Google has moved billions of dollars in chip risk off its balance sheet by creating a complex financing deal with Broadcom and Morgan Stanley to fund $35 billion of its proprietary AI hardware for startup Anthropic.
The agreement relies on a special-purpose vehicle to purchase the chips using outside investor capital. Broadcom acts as the guarantor while Anthropic leases the equipment. This structure removes the hardware from the books of all parties involved.
Avoiding balance sheet strain
Anthropic requires vast amounts of computing power but lacks a credit rating. Banks will not lend the startup the necessary funds. Google, Broadcom, and the other involved firms all wish to keep the expensive assets off their own financial statements.
Google is currently spending record sums and does not want added strain on its books. Broadcom similarly wants to avoid tying up capital in the Google chips it resells.
Morgan Stanley established the financing vehicle that buys the chips and leases them to Anthropic. Apollo and Blackstone provided the money. A special purpose vehicle called Compute SPV will buy about one gigawatt of TPU hardware for $35 billion starting in June. That equates to roughly one million TPUs.
Broadercom provides a backstop covering about $30 billion of the purchase if Anthropic stops making lease payments.
The model now serves as a template for further agreements. The largest deal so far is an April agreement covering Google’s sale of another 3.5 gigawatts of TPU hardware to Broadcom for Anthropic. Broadcom’s financial filings list $128 billion in purchase commitments through 2028. Sources say nearly all of that amount is tied to Google TPUs.
Securing power from crypto miners
Financing the chips solves only half the problem. Google also needs data centers with enough power to run them. The company is turning to crypto miners that have already secured access to large amounts of electricity.
TeraWulf is the first to receive a Google guarantee, which covers a 360 megawatt data center in New York. Morgan Stanley packages the guarantee into a $3.2 billion construction bond. Google receives an ownership stake in TeraWulf in return.
The same model extends to other crypto miners, including Cipher Digital and Hut 8. Google has backed ten projects with a combined capacity of 2.4 gigawatts so far.
Google could face up to $44 billion in obligations if every lease defaults. Yet the company records only $815 million of that liability on its balance sheet, leaving most of the exposure off its books.
Lower costs and higher exposure
Google’s financial backing is changing the economics of AI infrastructure. Data center projects backed by Google borrow at a median interest rate of 7.1 percent. That compares with 9.3 percent for neocloud operators that rely on Nvidia chips.
Jefferies analysts call the gap a structural cost-of-capital disadvantage for companies in Nvidia’s ecosystem.
The arrangement carries substantial risk because $200 billion in contracts depend on Anthropic continuing to make its lease payments. Google sits on both sides of the deal as an investor in Anthropic and the supplier of its chips.
An earlier report from The Information says Anthropic has committed to spending about $200 billion on Google Cloud over five years in exchange for five gigawatts of server capacity. That deal accounts for more than 40 percent of Google’s committed future cloud revenue.
Together with OpenAI, Anthropic accounts for roughly half of the $2 trillion in cloud backlogs at Amazon, Microsoft, Google, and Oracle. Both startups are counting on their revenue growing 20 to 30 times by 2029. The entire structure could unravel if that growth slows or stalls.
What it means
For people building AI systems, this means the cost of access to Google’s Tensor Processing Units is dropping compared to using Nvidia hardware. The risk of not getting the chips falls on the investors and the startup, not the chip maker. However, the entire ecosystem now hinges on Anthropic’s ability to generate enough revenue to keep paying for the leases. If growth stops, the financing model breaks.
