Anthropic announced it’s putting weekly usage caps on Claude subscriptions starting August 28.
They say it’ll affect fewer than 5% of users, but that’s cold comfort to anyone running Claude Code 24/7 or building products on top of it. Some of these folks were pushing usage worth tens of thousands of dollars—on a $200/month plan.
The goal? Stop account sharing, resale, and flat-out abuse. But not everyone’s happy about it.
Get the Latest on AI
Stay updated on real AI changes, join the We Are Founders mailing list.
What’s Actually Changing?
If you're on the Pro or Max plan, here’s the shift:
- Current system: Sessions reset every 5 hours.
- New system: A weekly limit gets layered on top.
- Max 20x plan users get 240–480 hours of Sonnet 4 and 24–40 hours of Opus 4 each week.
Sounds generous, until you realize you might blow through your Opus time in 2–3 days if you're a power user. Oh, and if you hit the cap, you can pay extra at API rates or… wait. For days.
The Reddit Fallout
Founders, devs, and Claude die-hards on Reddit had thoughts—and lots of them:
- Frustration with lack of transparency
You can’t check your usage. No tracker, no dashboard. Just vibes and prayer. - Backlash against collective punishment
Many felt Anthropic should’ve targeted abusers directly instead of penalizing everyone. - Suspicion of upcharging
Some suspect this is a soft nudge toward higher-priced plans or API upsells. - Confusion over what “hours” mean
Is it active usage? Session windows? Token consumption? Nobody’s sure—and that’s a problem.
What Founders Should Take From This
Whether you're building a dev tool, a subscription service, or anything vaguely SaaS-y, this whole mess is a goldmine of lessons:
- Unlimited isn’t sustainable, don’t fake it
If your pricing depends on people not fully using what they pay for, expect chaos later. Be honest up front. - Show usage data
Users want to feel in control. If you're enforcing limits, show them how close they are. No one likes hitting a wall they didn’t see coming. - Explain your decisions clearly
Anthropic’s “less than 5%” line might be true, but it feels dodgy when the silent majority never uses the service, and active users take the hit. - Don’t punish your best users without a safety net
Founders pushing the limits often drive product feedback and growth. If you throttle them, at least offer something better (like API access or an "ultra" plan).
Final Thought
Anthropic’s decision isn’t surprising. The writing was on the wall as soon as people started spinning up Claude sub-agents and pushing 24/7 compute.
But the rollout shows how quickly goodwill can erode when limits feel vague, sudden, and out of step with how people actually use the product.
Founders: build your limits with clarity, not hope. And if you say “unlimited,” mean it—or don’t say it at all.
Want more stories like this? We interview founders building everything from niche tools to million-dollar products. Subscribe to We Are Founders for real stories.
