Anthropic

What they built

Anthropic is the only foundation-model lab whose product positioning is indistinguishable from its research positioning: safety-first, enterprise-native, long-context, tool-use-native. The Claude model family — Claude Opus, Sonnet, and Haiku — now runs in production at roughly half of the AmLaw 100, every major consultancy, and a growing share of banks. Claude Code, the developer product launched in 2024, hit $2.5B ARR in early 2026 almost entirely on self-serve and is the fastest-growing line in the company.

How they got here

The founders — mostly former OpenAI leadership — started Anthropic in 2021 with a thesis that was unfashionable at the time: models would get much more capable very quickly, which meant alignment and interpretability had to ship alongside capability, not after. For the first two years that looked slow next to OpenAI’s go-to-market velocity. By 2024 the thesis was proven: the enterprises that moved from OpenAI to Claude, or ran both in parallel, cited safety behavior, steerability, and the long-context memory budget as the deciding factors.

Revenue caught up with the thesis fast. $1B ARR by end of 2024. $9B by end of 2025. $30B by March 2026 — growth that outpaced every SaaS cohort in history at the same revenue scale. The February 2026 Series G closed $30B at $380B post-money, led by GIC and Coatue, with reported preemptive offers at $800B+ already circulating.

What’s ahead

The two questions that matter: (1) Can Anthropic’s enterprise-first revenue mix continue to grow at the current rate without a flagship consumer product? Management has been explicit that consumer is not the priority; the question is whether that discipline holds if ChatGPT-class consumer revenue becomes too tempting to ignore. (2) Does the Amazon partnership scale — the Trainium-based infrastructure commitment — fast enough to keep training cost parity with OpenAI’s Azure + Oracle stack?

Why it matters

For enterprise-AI founders, Anthropic is the reference customer-acquisition motion: sell to a small number of large, sophisticated accounts, let them specify the product, and let net-dollar-retention compound. For investors, the Claude Code line alone ($2.5B ARR and doubling) is proof that a well-built AI dev product can grow faster than any SaaS category ever has — which resets every comp multiple in the space.