Does Just a moment... have a moat?

Training and serving frontier models takes rare research talent, safety and alignment methods, and years of experience. Menlo Ventures put Claude at about 42% share of code generation among developers, roughly double OpenAI's.

To cross it: Hire a top research team and train a model that beats Claude on agentic coding. Only a few labs worldwide (OpenAI, Google DeepMind, xAI, DeepSeek) have managed it.

Reported raises include a $30B Series G at $380B and a $65B Series H at a $965B valuation in May 2026. Bloomberg reported a run rate above $65B and positive adjusted operating income in Q2 2026.

To cross it: Raise tens of billions on proof of a competitive model, which in practice only hyperscalers and a handful of labs can do.

Anthropic has multi-gigawatt compute deals: a Google and Broadcom TPU agreement worth tens of billions, plus AWS Project Rainier and up to 5 GW of new Amazon compute.

To cross it: Lock up gigawatts of GPU or TPU capacity years ahead, competing for the same scarce chips and power.

Anthropic reports 300,000+ business customers, and Menlo Ventures estimated about 40% of enterprise LLM API spend in 2025. It has SOC 2, HIPAA-eligible offerings and distribution through AWS Bedrock and Google Vertex AI.

To cross it: Earn enterprise security reviews, cloud marketplace listings and years without incidents, while matching the model's quality.