Anthropic
Safety-First Frontier AI, Constitutional AI, and Claude Reasoning Architectures
Datacenter & Compute Infrastructure
Amazon Web Services (AWS Trainium & Inferentia + EC2 P5 H100 clusters); Google Cloud TPU v5e/v5p environments.
Founded by former OpenAI research executives Dario and Daniela Amodei, Anthropic focuses on AI safety, mechanistic interpretability, and steerable models. Backed by Amazon and Google, it develops the Claude series (Claude 3.5 Sonnet, Claude Opus 5).