Anthropic has agreed to spend $100 billion for up to 5 gigawatts of compute from Amazon, a deal that highlights how the AI race is shifting toward industrial-scale infrastructure and long-term capacity planning.
Google used its April 22 Cloud Next event to introduce eighth-generation Tensor Processing Units designed for the next wave of enterprise AI agents, pairing them with new Gemini Enterprise tools and security products. The announcement highlights how the AI race is shifting from model launches toward the infrastructure needed to run them at scale.
CoreWeave said on April 10, 2026, that it had reached a multi-year agreement with Anthropic to supply cloud computing capacity, with service expected to come online later this year. The deal adds to a fast-moving wave of AI infrastructure contracts as model developers race to secure enough compute to train and run larger systems.
Amazon is investing another $5 billion in Anthropic and may add $20 billion more, a deal that turns one of AI’s most important rivalries into a long-horizon bet on compute, cloud capacity and model scale.
Anthropic has expanded its partnership with Google and Broadcom in a move that will add multiple gigawatts of TPU capacity beginning in 2027. The deal underscores how the AI industry’s infrastructure race is shifting
