Google and Anthropic in talks for multi-billion dollar cloud deal (finance.yahoo.com)

Anthropic is in early talks with Google for a cloud-computing arrangement reportedly worth tens of billions of dollars, a deal that would scale the AI startup’s access to Google Cloud resources. The conversations — not finalized and subject to change — follow Google’s roughly $3 billion in prior investments in Anthropic and come amid competition from Amazon, which has also funded Anthropic and supplied AWS compute. Market reaction was immediate: Alphabet shares rose ~3.1% after-hours, while Amazon dipped amid the news and a recent 15-hour AWS outage affecting many customers. For the AI/ML community, the potential deal matters because it signals where large LLM providers may anchor their infrastructure: preferential access to GPUs/TPUs, cloud-optimized model-serving stacks, and committed capacity can materially reduce training/inference latency and costs, accelerate model iteration, and shape commercial deployment strategies. Concentrating Anthropic’s compute with Google could give it operational advantages versus rivals (e.g., tighter integration, performance tuning, and security tooling), but also raises questions about vendor lock-in, resilience (single-provider outages), and competitive dynamics in cloud and model markets. Since both companies declined comment and terms aren’t set, the key technical implications — hardware types, pricing, SLAs, and multi-cloud strategies — remain uncertain but consequential.