Moonshot AI, the Beijing-based developer of the Kimi family of AI models, has confidentially filed for a Hong Kong initial public offering targeting roughly $3 billion, people familiar with the filing told reporters this week — a listing that would value the four-year-old startup at around $50 billion, up from roughly $30 billion just weeks earlier.
The jump follows a fresh funding round in which Moonshot raised more than $2 billion from Meituan, China Mobile and private equity firm CPE, according to Tech Startups' reporting on the filing. That round pushed Moonshot's total funding past $5.5 billion from a backer list that includes Alibaba, Tencent, IDG Capital and HSG, the firm formerly known as Sequoia Capital China. Goldman Sachs, CICC and Deutsche Bank are managing the proposed offering, though the timing and final size of the deal remain subject to regulatory approval and market conditions.
A crowded field of Chinese AI listings
Moonshot's push toward Hong Kong follows two rivals that have already gone public there this year: Z.AI, formerly known as Zhipu, which now carries a market capitalization near $66 billion, and MiniMax, which listed earlier in 2026. Chinese AI-search rival DeepSeek, still private, is valued at roughly $74 billion — meaning that even at $50 billion, Moonshot would list as one of the smaller members of China's newly public AI cohort, despite the speed of its valuation climb.
The company's growth has been driven by Kimi K3, unveiled July 17 at the World Artificial Intelligence Conference in Shanghai and described by Moonshot as the world's largest open-weight model, with roughly 2.8 trillion parameters in a mixture-of-experts architecture and a one-million-token context window. The model has drawn strong developer interest — enough, according to reporting cited by Ground News, to push Moonshot's annualized revenue run rate to roughly $300 million — but demand has also strained the company's available computing capacity, a constraint shared by nearly every fast-growing AI lab racing to keep pace with user growth.
Moonshot was founded in Beijing in 2023 by Yang Zhilin, a 34-year-old researcher who studied computer science at Tsinghua University, completed a machine learning doctorate at Carnegie Mellon in four years, and interned at Google Brain and Meta AI before co-authoring Transformer-XL and XLNet, two papers that shaped how the industry approaches long-context language models — expertise that shows up directly in Kimi K3's million-token context window. Yang has said in an internal memo, according to Crypto Briefing, that the company was not rushing to go public, framing the fresh capital as giving Moonshot the option to wait rather than an obligation to list immediately.
A possible U.S. tie-up
Perhaps the most striking detail in the filing reports is that Moonshot is said to be negotiating revenue-sharing arrangements with Microsoft, Amazon and Google to host Kimi K3 on their cloud platforms — a deal that, if finalized, would mark an unusual moment of commercial cooperation between a Chinese frontier AI lab and the three largest American cloud providers at a time when Washington and Beijing remain locked in an escalating contest over AI chips and export controls. None of the three U.S. companies has confirmed such talks publicly.
Moonshot's timing also reflects a broader recalibration in how investors are pricing Chinese AI startups. A year ago, IPO windows for Chinese tech companies in Hong Kong were narrow and skittish; the successful listings of Z.AI and MiniMax appear to have reopened the market, giving Moonshot's bankers confidence that a $3 billion raise is achievable even as questions persist globally about whether AI valuations broadly have run ahead of revenue.
No listing date has been set, and a confidential filing gives Moonshot flexibility to adjust the offering's size or delay it entirely if market conditions sour. If the IPO proceeds at anything close to its targeted valuation, it would rank among the largest technology listings in Hong Kong this year and would hand early backers including Alibaba and Tencent a substantial return on a company that was, until recently, a distant also-ran in China's crowded AI field.
