Capital

Kimi's Maker Is About To Publish Its Real Numbers

Moonshot AI is raising at up to 50 billion dollars before a Hong Kong listing, with revenue up from 200 to 300 million dollars in two months. A prospectus is the first real diligence a Kimi buyer will get.

CapitalBy Servola Tech Desk2026-07-254 min read

AI-assisted, edited by humans. | Editor-in-Chief: Leon Soliman | Editorial standards

Kimi's Maker Is About To Publish Its Real Numbers

Key takeaways

Moonshot AI's annual recurring revenue went from 200 million dollars in April to 300 million in June, and a Hong Kong listing will force the audited disclosure that no European buyer of Kimi K3 can obtain today.
The company is closing a round at a 31.5 billion dollar pre-money valuation, opening talks in August for up to 50 billion, and is unwinding its offshore holding structure so it can file in Hong Kong within roughly six months.
Goldman Sachs and China International Capital Corporation are in talks to underwrite, which is the signal that this is a real filing timetable rather than a valuation rumour.
Do not sign a multi-year Kimi commitment on pre-IPO pricing this quarter; keep the term short, demand written notice on price changes, and reopen the decision when the prospectus lets you diligence the vendor properly.

A vendor that stopped taking customers is raising at 50 billion

Moonshot AI spent last week turning new subscribers away because demand for Kimi K3 outran its serving capacity. It is spending this week preparing to raise money at a valuation of up to 50 billion dollars. Both things are true at once, and the combination is the most useful thing an operator can know about the company right now.

The mechanics are advanced rather than speculative. Moonshot is closing its current round at a pre-money valuation of 31.5 billion dollars, and according to Bloomberg it plans to open talks in August for a final pre-listing round at up to 50 billion. It is unwinding the offshore holding structure that Chinese technology companies traditionally used, a prerequisite for a clean Hong Kong listing under the securities regulator's revised rules, and it intends to file within roughly six months. Goldman Sachs and China International Capital Corporation are in talks to underwrite.

The revenue line explains the hurry

Annual recurring revenue went from 200 million dollars in April to 300 million in June. That is a 50 percent increase in two months, driven by a 2.8-trillion-parameter open-weight model that independent evaluations placed alongside the frontier systems from Western labs. A company growing at that rate has a narrow, obvious window: raise while the benchmark headlines are fresh, then list before a competitor resets the comparison.

Underwriters do not join a timetable they think will slip. The presence of Goldman Sachs and CICC is the part of this story that separates it from the valuation chatter that surrounds every Chinese AI lab. Treat the six-month filing window as a planning date rather than an aspiration.

A prospectus is the diligence you cannot buy today

Here is the part that matters to a European buyer, and it runs against the usual reading of an IPO. If your team has standardised on Kimi K3, you currently cannot answer basic vendor questions. How much serving capacity exists. What share of revenue comes from the handful of largest customers. Whether inference is sold above or below cost. How much of the compute sits on hardware that an export rule could reach. A private Chinese company owes you none of that, and last week's subscription pause showed exactly why the answers matter.

A Hong Kong listing changes that. A prospectus and subsequent audited reporting put capacity, concentration, margin and geography into the public record, on an exchange with real disclosure obligations. For once, the corporate-finance event is the procurement event: the first genuine diligence file on this vendor will be written by its own bankers.

The same listing works against your price

Cheap inference is a pre-IPO condition, not a permanent feature. Loss-leading on price to win developer share is a rational move for a private company chasing a valuation. It is a much harder position to defend to public shareholders who read gross margin every quarter. The pricing that makes K3 attractive today is being funded by the same investors who will want it repaired after listing.

There is a second effect worth naming plainly. A listed national champion is a more visible object than a private startup, and visibility is what draws export-control and procurement attention. Neither risk argues against using the model. Both argue against treating this quarter's price as the price you will pay in 2028.

Set the contract term around the filing, not the benchmark

The decision in front of you is not whether K3 is good. Independent testing has largely settled that. The decision is contract term, and the listing timetable gives you an unusually clear calendar to work against. Keep commitments inside the pre-IPO window short, twelve months at most, so you are not locked into an arrangement made before the numbers were public.

Three specifics. Require written notice on any price or rate-limit change, with enough lead time to move workloads. Keep a second model wired and periodically exercised, because the capacity pause proved that availability, not quality, is the failure mode here. And diarise the prospectus: when it lands, read the capacity and customer-concentration disclosures before you renew, because that will be the first time you can price this vendor on evidence rather than on benchmark scores.

Frequently asked questions

Does a Hong Kong listing make Kimi K3 safer to depend on?

It makes the vendor legible, which is not the same as safe. You gain audited figures on capacity, revenue concentration and margins, which is diligence you cannot get today. You do not gain protection from price increases or from export controls, and public-market margin pressure makes a price rise more likely rather than less.

Should we delay adopting Kimi K3 until after the IPO?

Not necessarily. Delay the length of the commitment, not the adoption. Running the model on a short term with a tested fallback lets you benefit from current pricing while keeping the option to renegotiate once the prospectus discloses the economics. Signing a multi-year deal now is the move that removes that option.

How firm is the six-month filing timetable?

Firmer than a typical valuation report, because Moonshot is already unwinding its offshore holding structure to satisfy the Chinese regulator's listing rules and has Goldman Sachs and CICC in underwriting talks. Those are steps taken by companies that intend to file. Timetables still slip, so treat it as a planning date and check it again when the August funding round closes.

The interesting number is not 50 billion dollars, it is the audited page count that has to exist before anyone can charge it. Operators who put their renewal date after the prospectus will be the only Kimi customers negotiating with evidence.

Moonshot AIKimi K3Hong Kong IPOAI procurementvendor diligenceGoldman Sachsopen weightscontract terms

Sanctions Are Now on the Table for Kimi K3

A White House official says Moonshot distilled Anthropic's Fable to build Kimi K3 and used banned Nvidia chips via Thailand. Treasury put sanctions on the table. Here is what it means for anyone running Chinese open weights.

6 min read

AI Policy

Your Cloud Contract Decides If You Keep Kimi K3

Treasury Secretary Scott Bessent said Washington can sanction foreign AI models for IP theft. For European operators the block arrives through suppliers, not weights.

5 min read

Law & Regulation

The EA Buyout Now Hinges on Brussels

The largest take-private in history has cleared EA shareholders and US antitrust. The last real gate is the EU - and not classic merger review, but the Foreign Subsidies Regulation, facing its first blockbuster test.

3 min read

← All articles