Anthropic adjusts Claude session limits and says users will hit their limits faster during peak hours, amid compute strain due to Claude's new popularity
Follow Brent D. Griffiths … - Anthropic adjusted Claude usage caps during peak hours. — The changes show that the AI company is experiencing some compute strain.
Business InsiderBrent D. Griffiths
Context & Ripple Effects
Anthropic has been refining access controls for heavy Claude use for some time: Claude Code users previously reported restrictive caps, and the company later outlined new Pro and Max rate limits for a small group of continuous users. The latest peak-hour warning makes capacity management visible to a broader set of users rather than confining it to exceptional workloads.
Claude users will exhaust session allowances more quickly during busy periods, making access less predictable for time-sensitive and high-volume work.
Anthropic can ration constrained inference capacity immediately through usage policy rather than waiting for additional capacity or relying solely on blanket service degradation.
Second-order effects
Teams that depend on Claude Code or other sustained workflows may shift work outside peak periods, reduce usage, or evaluate alternative model providers when caps interrupt production.
Stronger peak-hour controls put more weight on plan design and usage credits: paid tiers may not by themselves guarantee uninterrupted access when demand is high.
Third-order effects
If repeated, peak-sensitive limits would make frontier-model subscriptions look increasingly like managed-capacity products, where nominal access is qualified by demand, workload type, and time of use.
The episode reinforces the compute-economics squeeze: model providers must balance user growth and feature rollout against the cost and availability of inference capacity, potentially making allocation policies a more important competitive differentiator.
The trend: Consumer and enterprise AI services are moving toward more dynamic inference allocation as usage growth outpaces the capacity available for consistently uncapped access.
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged. During weekdays between 5am-11am PT / 1pm-7pm GMT, you'll move through your 5-hour session limits faster than before.
We've landed a lot of efficiency wins to offset this, but ~7% of users will hit session limits they wouldn't have before, particularly for pro tiers. If you run token-intensive background jobs, shifting them to off-peak hours will stretch your session limits further.
I think people are overreacting to this. It sucks, but they clearly have a real compute shortage and this was the best path they had without cutting your usage massively
Overall weekly limits stay the same, just how they're distributed across the week is changing. I know this was frustrating. We're continuing to invest in scaling efficiently. I'll keep you posted on progress.
You're going from rewarding users using it during off-peak hours (double usage which ends March 28th) to punishing them for using during peak hours. I think that's a terrible move if you want happy customers.
Thariq: “I need to give our customers less for the same price. How can I obfuscate that as much as possible” Claude: “How about 'You'll move through your 5-hour session limits faster than before'”
Devs who can code also WITHOUT AI as well looking to became 10x more valuable They are the ones who won't panic or be idle when their Claude quota runs out... So much for all the advice on how learning to code is not worth it any more...
meanwhile, you can enjoy Codex with 2x rate limits for the next week 24x7 across all ChatGPT subscriptions! npm i -g @openai/codex time to build is NOW!!
You make this announcement BEFORE you make the change. Good riddance... what a shame @AnthropicAI and @claudeai. Customer service isn't hard, @DarioAmodei. This was a rug pool and y'all got caught. I hope the @FTC fines you.
My counterpoint to Citrini Research's prediction of mass unemployment by 2028 due to AI agents is we don't have the data center capacity. — Anthropic is tightening usage caps and expects 7% of users to hit peak-hour limits. Token cost and compute are still real bottlenecks to …