China's Moonshot AI Halts Kimi Subscriptions Amid Demand Surge, Hong Kong IPO Plans

Chinese AI startup Moonshot AI has temporarily suspended new subscriptions for its popular Kimi chatbot, citing overwhelming demand that has strained its computing resources. The service disruption comes as the company, recently valued at $30 billion, is reportedly preparing for a potential initial public offering in Hong Kong.

Demand for Kimi K3 Exceeds Capacity

Moonshot AI announced Sunday that a surge in user requests following the release of its new Kimi K3 model had pushed its systems to their limits. In a statement, the company described the demand as creating "unprecedented compute challenges," with user activity over the past 48 hours "sharply exceeding forecasts."

The capacity crunch follows the launch of Kimi K3 on Friday, which Moonshot unveiled as a 2.8 trillion-parameter model, positioning it as the world's largest open-weight AI system. In response to the bottleneck, Moonshot is pausing new consumer sign-ups to allocate available computing power to its existing paid users, who the company said would not be affected.

IPO Preparations and Fundraising

The operational challenge coincides with significant corporate finance activity. According to multiple sources familiar with the matter cited by Reuters, Moonshot is preparing for a Hong Kong IPO and is in the process of unwinding its current offshore corporate structure.

Ad

The startup has reportedly engaged financial advisers including Goldman Sachs and China International Capital Corp (CICC) to discuss the listing, though the timeline remains fluid. This follows a successful funding round in May where Moonshot raised over $2 billion from investors including Meituan and China Mobile. The company is now reportedly seeking up to $2 billion in fresh capital.

A Sector-Wide Compute Challenge

Moonshot's subscription pause highlights a critical issue facing China's rapidly growing AI industry: the intense competition for costly and scarce computing infrastructure. As Chinese firms like Moonshot, DeepSeek, and Alibaba race to close the gap with U.S. rivals by releasing increasingly powerful models, their need for advanced hardware escalates.

This infrastructure demand is further complicated by U.S. export controls on advanced semiconductors, creating a significant constraint for Chinese AI developers aiming to scale their services. The incident underscores the direct link between a model's capabilities and the immense, costly computing power required to serve it to a mass audience.