Anthropic is an AI safety company that builds Claude - one of the most capable AI assistants globally. Founded by ex-OpenAI researchers Dario and Daniela Amodei, it has raised $8B from Amazon and $2B from Google. Its Constitutional AI approach prioritizes safety over raw capability. Amazon uses Claude across Alexa and AWS services. Anthropic's interpretability research is among the most cited in academic AI safety circles.
Key relationships in depth
Anthropic & Apple
Apple integrated Anthropic's Claude into Apple Intelligence in 2026 alongside OpenAI's ChatGPT, giving iPhone and Mac users the option to route complex queries to Claude models for AI responses. Anthropic's Claude became one of multiple third-party AI providers Apple supports in its AI orchestration framework, positioning Apple as a neutral platform that offers users choice among leading AI models while preserving Apple's privacy architecture.
Why it matters: Apple's multi-model strategy for Apple Intelligence allows it to maintain competitive positioning as AI model quality evolves, and Anthropic's Claude models are recognized for their safety, reasoning, and instruction-following capabilities that complement OpenAI's GPT models. Anthropic gains access to Apple's device ecosystem and the implicit trust transfer that Apple's brand provides to consumers who may be hesitant to engage with unfamiliar AI providers directly.
Microsoft holds a minority investment in Anthropic as part of a broader AI portfolio strategy that includes its dominant OpenAI partnership. While Microsoft's primary generative AI relationship is with OpenAI (through a $13B+ investment and exclusive Azure integration), the Anthropic stake provides Microsoft exposure to a leading safety-focused AI lab and a hedge against over-concentration in OpenAI's model family. Claude models are also available on Azure AI Studio, giving Microsoft enterprise customers access to Anthropic's Constitutional AI approach.
Why it matters: Microsoft's investment in Anthropic, reported at around $300M as part of a $7.3B funding round, reflects its strategy of maintaining optionality across the leading frontier AI labs rather than betting exclusively on OpenAI. As enterprise AI adoption matures, customers increasingly want to mix and match models from different providers for different tasks, and Microsoft's multi-lab portfolio allows Azure to serve as the neutral enterprise AI platform.
NVIDIA and Anthropic have a technology and investment relationship in which Anthropic runs Claude model training and inference on NVIDIA H100 and A100 GPU clusters, and NVIDIA has participated as an investor in Anthropic's funding rounds. NVIDIA NIM microservices are compatible with Anthropic's API standards, and the two companies collaborate on AI safety research tooling that requires GPU-accelerated interpretability experiments.
Why it matters: Anthropic's frontier model training requires tens of thousands of GPUs for each major training run, making NVIDIA the de facto infrastructure provider for Anthropic's research ambitions. NVIDIA's investment in Anthropic diversifies its strategic AI ecosystem beyond Microsoft and OpenAI, giving NVIDIA access to safety-focused AI development approaches and ensuring Anthropic's architecture choices remain GPU-friendly as model sizes scale.
Google committed up to $40B in investment into Anthropic in April 2026, with $10B in cash at Anthropic's $350B valuation and up to $30B contingent on performance targets. Separately, Anthropic agreed in October 2025 to expand its use of Google Cloud TPUs, gaining access to over 1 million TPUs and more than 1 gigawatt of AI compute capacity. By 2027, Broadcom and Google are providing Anthropic 3.5 gigawatts of TPU-based compute.
Why it matters: Google's investment in Anthropic hedges its Gemini development risk. If Gemini underperforms, Google still profits from Claude's success through its investment. For Google Cloud, Anthropic's $100B+ cloud commitment to AWS (primary) and Google (secondary) is an enormous workload anchor for GCP's AI business.
Relationships are compiled by VexMap from company filings, announcements and reporting, and last reviewed September 2026. Spotted something wrong? Tell us. Not investment advice.