Moonshot AI Files Confidentially for Hong Kong IPO, Targets $3 Billion Raise

Moonshot AI, the Beijing-based startup behind the Kimi K3 large language model, has confidentially filed for an initial public offering in Hong Kong, according to multiple people with knowledge of the matter. The company is aiming to raise $3 billion, one person told Business Standard, in what would be one of the most anticipated near-term listings by a Chinese AI firm.

The timetable for the IPO is subject to regulatory approvals, and financial details including proceeds sought could change depending on market conditions, sources said. According to the South China Morning Post, which also reported the confidential filing, the company is targeting a listing as early as the first quarter of 2027, though the timeline remains dependent on approvals and market conditions.

Moonshot is valued at roughly $50 billion in an ongoing funding round, a figure confirmed by two separate sources to Business Standard and echoed by a Quartz summary. That valuation sits below some domestic rivals: DeepSeek is valued at about $74 billion and Z.AI at $66 billion, sources have told Business Standard.

The startup made headlines in July with the release of Kimi K3, an open-weight model with more than 2.8 trillion parameters — the world's largest in its category, according to the company. The model has drawn positive reviews and strong demand, but has also strained Moonshot's computing capacity, sources said.

Revenue-Sharing Talks and Scrutiny

Moonshot is in discussions with Microsoft, Amazon and Google over revenue-sharing agreements that would allow the U.S. cloud companies to host Kimi K3, Business Standard reported, citing sources. Any such deal would mark the first major revenue-sharing pact between a Chinese AI firm and a major U.S. cloud provider.

The company has also attracted scrutiny from senior U.S. officials. U.S. Treasury Secretary Scott Bessent has said he may add Moonshot to a trade blacklist, according to Business Standard, which noted that the startup faces allegations it used restricted Nvidia chips and distilled capabilities from Anthropic's models. Moonshot has disputed claims that K3's performance was achieved through distillation.

Regulatory Hurdles and IPO Structure

To gain regulatory approval for the Hong Kong listing, Moonshot has had to unwind its offshore "red-chip" incorporation structure and shift to an onshore China domicile before the filing, two sources familiar with the matter told Business Standard.

The startup was founded in 2023 by Yang Zhilin, an AI researcher who pursued doctoral studies at Carnegie Mellon University.

Investor Backing and IPO Banks

Moonshot's investors include Alibaba, Tencent, IDG Capital and HSG (formerly Sequoia Capital China), according to Business Standard. The company raised more than $2 billion in May from investors including Meituan, China Mobile and Chinese private equity firm CPE, bringing total fundraising to more than $5.5 billion, per a fundraising document seen by Reuters and reported by Business Standard.

For the IPO, Moonshot is working with banks including Goldman Sachs, CICC and Deutsche Bank, Business Standard reported. CICC did not immediately respond to a request for comment; Goldman and Deutsche Bank declined to comment.

Context

The listing adds to a hectic pace of fundraising by AI companies globally. Anthropic, developer of the Claude model, is planning to launch an offering in the coming days and complete its listing by October, Business Standard reported. Among Chinese firms, Z.AI (formerly Zhipu) and MiniMax listed in Hong Kong this year.

Funds raised via Hong Kong listings totaled $41.2 billion as of mid-August, up 142% year-on-year, with Chinese technology companies accounting for most of that, according to LSEG data cited by Business Standard.

Moonshot said it "doesn't comment on market rumours or speculation" in response to a query from the South China Morning Post. Chinese tech-focused publication LatePost first reported the confidential IPO filing on Wednesday, Business Standard noted.