China’s Moonshot AI pauses Kimi subscriptions as demand surges sixfold ahead of Hong Kong IPO
The Beijing-based AI startup's revenue tripled in three months to $300 million ARR, and now it's eyeing a public listing that could value it north of $30 billion
Moonshot AI, the Beijing-based company behind the Kimi chatbot, did something most startups would consider unthinkable. It told new customers to stop giving it money.
The company paused new subscriptions to its Kimi K3 model over the weekend of July 18-19, after daily sales surged at least sixfold following the model’s launch around July 17.
From zero to $300 million in record time
Moonshot AI’s annualized recurring revenue hit $300 million in June 2026. That figure was $100 million in March. And $200 million in April.
The revenue surge has been driven primarily by paid subscriptions and enterprise access to products like Kimi Work.
Kimi K3 packs 2.8 trillion parameters and outperforms major competitors on key coding benchmarks. The model’s open-weight architecture means developers can tinker with it, which partly explains the stampede of new subscribers that forced the company to hit pause.
Moonshot AI was founded in 2023.
The Hong Kong IPO playbook
Moonshot is now executing plans for an initial public offering in Hong Kong within the next six months. The IPO preparation involves two critical steps: securing shareholder approval, and dismantling the company’s variable interest entity structure to meet the compliance requirements Beijing now mandates for overseas listings.
The company’s most recent private valuation topped $20 billion after a $2 billion funding round in May 2026. That round featured Meituan, Alibaba, and Tencent. Future fundraising efforts are reportedly targeting a valuation past $30 billion.
What this means for the broader AI investment landscape
Moonshot’s revenue growth, from $100 million to $300 million ARR between March and June, suggests that Chinese enterprises and consumers are adopting AI tools at a rapid pace.
Kimi K3’s benchmark performance adds competitive pressure to the broader AI landscape. When a model with 2.8 trillion parameters is beating competitors on coding tasks, it reinforces that compute demand is significant. That is relevant for projects positioning themselves as decentralized alternatives to centralized cloud providers.
Hong Kong’s role as the IPO destination is notable. The city has been actively positioning itself as a hub for both tech listings and crypto-friendly regulation.
The VIE restructuring required for the Hong Kong listing introduces legal complexity that has tripped up Chinese companies before. And going from a $20 billion valuation to a target above $30 billion assumes sustained revenue acceleration. The backing from Alibaba, Tencent, and Meituan provides credibility, but public market expectations tied to ARR growth remain a key variable.