Google to Invest Up to $2 Billion More in Anthropic

Google has agreed to invest up to $2 billion more in Anthropic, it emerged today, extending a financial relationship that goes back to 2022 and cementing the Mountain View giant's role as one of the leading backers of the AI lab founded by Dario and Daniela Amodei.

A partnership years in the making

This isn't Google's first move in Anthropic's story. Back in early 2023, it was reported that Google was prepared to put up to $300 million into the company in exchange for a stake of roughly 10%, building on a deal first struck in 2022. Today's figure — up to $2 billion — dwarfs that earlier commitment several times over and makes clear Google has no intention of sitting on the sidelines as Anthropic gains ground as an alternative to OpenAI.

Anthropic, founded in 2021 by former OpenAI research and safety leaders, has built a reputation distinct from its rival's. The company emphasizes what it calls "constitutional AI," a method for aligning model behavior with an explicit set of ethical principles rather than relying solely on human feedback. Its Claude family of models, with Claude 2 launched this past July, competes directly with GPT-4 on reasoning, summarization and text generation, though its user base still lags well behind ChatGPT's.

Not the only tech giant at the table

Google isn't the only heavyweight betting big on Anthropic. Last September, Amazon announced it would invest up to $4 billion in the company in exchange for a minority stake, a move aimed both at gaining exposure to the generative AI race and at ensuring Anthropic's models run preferentially on AWS infrastructure. Google's involvement, running in parallel, leaves Anthropic backed by two of the world's largest cloud infrastructure providers — an unusual position for an AI startup.

For Google, the calculus is twofold. On one hand, a stake in Anthropic gives it an indirect way to compete in the large language model race without relying exclusively on its own in-house development, led by DeepMind and the still-in-progress Gemini project. On the other, it pushes Anthropic toward using Google Cloud as its computing backbone, in a market where training each frontier model consumes staggering amounts of chips and electricity.

The shadow of Microsoft and OpenAI

All of this is playing out under the shadow of the Microsoft-OpenAI deal, which since 2019 has funneled billions of dollars to Sam Altman's company and made Azure the go-to infrastructure for ChatGPT and GPT-4. That arrangement forced every other major tech company to respond: Google, Amazon and Meta have all sought their own footholds in the generative AI landscape, whether by building their own models or by bankrolling others.

Google's move with Anthropic follows exactly that logic. If Microsoft has OpenAI, Google wants, at minimum, a solid position in Anthropic on top of its own internal efforts. The key difference from the Microsoft-OpenAI model is that Google hasn't pursued — at least not publicly — an exclusive integration of Claude into its products, opting instead for a relationship built more around infrastructure and financial stake.

What it means for the industry

With Amazon and Google both behind it financially, Anthropic finds itself in a position of strength unusual for a startup barely two years old. That gives it room to keep training ever-larger, ever-costlier models without depending on a single venture capital round — something most of its competitors simply can't afford.

For the rest of the ecosystem, the message is clear: the race for frontier models is no longer being fought solely between research labs, but between the bank accounts of the tech giants funding them. How much of that $2 billion actually reaches Anthropic's coffers, and under what conditions, is something that will likely become clearer in the months ahead.