SI Agent News
DevelopingClaudeOpenAIChipsBusiness

Anthropic's IPO filing reveals non-cancelable spending deals with tech giants

The AI startup's go-public documents show it is legally bound to pay billions to Alphabet, Nvidia, Microsoft, and Amazon regardless of actual usage.

$114.5 billionAnthropic's minimum spending commitment to Alphabet

The short version

Why it matters

For investors, these contracts provide a guaranteed revenue floor for big tech companies, even if the AI market cools down. For Anthropic, the IPO provides the necessary capital to fulfill these massive, non-cancelable financial obligations.

The cost of scaling AI

Anthropic is preparing to go public under the ticker symbol 'ANT' following a period of explosive growth. The company reported $16.3 billion in revenue for the six months ending June 30, which represents a 1,300% increase compared to the previous year. However, this rapid expansion came with significant costs, as the company posted a net loss of $80.7 billion during that same timeframe.

Despite these heavy losses, Anthropic is highlighting what it calls 'compute efficiency' to potential investors. In the three months ending June, the company noted that revenue grew by 1,368% while its infrastructure and compute expenses rose by 300%. This means that for every dollar earned, the company spent less on the underlying hardware and processing power than it did a year earlier.

Infrastructure remains one of the most significant expenses for the AI developer, accounting for roughly half of its total operating costs. During the first half of the year, Anthropic spent $8.9 billion on these essential resources. The company's upcoming public offering is expected to provide a large pool of new capital to help manage these ongoing and escalating bills.

Non-cancelable spending floors

The company's IPO filing reveals a unique set of multiyear, non-cancelable agreements with four of the largest technology companies in the U.S. These contracts commit Anthropic to spending billions of dollars on products and services from Alphabet, Nvidia, Microsoft, and Amazon. Because these deals are non-cancelable, Anthropic must pay the agreed amounts even if it chooses not to use the services provided.

For Alphabet, the specific terms are the most transparent in the filing. Anthropic has committed to spending at least $114.5 billion on Alphabet's services between April 2026 and July 2033. If Anthropic's actual spending falls below this threshold, it is legally required to pay Alphabet the difference. This structure creates a spending floor that protects Alphabet's revenue regardless of how much Anthropic actually scales its operations.

While the specific dollar amounts for the deals with Nvidia, Microsoft, and Amazon are not disclosed in the filing, they are described as being similar in structure. These agreements ensure that these tech giants stand to gain from Anthropic's growth. For shareholders of these four companies, these contracts offer a form of indirect exposure to Anthropic's success before the company officially begins trading its own shares.

Hardware and cloud ties

Anthropic relies on a diverse array of computing resources to train and run its models. The filing lists several routes for obtaining hardware, including multi-year agreements with SpaceXAI, Nscale, and Lambda. It also specifically mentions using Nvidia and AMD graphics processors alongside specialized chips like Amazon's AWS Trainium and Alphabet's TPUs.

The relationship with Nvidia is particularly deep, involving plans for joint engineering and design work. Anthropic intends to run and scale on Nvidia infrastructure, specifically using Grace Blackwell and Vera Rubin systems. A primary goal of this partnership is to tune Anthropic's models for better performance while adapting Nvidia's architectures to handle Anthropic's specific workloads.

Amazon's involvement is multifaceted because it serves as both a supplier and a shareholder. Anthropic buys a large amount of services from both Amazon's cloud computing unit and its chip operations. Additionally, Amazon owns nonvoting shares in Anthropic, though it does not have a seat on the company's board or any formal role in steering its corporate direction.

What comes next for ANT

The upcoming listing of Anthropic will be a significant test of the company's ability to manage its massive debt and spending obligations. Investors will be watching closely to see if the capital raised from the IPO is sufficient to cover the billions owed to its primary partners. The non-cancelable nature of these contracts shifts the risk of changing business plans entirely onto Anthropic.

If Anthropic's needs shrink or it decides to move its operations to different providers, it remains legally bound to pay its current partners. This makes the IPO a critical milestone for ensuring the company can meet its financial commitments. The success of the listing will determine how reliably those committed dollars will arrive for the tech giants that currently hold these contracts.

As the company moves toward its public debut, the focus will remain on its 'compute efficiency' and its ability to turn revenue into sustainable profit. While the market remains mixed for technology stocks, Anthropic's unique web of guaranteed contracts with the biggest names in the S&P 500 provides a distinct financial framework for its future.

What to watch

Try it

Look up the ticker symbols GOOGL, NVDA, MSFT, and AMZN to see how these companies might be indirectly linked to Anthropic's upcoming IPO.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by qz.com · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news