In three months, the market attitude towards the dark side of the moon has changed.

Ask AI · Why Soaring Funding for Moonshot AI Is Prompting a Reassessment of Its Listing?

China Business News Reporter Li Kunkun Li Zhenghao Beijing Report

Wu Qing: In less than three months, the issue of Kimi’s parent company—Moonshot AI—going public has taken a U-turn.

In a company-wide letter to all employees at the end of 2025, Moonshot AI founder Yang Zhilin wrote: “We are not in a hurry to go public in the short term, and going public is not our purpose.” The backdrop is that Moonshot AI had just completed a $500 million Series C funding round at that time, with a valuation of more than $4 billion and cash on the books of over 10 billion RMB.

By the end of March 2026, media outlets citing people with knowledge of the matter reported that Moonshot AI was evaluating the possibility of listing in Hong Kong and had already been in touch with China International Capital Corporation and Goldman Sachs regarding listing matters.

A reporter from China Business News contacted the Moonshot AI team regarding its preparations for going public and its development plans. The company claimed, “There is no information we can provide.” However, an internal employee at Moonshot AI revealed that the company had been conducting various internal reviews since the end of last year, including cost accounting.

Over the past three months, what caused Moonshot AI to change its plan? First, in February 2026, Moonshot AI completed another $700 million of financing, and it has recently been pushing forward a new round of $1 billion financing; industry chatter suggests its post-investment valuation could surge to $18 billion. Second, Zhipu AI and MiniMax—both also in the domestic “first tier” of large model companies—had already listed on Hong Kong equities in January 2026, and after listing their stock prices have continued to rise; their valuations are currently stable at around $40 billion. In addition, 2026 has been defined by capital markets as the year of IPOs: private companies with a total value of about $2.9 trillion are preparing to go public. AI is the main character here, and the market window does not wait for anyone.

Performance Reversal

According to Tianyancha, at the end of 2025, Kimi completed a $500 million Series C funding round, with a post-investment valuation of $4.3 billion. During this year’s Spring Festival period, it raised more than $700 million, with a valuation of $10 billion. It is now in the process of a new round of $1 billion financing.

At the beginning of 2025, DeepSeek burst onto the scene. With its high-performance, low-cost advantages brought by the MoE architecture and its fully open-source strategy, it sparked a wave both domestically and internationally, stirring up the entire industry. The impact on Kimi was direct and obvious. At that time, the company was facing industry criticism over “burning money to acquire customers,” and later Moonshot AI significantly cut its product placement budget.

According to QuestMobile data, Kimi App’s monthly active users fell from 21.653 million in Q1 2025 to 9.027 million in Q4 2025. By the end of 2025, even some tech media used “monthly active users are less than 5% of Doubao” as a standard to put Kimi into the discussion lineup of “the most fallen AI companies.”

At the beginning of 2026, the Lobster OpenClaw went viral. It allows all large models to be equipped with “hands and feet,” enabling AI to make a core leap from answering questions to completing tasks.

And what surprised the industry was that the one standing at the center of this wave—having taken in enormous dividends—was the long-dormant Kimi.

In February 2026, OpenClaw announced that it would set Kimi K2.5 as its official flagship model. This decision became the key turning point in Kimi’s comeback.

Kimi K2.5 was officially released and open-sourced—Moonshot AI’s “killer move” after shifting strategy. Built on a trillion-parameter foundation, it uses 15 trillion visual-and-text mixed tokens for native multimodal training, making important progress in areas such as agent intelligence, code generation, and visual understanding. A few weeks later, the Kimi official website also launched Kimi Claw, directly moving the agent entry point into the browser.

Previously, multiple media outlets reported that since the end of January this year, under the dual impact of the explosive popularity of the Kimi K2.5 model and Kimi Claw, it was revealed that Kimi’s total revenue in the past 20 days had already exceeded its total revenue for all of 2025.

According to data from global payments giant Stripe, the number of payment orders from Kimi’s individual subscription users surged 8,280% month-over-month in January 2026, and rose another 123.8% month-over-month in February. Kimi’s ranking on Stripe’s global leaderboard jumped from outside the top 100 to 9th place, becoming the first Chinese AGI product to enter the top ten of that list.

As reported by Jiemian News on March 30, just one month after the release of K2.5, Moonshot AI’s ARR (annual recurring revenue) had already exceeded $100 million, making it the first company among the “AI Six Little Tigers” (Zhipu AI, MiniMax, Baichuan Intelligence, Moonshot AI, Step Star, and Zero One All Things) to reach that milestone.

Worth noting is that after the launch of K2.5, Kimi’s overseas revenue has surpassed domestic revenue: overseas API revenue grew 4x, and the month-over-month growth rate of global paying users exceeded 170%.

A Bright Future Ahead

Now, CoreWeave has already gone public on the U.S. stock market, raising $1.5 billion. Anthropic’s valuation has soared to $380 billion, and it is planning a super IPO to raise more than $60 billion. OpenAI, with a valuation of $830 billion, is pushing forward with its listing plan. Global AI companies are collectively sprinting toward capital markets.

Meanwhile, Zhipu and MiniMax both listed on the Hong Kong stock market in January 2026. After listing, their stock prices continued to rise, and their market caps have remained stable at around $40 billion—demonstrating that the Hong Kong Exchanges and Clearing’s Chapter 18C (the listing mechanism for specialized technology companies) is a feasible channel for unprofitable AI companies. But the window for Hong Kong listings may not stay open forever.

Currently, besides overseas markets, there are also many domestic large model companies. How can Kimi stand out?

Speaking about Kimi’s strengths and weaknesses, senior columnist Ma Jipeng told reporters: “I think this large model is, at present, moving from general-purpose toward differentiation. It can roughly be divided into two tiers: one is the big internet companies, including Tencent, ByteDance, Alibaba, and Baidu; the other is startups, including Kimi, Zhipu AI, and so on. Compared to them, Kimi’s advantage mainly lies in its handling of long texts. In academia—and in some fields where there is strong demand for long-text processing—there is still substantial demand. Doubao’s influence on the consumer side, including its ability to process videos and images, might be even stronger.”

Regarding its development goals for 2026, Yang Zhilin also mentioned in a public letter at the end of 2025: “Achieve growth in revenue scale at a magnitude level; in product and commercialization, focus on Agent; do not take absolute user numbers as the goal, but pursue the ceiling of intelligence and create greater value in productivity.”

Yang Zhilin recently also said that starting from 2026, and in the coming years, the way AI R&D is conducted will undergo major changes: more of the research will be led by AI itself. Each researcher will be equipped with a large number of AI Tokens, and these AI Tokens can help them synthesize new tasks, synthesize new environments, help the company define what the best and most suitable reward function is in that environment, and even help the company explore what new network architectures could look like.

“Personally, I’m still optimistic about Kimi’s future prospects. Because its current operating performance is good, and its development momentum is also quite strong. This is a relatively good path to verticalize within its own domain, avoiding direct competition with these giants. Its total user base may not be that large, but by going deep into an industry and fully understanding it, there is still real potential,” Ma Jipeng said.

For Kimi, if it wants to achieve a high valuation in the market after listing, it must prove to the market that the growth in its Token consumption is sustainable and not dependent on short-term explosive surges in a single agent scenario; that its revenue growth can be translated into a continuous improvement in gross margin, rather than simply chasing volume by cutting prices.