The deal confirms that the race for frontier models is also a race for capital, energy and secured access to compute.
Why it matters
This deal turns AI infrastructure into a new financial asset class. It also strengthens Google's position against Nvidia, because it can bundle TPUs, cloud, financing and power capacity into a single offer. The risk is concentration: an important part of the structure depends on Anthropic's future growth and on inference demand justifying investments of historic scale.
Practical implications
- Companies should evaluate AI vendors by financial stability, capacity and contracts, not only by model quality.
- Agencies and consultants should avoid architectures that depend on a single provider or model.
- Projects with heavy inference use need budgets that allow for changes in price and availability.