Google Plans Up to $40 Billion Anthropic Investment, Secures AI Compute Capacity

Alphabet’s Google is deepening its bet on Anthropic with a deal that could total up to $40 billion, underscoring how the AI arms race is increasingly being fought not only with models, but with capital and computing power.

Bloomberg reported on Friday that Google has agreed to invest an initial $10 billion in Anthropic at a $350 billion valuation, with the option to invest an additional $30 billion if specified performance milestones are met. The agreement also includes Google Cloud supplying Anthropic with up to 5 gigawatts of computing capacity over the next five years, with room to expand depending on demand.

Anthropic, founded in 2021 by siblings Dario Amodei and Daniela Amodei—both former OpenAI executives—develops and operates the Claude family of AI assistants. The new investment package builds on a partnership arrangement signed earlier this month involving Anthropic, Google, and Broadcom ($AVGO), reflecting growing coordination across AI software, cloud infrastructure, and specialized hardware.

For Google, the structure of the deal signals a two-pronged push: capture more AI-driven cloud workloads while widening deployment for its in-house AI accelerators, the 'TPU' (Tensor Processing Unit) chips. Google this week unveiled two new TPU variants aimed at what it describes as the emerging era of AI agents—TPU 8t for training and TPU 8i for inference—moves widely seen as a direct attempt to erode Nvidia’s ($NVDA) dominance in foundational AI compute.

The size and design of the Anthropic agreement also highlight the strategic value of compute supply at a time when leading model developers require vast, stable access to high-end infrastructure. A 5GW commitment is extraordinary by cloud standards, reflecting both the scale of next-generation AI systems and the competitive leverage cloud providers gain by locking in marquee customers with long-term capacity guarantees.

Google is not the only hyperscaler using capital to tighten ties with top AI labs. Amazon has previously invested $5 billion into Anthropic at the same $350 billion valuation and secured an option to invest up to an additional $20 billion, reinforcing a pattern in which major cloud players compete to anchor fast-growing AI companies to their platforms.

At the same time, the deal is likely to intensify criticism around what some industry observers describe as a 'circular financing' dynamic: cloud providers invest in AI startups, and those startups then spend heavily—often on the same provider’s infrastructure—effectively recycling capital into cloud revenue. Supporters argue the arrangements accelerate model development and reduce supply uncertainty; detractors warn they can blur competitive lines and inflate valuations in a market already characterized by runaway spending.

With rival ecosystems coalescing around different combinations of models, chips, and cloud capacity, Google’s expanded commitment to Anthropic signals that AI competition is moving beyond algorithmic breakthroughs into a contest for 'liquidity inflow' and long-term compute entitlements—resources that could determine which platforms dominate the next phase of enterprise and consumer AI adoption.