Anthropic Signs Compute Deal with Google and Broadcom as Run-Rate Revenue Exceeds $30B

Anthropic has officially signed a massive compute deal with Google and Broadcom to create multiple gigawatts of AI processing power using TPUs. This will allow the company to power its future Claude language models. The company says that the processing facilities should be online by 2027.

The push for more AI compute is getting serious

The announcement follows a significant increase in the demand for Anthropic's Claude models in 2026. Notably, the company also announced that it has surpassed $30 billion run-rate revenue mark, almost tripling the $9 billion reported in 2025. Additionally, the company saw a doubling of the number of organizations paying them over $1 million annually, climbing to more than 1,000 in the last few months.

This new agreement builds on Anthropic's earlier partnerships, especially with Google Cloud and Broadcom. Most of the new infrastructure will be based in the United States, expanding its earlier $50 billion commitment toward domestic compute investments.

Anthropic is not relying on a single ecosystem either. It continues to train and run Claude across multiple platforms, including AWS Trainium, Google TPUs, and NVIDIA GPUs. This multi-platform approach likely helps balance performance, cost, and reliability depending on workload needs.

Speaking of competition, this move highlights how aggressively AI companies are locking in compute supply. Google strengthens its TPU position here, while Broadcom benefits from deeper infrastructure integration. At the same time, Amazon remains Anthropic's primary cloud partner through AWS and Project Rainier.

Claude also stands out as one of the few frontier models available across all major cloud platforms, including AWS Bedrock, Google Vertex AI, and Microsoft Azure. All that said, it's still early to tell how this capacity will translate into real-world performance gains. The 2027 timeline suggests a longer runway, and demand may shift further by then.