Analysis
Corrected Sep 27, 2026: an earlier version credited the $3 billion target raise to TechNode, which does not state it (the figure comes from Reuters), and described $50 billion as the IPO valuation; Reuters reports $50 billion as the valuation in an ongoing private round.
Moonshot AI, the Beijing-based developer of the Kimi chatbot, has submitted a confidential A1 application to list on the Hong Kong Stock Exchange, Reuters reported, via Business Standard, and TechNode separately reported the confidential A1 filing -- the concrete step behind a plan Pulse first covered in July, when Moonshot said it intended to list within six months after Kimi K3, a 2.8-trillion-parameter open-weight model, briefly rattled US AI stocks on release:
- Target raise -- $3 billion, one of the people told Reuters
- Private-round valuation -- $50 billion in an ongoing funding round, per Reuters, up from the roughly $31.5 billion pre-IPO round it was closing in July
- What's new -- the company has actually filed and named its underwriting banks, versus July's stated intent to list
“For US-based LPs with any exposure to China-focused funds, the sanctions overhang matters more than the valuation.”
Moonshot AI, founded in 2023 by Yang Zhilin and a small team of Tsinghua- and Carnegie Mellon-trained researchers, built its reputation on long-context models before Kimi K3 -- an open-weight release large enough to compete with frontier releases from OpenAI, Anthropic and Google on several public benchmarks -- turned it into one of China's most closely watched AI labs. Before it could file, Moonshot had to unwind its offshore holding structure and shift to onshore China domicile, Business Standard reported, a restructuring step that adds regulatory friction most Silicon Valley IPO candidates never encounter.
Moonshot is one of several Chinese labs racing toward Hong Kong listings in the same window: DeepSeek, Alibaba's Qwen team and Zhipu AI are all pursuing variations of the same path, part of what Pulse has tracked as a broader China AI IPO rush. That domestic competition is compounded by a harder external problem -- Moonshot faces US allegations that Kimi K3 was trained partly on export-restricted Nvidia chips and that some of its outputs were distilled from Anthropic's models, accusations serious enough that Treasury Secretary Scott Bessent has said he's considering adding the company to a trade blacklist.
A $50 billion target valuation against a $31.5 billion round the company was still closing in July is roughly 60% appreciation in under two months -- fast even by 2026's AI-valuation standards, and a bet that Hong Kong investors will price Moonshot closer to how US markets have priced OpenAI's and Anthropic's own pending IPOs than to how Chinese tech stocks have traded historically. Chinese AI names have generally carried valuation discounts to US peers on governance and capital-controls concerns; Moonshot's bankers are effectively betting that discount narrows before the roadshow.
For US-based LPs with any exposure to China-focused funds, the sanctions overhang matters more than the valuation. A Treasury blacklist designation wouldn't just hit Moonshot's US-facing partnerships -- it would likely force any US-domiciled fund holding pre-IPO shares to unwind the position, turning a potential Hong Kong listing gain into a forced, poorly timed sale.
The risk the $50 billion figure overstates is certainty: confidential filings routinely stall, get repriced down before a roadshow, or get withdrawn entirely if market conditions or regulatory scrutiny worsen -- and unlike a US S-1, a confidential HKEX A1 application gives outside observers almost no visibility into financials, burn rate or the specific chip-sourcing disclosures that would settle the export-control question one way or the other. That's a real limitation on how much conviction anyone should attach to the valuation before a roadshow prices it.
Whether Moonshot prices before or after any Treasury action is now the real thing to track -- a listing that closes ahead of a blacklist decision effectively forces US regulators' hand after the fact, while a delayed filing gives Washington time to act first.
