Three weeks after hitting #1 on the App Store and running a 2x usage promotion to welcome a wave of new users, Anthropic is facing sharp criticism from its paying subscriber base over peak-hour rate limits on Claude Pro.
The core complaint from Pro subscribers ($20/month): during peak hours, as few as two standard prompts can exhaust an entire 5-hour usage window. Some users report that Claude's free tier running Sonnet is functionally more usable during peak periods than the paid Pro plan, since free users face different throttling mechanics. The promotion that doubled Pro usage is set to expire March 29, which many expect will make the situation feel worse, not better.
The Growth-vs-Capacity Tradeoff
This is the classic scaling problem every AI company faces. Anthropic successfully attracted a massive influx of users, but GPU capacity doesn't scale on the same timeline as app downloads. The result: existing paying customers feel like they're subsidizing free-tier access while getting a degraded experience themselves.
Several users have proposed what seems like an obvious fix: restrict free-tier access during peak hours rather than throttling paid subscribers. The logic is simple - if compute is scarce, prioritize the people paying for it. Anthropic hasn't publicly responded to these specific suggestions.
The timing is particularly sensitive. Many of these frustrated users explicitly left ChatGPT for Claude and committed to paid plans. Feeling punished for that loyalty, especially right after a promotional period designed to onboard exactly these users, creates the kind of resentment that's hard to walk back.
For Anthropic, this is a pricing and infrastructure problem that needs a visible solution - transparent usage meters, guaranteed minimums for paid users, or tiered priority access. The worst outcome would be losing the exact users they just worked so hard to acquire.