Three Days After Kimi K3, Moonshot Kicked Off a $30 Billion IPO Run. The Biggest Open Model Is Now a Prospectus.

TL;DR

Moonshot AI, the Beijing lab behind Kimi K3, has officially distributed a shareholder resolution seeking approval for a Hong Kong listing that could come as soon as the next six months, according to a Bloomberg report citing people familiar with the matter. A concurrent funding round could value the company above $30 billion, roughly 7x its worth at the end of 2025. Annual recurring revenue reportedly hit $300 million in June, up from $200 million in April, and CICC and Goldman Sachs are in talks to work on the offering. The trigger, per Bloomberg, was K3 itself: the 2.8-trillion-parameter model Moonshot launched on July 16 and whose open weights are promised by July 27. The lab that plans to give away the biggest set of weights in history is about to ask public markets to price the giveaway.

The 72-Hour Turnaround

The sequencing is the story. K3 launched on Wednesday, July 16. By Sunday the 19th, Bloomberg was reporting that the shareholder resolution for an IPO had gone out to Moonshot's investors, that the company has been unwinding the offshore holding structure it would need to restructure before a listing (a process reports trace back to May), and that two heavyweight banks are circling the mandate. Chinese outlet PEdaily separately reported the same timing. Going from "weights coming soon" to "IPO coming soon" inside 72 hours is a remarkable promises-per-week ratio, even by 2026 standards.

To be precise about what is and is not confirmed: there is no filing with the Hong Kong exchange yet, and Moonshot has not publicly commented. What exists is a concrete corporate action (the resolution to shareholders), reported by two independent outlets, plus named banks in talks. That is well past rumor, but short of a prospectus.

Seven Times in Seven Months

The valuation staircase Moonshot climbed to get here is steep even by frontier-lab standards. The company ended 2025 worth about $4.3 billion. A $700 million raise in early 2026 took it to $10 billion. In May it closed $2 billion at a $20 billion valuation in a Meituan-led round, and by June it was already back out seeking $30 billion. The IPO-adjacent round now under discussion would lock that number in, a 50% markup on May in about ten weeks.

The Revenue Doing the Lifting

Underneath the multiple expansion there is real, fast-compounding revenue. Bloomberg puts Kimi's annual recurring revenue at $300 million in June, up from $200 million in April. That is 50% growth in two months, driven by paid subscriptions and API usage, and it is the number Moonshot will lead with when it walks into investor meetings.

Do the arithmetic and $30 billion is about 100 times current ARR. Growth like this can justify a lot, but a public listing means quarterly disclosure, and public markets will get to vote every three months on whether 100x revenue is a valuation or a vibe.

From Launch to Listing in Three Days

Bloomberg is explicit that K3 was the catalyst: the model, per its report, prompted a broad reassessment of Chinese AI capability, and Moonshot moved to seize the moment. That puts the IPO clock and the open-weight clock on the same wall. K3's checkpoints are promised by July 27 on Moonshot's Hugging Face org, under a reported modified-MIT license.

For builders waiting on those weights, the IPO cuts both ways. It makes the July 27 promise reputationally load-bearing: the listing story is "the open-weight lab that caught the frontier," and walking back the release would knife that narrative during the most scrutinized stretch of the company's life. But it also means the giveaway strategy will face quarterly earnings pressure from here on, and giveaways are historically the first line item that CFOs learn to hate.

Hong Kong Likes This Trade

Moonshot is not testing cold water. Zhipu AI listed first among China's LLM startups, and MiniMax doubled on its Hong Kong debut in January, raising about $620 million and closing its first day up 109%. By May, MiniMax was already preparing a mainland dual listing. Hong Kong's market has been rewarding Chinese AI listings all year, and a $30 billion Moonshot would arrive as the largest of the cohort by a wide margin.

The context that makes this more than a finance story: the flow of frontier-adjacent open weights that AI builders now take for granted (K3, GLM, Qwen, DeepSeek) is increasingly funded by this exact capital-markets machine. Whether that machine keeps paying for free weights is a question that directly shapes what you get to self-host in 2027.

What It Means If You Build on Kimi

The IPO makes explicit what the pricing already implied: open weights are the funnel, not the product. It works like the free-samples counter at a warehouse store: the tasting costs nothing, the cart at checkout is where the money is. The downloadable checkpoints seed the ecosystem and the benchmarks seed the headlines, while the paid Kimi subscriptions and the $3-per-million-token API (the version most production traffic actually uses, because almost nobody self-hosts 2.8 trillion parameters) ring the register.

Three practical readings. First, K3's aggressive pricing is likelier to survive than not in the near term, since the growth curve is the IPO's centerpiece and price hikes flatten growth curves. Second, the weights promise just got a $30 billion co-signer, which is the strongest guarantee short of the files actually existing. Third, watch the license text when the weights land: a modified-MIT license drafted the same month as an IPO resolution deserves a careful read of the modifications.

Caveats

All of this is sourced reporting on a private process, not a public filing. The six-month timeline is "as soon as," which is banker language for "could slip." The $30 billion figure is a target for an unpriced round, valuations in funding talks routinely move, and Moonshot itself has not confirmed any of it. IPO plans get shelved when markets turn, and Hong Kong's AI window, however open today, is a window. Treat the direction as solid and every specific number as provisional until a prospectus exists.

Key Takeaways

• Bloomberg reports Moonshot AI has distributed a shareholder resolution for a Hong Kong IPO that could come within six months, with CICC and Goldman Sachs in talks on the mandate; PEdaily corroborated the timing.
• A concurrent funding round could value Moonshot above $30 billion, up from $20 billion in May and roughly 7x its value at the end of 2025.
• Kimi's reported ARR hit $300 million in June, up from $200 million in April, which puts the target valuation at roughly 100 times current revenue.
• Bloomberg names Kimi K3, launched July 16 with open weights promised by July 27, as the catalyst; the IPO narrative now depends on that release actually shipping.
• Precedent is friendly: MiniMax doubled on its January Hong Kong debut, and Zhipu listed before that. Moonshot would be the largest Chinese LLM listing yet.
• Nothing is filed and nothing is priced. The direction is well-sourced; every number is provisional until there is a prospectus.

Sources: Bloomberg, Yahoo Finance (Bloomberg syndication), The Next Web, TechCrunch, South China Morning Post, CNBC, Kimi K3 launch post