Google Guarantees $15 Billion for Anthropic Debt, Takes Up to 20% Stake in Data Center

Google has agreed to provide a debt guarantee for Anthropic's $15 billion financing round. In the deal, led by Morgan Stanley, Google secures a 20% stake in a Texas data center in exchange for providing the guarantee, and supplies the AI chips that will go inside it with its own TPUs.

This isn't lending money — it's guaranteeing the borrower's debt. Anthropic raises capital on terms it couldn't have secured on its own credit alone, while Google, without putting up a single dollar, locks in long-term demand for its own chips.

One deal, three threads tied together

Thread	What Google gives	What Google gets
Financing	Debt guarantee on $15 billion raise	Deal execution / effective reduction in borrowing costs
Real estate	Guarantee risk exposure	20% stake in Texas data center
Semiconductors	TPU supply volume	Locked-in long-term demand for its own chips

The details of the guarantee are specific. If Anthropic fails to meet its debt obligations, Google will cover the data center's rent and power costs on its behalf. The fact that power costs are written into the contract alongside rent shows how large a share electricity now occupies in the cost structure of AI data centers.

This financing round is separate from the $35 billion chip purchase agreement signed earlier this year. Combined, the two deals represent a much larger commitment of Google's stake in Anthropic's infrastructure. Around the same time, a structure in which Broadcom shares chip costs was also reported.

Why this kind of structure emerged

As AI infrastructure investment scales up, financing itself becomes the bottleneck. A single data center facility can cost trillions of won, with returns not materializing for years. Model companies are largely unprofitable, making it difficult for them to pass traditional loan underwriting.

When a chip company provides the guarantee, three problems get solved at once.

1. The model company raises capital on terms it couldn't obtain on its own credit.
2. The chip company locks in demand for its own products through contract, without spending cash.
3. Lenders treat the debt as backed by a major corporate guarantee, lowering perceived risk.

But the risk doesn't disappear — it concentrates in one place. If Anthropic falters, Google absorbs the rent and electricity bills. Taking the 20% stake alongside the guarantee looks very much like securing compensation for that exposure in advance.

What's happening outside NVIDIA's orbit

The AI infrastructure narrative has largely been summarized as a race to secure GPUs. This deal shows a different axis: a company with its own accelerator using financing as leverage to manufacture demand for its own chips. Instead of simply selling chips, Google is becoming the financier behind the building that will house them.

TPU is Google's own AI-specific accelerator, designed in-house. It has historically been used more within Google's own cloud than sold externally, but that changes once a major model company takes on volume through a long-term contract. It's a structure that fills the gap created by criticism that TPU's ecosystem is weaker than general-purpose GPUs — filled in through contract instead.

The degree of vertical integration, in table form

Layer	Means of integration	Appearance in this deal
Semiconductors	Proprietary accelerator	TPU supply
Power/real estate	Data center stake	20% in Texas
Financing	Debt guarantee	$15 billion
Demand	Long-term purchase agreement	Separate $35 billion

It's unusual for a single company's name to appear across all four layers.

What Korean practitioners should watch

First, dependency structures are deepening. When adopting AI services, it's no longer enough to look only at the model provider. Which cloud and which accelerator the model runs on, and the terms binding that contract, directly affect pricing and availability. It's worth checking whether contracts include notification clauses for changes in model, region, or accelerator.

Second, power is the bottleneck. It's no coincidence that power costs are included in the guarantee terms. For anyone evaluating a data center project in Korea, grid connection timing and rate structures will determine half of the project's viability.

Third, memory demand moves in tandem. Anthropic CEO Dario Amodei recently noted that Korean memory manufacturers have already made substantial investments. As this pipeline thickens, the volume flowing into Korea's semiconductor and data center industries will move with it.

What remains unconfirmed

Whether the financing has fully closed, the composition of the lending syndicate, and the precise trigger conditions of the guarantee all fall outside the scope of what has been reported. The size of the data center is also described differently across reports. This article reflects only what has been confirmed.

Source: Compiled from Maeil Business Newspaper and other domestic reports.