What’s most surprising is that Google is backing a direct competitor to its own Gemini AI. Claude has been fiercely competing with Gemini in the chatbot market, and Bloomberg reports that even within Google, there are growing concerns about falling behind Anthropic in certain areas.

Internal Concerns

One major source of anxiety is Claude Code, Anthropic’s AI programming tool that has quickly become the go-to choice for developers. The rise of “vibe coding”—where AI handles most of the software-building process—has seen even Google employees turning to Claude Code over Gemini-based internal tools.

Rather than fighting head-on, Google is pursuing a different strategy. By positioning itself as Anthropic’s primary infrastructure provider, Google ensures that whether Claude or Gemini wins the AI race, it profits from the computing backbone behind them.

The AI race is increasingly defined by access to the massive computing power required to train and deploy these systems. OpenAI has already secured hundreds of billions in deals with cloud providers, chipmakers, and energy suppliers. Anthropic is following suit, racing to lock down similar resources.

In recent weeks, Anthropic faced backlash over unusually low usage limits for Claude. To address this, the company has rapidly expanded infrastructure agreements. Its revenue has skyrocketed from $9 billion at the end of last year to over $30 billion this month, putting immense pressure on its systems.

Earlier this month, Anthropic signed a deal with cloud provider CoreWeave for data center capacity. This week, it secured a $25 billion investment from Amazon, part of a broader plan to spend up to $100 billion to secure roughly 5 gigawatts of computing power over time.

A Complex Relationship

Despite being rivals in AI models, Google is also Anthropic’s key infrastructure supplier. Anthropic relies heavily on Google Cloud for chips and computing, including access to Google’s tensor processing units (TPUs)—specialized AI chips considered among the best alternatives to NVIDIA’s sought-after processors.

Earlier this month, Anthropic announced a partnership with Google and chipmaker Broadcom to access several gigawatts of TPU-based computing starting in 2027. Broadcom later disclosed the figure as 3.5 gigawatts. Google’s new investment expands that deal, committing to provide 5 gigawatts over the next five years, with potential for further growth.

The agreement also guarantees Google continued access to Anthropic’s most powerful model, Mythos. Anthropic describes Mythos as its strongest system yet, with critical cybersecurity applications. Due to potential misuse, access has been restricted to select organizations for evaluation, though leaks have already placed the model in unauthorized hands.

Anthropic’s valuation stood at $350 billion in February, but investors are now eager to back the company at $800 billion or higher, according to Bloomberg. The startup is reportedly considering an IPO as early as October.

Google’s $40 billion bet signals its commitment to the long game. Even as Gemini and Claude battle for user attention, Google wins if either succeeds—so long as they run on its servers. For Anthropic, founded just a few years ago by former OpenAI executives, the support provides crucial breathing room to challenge the biggest names in tech.