Moonshot AI has temporarily suspended new consumer subscriptions for Kimi K3 after demand for its newly launched model exceeded expectations and pushed its compute infrastructure close to its limits.
In a statement titled "To Kimi Users: An Update on Compute Capacity Constraints and Subscription Suspension," the company said user requests in the 48 hours following the Kimi K3 launch surged far beyond projections, bringing its existing compute cluster close to maximum capacity. To preserve service quality for existing subscribers, Moonshot AI suspended new consumer subscriptions with immediate effect, dedicating all available compute resources to current members.
Kimi K3, a 2.8-trillion-parameter Mixture-of-Experts model released days ago, quickly topped a major coding leaderboard, driving an explosion of demand. The capacity crunch highlights the immense infrastructure required to serve frontier-scale models — and the particular pressure on Chinese AI labs facing export-controlled GPU access. Moonshot cannot simply buy its way out the way US rivals might.
The strategic silver lining: Kimi K3's open weights are still scheduled for release on July 27. Once public, anyone can self-host the model or use inference providers, shifting the capacity bottleneck from Moonshot to the broader ecosystem. The suspension, while inconvenient for new users, is a bullish signal disguised as bad news — real demand, real capability, and a real open-weight resolution on the horizon.
Sources
- technode.comTechNode — Kimi K3 overwhelms capacity, suspends new consumer subscriptions
- asia.nikkei.comNikkei Asia — China's Moonshot AI halts subscriptions for Kimi K3
- scmp.comSCMP — Kimi K3 developer suspends new subscriptions amid compute constraints
- the-decoder.comThe Decoder — Moonshot pauses new Kimi K3 subscriptions after GPU demand maxes out in 48 hours
