What's this about?
OrangeAI's assessment: Claude Code has moved into single-digit growth, developer penetration is close to saturated, the users who were going to migrate to Codex have already gone, and Cowork hasn't captured much of the general work market. His sharper point is about credibility — "back in May and June that team was out there evangelizing AI self-evolution, self-developing software, self-optimizing models. The real world is so very real."
What exactly does market saturation mean here?
Early growth in a developer tool comes from people who hadn't tried the category at all — a large pool, and easy to convert. Saturation is when almost everyone who would plausibly use a coding agent already uses one, so further growth must be taken from competitors rather than found. That's a fundamentally harder and slower game, and it's why growth rates fall sharply even when a product is doing nothing wrong. It also explains the capacity messaging this week: extending a usage promotion in two-week increments is what you do when you're defending a base, not expanding one.
Why is this worth noticing?
- This is one person's read, not a disclosure, and worth reading as informed commentary rather than data — Anthropic hasn't published these numbers.
- It's internally consistent with the week's other signals: the limits promotion extended rather than made permanent, capacity described as tight, a degraded-performance incident, and an HN thread full of subscribers naming their last month.
- The rhetorical point is the durable one. Peak-ARR evangelism about self-improving software is the kind of claim that gets quoted back, and this week it was.