China's Moonshot pauses Kimi subscriptions amid hot demand, IPO push

Chinese artificial intelligence startup Moonshot AI has temporarily suspended new subscriptions for its recently launched Kimi K3 model, citing unprecedented demand that has strained its computing capacity. This bottleneck emerges as the company is actively pursuing new funding and preparing for a potential listing on the Hong Kong stock exchange.

Moonshot AI is reportedly unwinding its current offshore structure in anticipation of a Hong Kong IPO. The company has engaged financial advisors, including Goldman Sachs and China International Capital Corp (CICC), to discuss the listing plan, though the timeline remains flexible. Founded in 2023 by AI researcher Yang Zhilin, Moonshot AI has rapidly become one of China's most closely watched AI startups.

The company has a strong funding history, having raised over $2 billion in May from investors like Meituan and China Mobile, bringing its total historical fundraising to more than $5.5 billion. In June, it was seeking up to $2 billion in fresh capital, with its valuation reaching $30 billion. This intense investor interest in China's AI sector is driven by the demand for increasingly powerful AI models, but it also necessitates substantial investment in costly computing infrastructure.

Moonshot's competitors, such as DeepSeek, are also seeking external capital to expand their compute capacity as Chinese AI firms strive to close the gap with their U.S. counterparts. Moonshot announced on its X account that user requests for Kimi K3 have sharply exceeded forecasts, pushing existing computing clusters to their limits. The company will pause new consumer subscriptions immediately and prioritize available computing power for existing paid users.

To better match resources with demand, Moonshot plans to split future memberships into two tiers, including one specifically for coding tasks. The Kimi K3 model, unveiled as a 2.8 trillion-parameter model, is described by the company as the world's largest open-weight AI system. Its size and focus on coding and agent-style tasks make it more resource-intensive to serve at scale, requiring repeated model calls and significant inference capacity.

“

Kimi K3 has received far more love than we expected, and our GPUs are feeling it.