In its pursuit of dominance in artificial intelligence, Google has not only developed its own technology but also strategically invested in leading A.I. start-ups. To maintain its competitive advantage, however, the company has kept its ownership stakes in these start-ups under wraps.

Recently obtained court documents, reviewed by The New York Times, shed light on Google’s investment in Anthropic and reveal upcoming changes in the financial arrangement between the two companies.

Legal filings submitted as part of a Google antitrust case disclose that the tech giant holds a 14% stake in Anthropic. Despite this investment, Google exerts minimal influence over the start-up—it is limited to a maximum 15% ownership and lacks voting rights, board representation, or even board observer status.

Nonetheless, Google is set to inject an additional $750 million into Anthropic in September via a convertible debt agreement, a type of loan that can be converted into equity. This investment is part of a broader commitment made in 2023. In total, Google has poured over $3 billion into the company.

The legal filings offer a rare glimpse into how major tech firms are backing A.I. start-ups, a trend that has drawn regulatory scrutiny. Authorities have questioned whether such investments give established corporations an unfair edge in the rapidly evolving A.I. landscape. In addition to Google, both Amazon and Microsoft have backed prominent A.I. firms, including Anthropic and OpenAI—the company behind ChatGPT.

“Google understands that A.I. is a race, and with its vast financial resources, it can afford to place bets on multiple contenders,” said Chris V. Nicholson, a venture capitalist at Page One Ventures specializing in A.I. investments.

The court documents surfaced amid a landmark antitrust case against Google over its dominance in online search. In August, a federal court ruled that the company had violated antitrust laws by maintaining a monopoly in the search market.

In response, the Justice Department initially proposed forcing Google to divest from any A.I. ventures that could compete with its search business—including its stake in Anthropic. The start-up’s chatbot, Claude, could potentially serve as an alternative to traditional search engines.

However, last Friday, the Justice Department walked back its initial proposal, suggesting instead that Google be required to notify authorities before making future A.I. investments.

Anthropic submitted its filings before this shift in government stance. While the publicly available court records contain redacted sections, an unredacted version was provided to The Times, which has advocated for greater transparency in the case.

In these documents, Anthropic argued that forcing Google to sell its stake would be detrimental to both the company and competition as a whole. Such a move, the start-up’s lawyers contended, would devalue Anthropic and make it harder to secure funding.

Co-founder Tom Brown warned that severing ties with Google would cause “grievous harm” to the start-up, according to the filings.

Google declined to comment on the financial details of its investment but stated that it backs various companies to generate returns and expand the market. Anthropic also declined to comment.

Founded in 2021 by Dario and Daniela Amodei—former OpenAI executives who departed over disagreements regarding the company’s funding and partnership with Microsoft—Anthropic aims to develop A.I. with strong safety measures. It is structured as a public benefit corporation, meaning its mission includes broader social and ethical considerations.

(The New York Times has filed a lawsuit against OpenAI and Microsoft, alleging copyright infringement related to A.I.-generated content. Both companies have denied the allegations.)

Court filings indicate that Anthropic has taken measures to prevent any single tech giant from dominating its ownership. To date, the start-up has raised over $14.8 billion from venture capital firms, including Menlo Ventures.

Google invested more than $2 billion in Anthropic in 2023, while Amazon contributed $4 billion that year and another $4 billion in 2024.

Today, Google’s stake in Anthropic is reportedly worth significantly more than the start-up’s most recent funding round, which closed this month with a valuation of $61.5 billion.

To build its A.I. models, Anthropic purchases vast amounts of computing power from both Google and Amazon. As part of its agreements with investors, it has committed to using their cloud computing and chip services—essentially directing some of the capital it raises back to its own backers.