Simon Willison · 2026-05-27 · notable
Simon Willison: I Think Anthropic and OpenAI Have Found Product-Market Fit
Willison argues both labs hit product-market fit through enterprise coding agents — Claude Code and Codex — pointing to the April 2026 shift to usage-based API pricing, pricier new flagship models, and enterprise-heavy hiring.
The case that AI's business model finally works — built on enterprise coding agents, not consumer chat subscriptions.
What is it?
An analysis post from developer and LLM commentator Simon Willison arguing that Anthropic and OpenAI have reached product-market fit. His core claim is that the money is in enterprise coding agents — Claude Code and Codex — billed by usage, not in $20/month consumer chat plans.
How does it work?
Willison stacks several signals: the April 2026 move of enterprise customers from flat per-seat pricing to API usage pricing; new flagship models priced higher than predecessors (GPT-5.5 at roughly 2x, Opus 4.7 at roughly 1.4x); his own documented $2,180 in monthly token value against a $200 subscription; ChatGPT's 900M weekly users with only about 5.6% paying; and enterprise-focused hiring at both labs. He reads recent budget-overrun and seat-cancellation stories as customers accepting premium pricing rather than as AI failure.
Why does it matter?
For two years the open question was whether frontier labs could turn heavy usage into sustainable revenue. Willison's framing reorients that debate around per-seat enterprise spend of $200+/user and locked annual contracts, and lands as Anthropic is rumored to approach its first profitable quarter.
Who is it for?
engineering leaders, AI-industry watchers