Anthropic’s Compute Capacity Doubles: 1,000+ Customers Spend $1M+

Anthropic is expanding its artificial intelligence capabilities with a new agreement for multiple gigawatts of computing power from Google and Broadcom, anticipating the infrastructure will be operational starting in 2027. This expansion is fueled by accelerating demand for its Claude models, with the company’s annualized run-rate revenue now exceeding 30 billion, a substantial increase from approximately 9 billion at the close of 2025. Anthropic reports its customer base of businesses spending over 1 million annually has doubled in under two months, now surpassing 1,000 clients. “This partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure,” said Krishna Rao, CFO of Anthropic. “We are building the capacity necessary to serve the growth we have seen in our customer base while also enabling Claude to advance AI development.”

This commitment builds upon previously announced increases to TPU capacity with Google Cloud last October and signifies a strategic approach to infrastructure scaling, designed to meet accelerating demand for its Claude models. The company reports a doubling of high-value customers in just under two months; over 1,000 businesses now spend more than 1 million annually on Claude services.

Anthropic utilizes a diverse range of AI hardware, including AWS Trainium, Google TPUs, and NVIDIA GPUs, to optimize performance and resilience for its customers, while also maintaining Claude as the only frontier AI model accessible across Amazon Web Services, Google Cloud, and Microsoft Azure.

This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure: we are building the capacity necessary to serve the exponential growth we have seen in our customer base while also enabling Claude to define the frontier of AI development.

Krishna Rao, CFO of Anthropic
Dr. Donovan

Dr. Donovan

Dr. Donovan is a futurist and technology writer covering the quantum revolution. Where classical computers manipulate bits that are either on or off, quantum machines exploit superposition and entanglement to process information in ways that classical physics cannot. Dr. Donovan tracks the full quantum landscape: fault-tolerant computing, photonic and superconducting architectures, post-quantum cryptography, and the geopolitical race between nations and corporations to achieve quantum advantage. The decisions being made now, in research labs and government offices around the world, will determine who controls the most powerful computers ever built.

Latest Posts by Dr. Donovan:

QCNNs Classically Simulable Up To 1024 Qubits

QCNNs Classically Simulable Up To 1024 Qubits

April 7, 2026
Bell states representing maximally entangled quantum bit pairs

Bell Nonlocality Connected To Integrable Quantum Systems

April 7, 2026
Quantum computing processor with qubits represented on the Bloch sphere

Inspira Targets Connectivity Bottleneck in Quantum Systems with 3D Architecture

April 7, 2026