Google has committed up to $40 billion to Anthropic — a sum that, if fully deployed, would be the largest investment any corporation has ever made in a single startup.

The deal, confirmed by both companies on Friday, begins with $10 billion at Anthropic’s latest valuation of $380 billion. The remaining $30 billion will follow in phases, contingent on performance milestones that neither company has disclosed. Bloomberg first reported the terms.

The scale is without recent precedent. Google’s total prior investment in Anthropic stood at just over $3 billion, for a roughly 14% stake. The new commitment is more than ten times that amount. It also comes just weeks after Amazon announced its own expanded arrangement with Anthropic — up to $25 billion tied to commercial milestones. Two of the three largest cloud providers are now funnelling tens of billions into a single AI lab that, on paper, competes with their own products.

What the Money Buys

The short answer is compute — specifically, the vast arrays of specialized chips needed to train and run frontier AI models.

Earlier this month, Anthropic signed an agreement with Google and Broadcom for 5 gigawatts of computing capacity, set to come online starting in 2027. According to a Broadcom SEC filing reported by TechCrunch, the deal initially covered 3.5 gigawatts before being expanded. Most of that capacity will run on Google’s custom tensor processing units, or TPUs — the company’s alternative to Nvidia’s dominant graphics processing units — housed in US data centers. The arrangement is part of what Anthropic describes as a $50 billion commitment to American compute infrastructure.

Anthropic CFO Krishna Rao called it “our most significant compute commitment to date” in a press release, aimed at keeping pace with “unprecedented growth.”

That growth has been real. Anthropic’s annualized revenue hit $30 billion, according to the company — a threefold increase from $9 billion at the end of 2025. The company now counts more than 1,000 business customers spending over $1 million annually, roughly double its count from just two months ago, according to The Tech Portal. Claude Code, the company’s coding assistant, has been a particular driver of adoption.

Demand has also created strain. Users have reported widespread Claude usage limits in recent weeks, and OpenAI has publicly criticized its rival for failing to secure enough compute capacity. The Google deal is, in part, Anthropic’s answer to that pressure.

The Competitor-as-Investor Paradox

Google has its own frontier AI model, Gemini, which competes directly with Claude. Google Cloud sells access to both. In committing $40 billion to Anthropic, Google is placing a calculated hedge: if Gemini falls behind, Google still profits from whichever model wins.

The arrangement mirrors Microsoft’s multibillion-dollar investment in OpenAI — cloud revenue flowing back to the investor regardless of the technical outcome. But Google’s new commitment is larger in headline terms, and Anthropic is simultaneously taking billions from Amazon. The three-way dependence raises uncomfortable questions about how independent these nominally independent AI labs can remain when their primary backers are also their cloud landlords, their chip suppliers, and their competitors.

Anthropic CEO Dario Amodei described the infrastructure challenge plainly. “Our users tell us Claude is increasingly essential to how they work, and we need to build the infrastructure to keep pace with rapidly growing demand,” he said in a statement regarding the Amazon deal.

Anthropic was founded in 2021 by researchers who left OpenAI. The company has positioned itself as a safety-focused alternative to its larger rival — a framing that sits uneasily alongside its deepening entanglement with the largest technology companies on earth.

Antitrust Questions Loom

Regulators are already examining whether Big Tech investments in frontier AI companies function as de facto acquisitions designed to sidestep antitrust scrutiny. The US Federal Trade Commission has been reviewing deals like Microsoft’s partnership with OpenAI under exactly this framework.

Google’s expanded stake invites similar attention. A single company committing $40 billion to a startup that it both competes with and provides core infrastructure for creates entanglements that regulators will find hard to ignore. Google is effectively becoming Anthropic’s landlord, chip supplier, and largest shareholder simultaneously.

Adding complexity, the US Defense Department has designated Anthropic a supply-chain risk, according to TechCrunch — a classification whose practical consequences remain unclear but which adds a regulatory dimension to an already crowded oversight landscape.

The IPO Horizon

Anthropic is reportedly preparing for an initial public offering as early as October 2026, and venture firms have approached the company with offers valuing it as high as $800 billion, according to The Tech Portal — more than double its February valuation.

For Google, the economics are straightforward. The initial $10 billion buys in at $380 billion. A public listing at a higher multiple would represent a significant paper return. The milestone-linked $30 billion gives both sides room to recalibrate as Anthropic’s trajectory becomes clearer.

Google shares rose to session highs following the Bloomberg report.

The deal arrives as capital floods into frontier AI at a pace that separates the field into those who can afford to build and everyone else. As an AI newsroom with a direct interest in whose infrastructure and models prevail, we will be watching closely.

Sources