Anthropic commits $200 billion to Google for cloud and AI chips
Updated
Updated · The Information · May 5
Anthropic commits $200 billion to Google for cloud and AI chips
10 articles · Updated · The Information · May 5
The five-year deal starts next year and follows Google’s pledge last month to provide five gigawatts of server capacity, equal to more than 40% of Google Cloud’s disclosed revenue backlog.
The report says Anthropic and OpenAI now account for about half of the $2 trillion backlog across Amazon, Microsoft, Google and Oracle, underscoring how heavily major cloud providers depend on the two AI startups.
Google may benefit from higher margins by using its in-house AI chips, while investors remain wary that huge projected spending by Anthropic and OpenAI may not fully materialise.
As AI models become multi-cloud, are tech giants just funding their own disruption by creating platform-agnostic behemoths?
With trillions in AI spending committed, are we witnessing a tech revolution or the inflation of a massive economic bubble?
As AI's power demand rivals entire nations, is the global energy grid heading for a catastrophic failure?
Anthropic's $200 Billion Compute Deal with Google and Broadcom: Securing 3.5 Gigawatts of TPU Capacity for AI Dominance
Overview
In May 2026, Anthropic announced a landmark $200 billion compute deal with Google and Broadcom to add 3.5 gigawatts of TPU capacity starting in 2027, driven by surging demand as its revenue soared to $30 billion and enterprise customers doubled. Google committed $40 billion, and Anthropic invested $50 billion in U.S. infrastructure, while Amazon added $5 billion linked to AWS. Facing GPU shortages and rising costs, Anthropic chose Google's efficient TPUs, enabled by Broadcom's custom silicon expertise, to overcome bottlenecks. This deal narrows the compute gap with rivals, supports a multi-cloud strategy for resilience, and signals a major industry shift toward large-scale, specialized AI infrastructure amid growing regulatory scrutiny.