China's Kimi K3 Halts Signups as Demand Overwhelms It

Moonshot AI just proved its newest model is either a massive hit or a massive miscalculation, possibly both. The Chinese AI company paused new subscriptions to Kimi K3 within days of its public release, after a surge in demand overwhelmed the computing capacity the company had allocated for it, according to BNN Bloomberg's reporting.

Kimi K3 launched last week as a 2.8 trillion-parameter model, making it the largest open-weight AI system released to date. Unlike closed models such as OpenAI's or Anthropic's offerings, open-weight models let developers examine and adapt the underlying system directly rather than accessing it only through a paid interface, a distinction we've covered in our broader look at what generative AI actually means for businesses choosing between vendors.

The scramble says less about Kimi K3's quality and more about how unprepared Moonshot was for its own success. According to Euronews's coverage of the launch, the model has already outperformed rivals including GPT-5.6 Sol and Claude Fable 5 on several benchmarks, including topping Arena's leaderboard for front-end coding capability following its public release.

Lian Jye Su, chief analyst at technology research firm Omdia, offered a blunt read on what happened, according to that same Euronews report. "This does show Moonshot AI does not have sufficient compute chips to serve the current surge in demand," he said, adding that the more likely explanation is Moonshot simply underestimated how popular K3 would become, since the model is unusually demanding in terms of compute, making capacity allocation both difficult and costly.

Kimi K3 isn't an isolated surprise. It's part of a wave of low-cost, openly available Chinese models that have gained real traction internationally in recent months, following DeepSeek's V4 release and the market shock caused by DeepSeek's original model in early 2025, a moment that first prompted many in the industry to view China as a serious AI rival to the U.S. That earlier disruption is worth revisiting in our DeepSeek statistics coverage if you want the fuller picture of how quickly this rivalry has escalated.

The release has already weighed on shares of major U.S. technology firms, driven by concern that cheaper Chinese alternatives could squeeze the pricing power American AI companies have relied on, a dynamic playing out despite existing U.S.-led export restrictions on advanced chips.

I've advised companies on AI adoption for four years, and this story is worth watching closely regardless of whether you use Chinese AI models directly. When an open-weight competitor can top leaderboards against Western frontier models while charging a fraction of the price, that pressure eventually filters into what every AI vendor charges, including the ones you're already using. Our generative AI market statistics page tracks how pricing dynamics like this are shifting the competitive landscape more broadly.

For businesses evaluating AI vendors, this is a reminder to stay flexible rather than locking into long-term contracts based on today's pricing. The gap between what U.S. and Chinese AI labs charge for comparable capability keeps narrowing, and that trend shows no sign of reversing.

Moonshot AI paused new signups for its Kimi K3 model days after launch, after demand overwhelmed the compute capacity it had allocated. The 2.8 trillion-parameter model has already topped some coding benchmarks against Western rivals like GPT-5.6 Sol and Claude Fable 5. The episode has renewed concerns about cheaper Chinese AI models squeezing pricing power across the industry.