On the eve of capital ebb, how did Moonshot AI pull off its "moon landing" feat?
In July 2026, large model startup Moonshot AI officially released Kimi K3, a new open-source model with 2.8 trillion parameters. The launch of this product has drawn widespread market attention to Moonshot AI's technological progress and commercialization.
Apart from technological iteration, Moonshot AI is also very active in financing activities in the primary market, and its valuation has risen steadily as the financing rounds increase. Recently, there have been multiple market rumors that Moonshot AI is preparing for an IPO with a relatively high target valuation.
However, while the popularity of the primary market is rising, the large model sector in the secondary market is undergoing value revaluation. Listed large model enterprises such as Zhipu (02513.HK) and MiniMax (00100.HK) have generally seen a sharp correction in their market capitalization recently due to the impact of share unlocking. Under the interplay of capital expectations and industry realities, Moonshot AI's high valuation logic, supply chain carrying capacity, and the objective environment of the listing window period may constitute important dimensions for analyzing the company's future development path.
Financing Process and Valuation Evolution: IPO Expectations Driven by Capital
Since 2026, both the financing scale and valuation level of Moonshot AI in the primary market have shown significant changes. Public information shows that Moonshot AI started financing with a valuation of 4.3 billion US dollars at the end of 2025, and completed three consecutive rounds of financing totaling about 19 billion US dollars from January to February 2026.
At this stage, Moonshot AI not only received support from existing shareholders, but also obtained additional investment from old shareholders including Alibaba (09988.HK), Tencent Holdings (00700.HK), 5Y Capital, and Andon Health (002432.SZ), and its post-money valuation quickly rose to 180 billion US dollars.
In May 2026, the company further completed the Series D financing of about 20 billion US dollars led by Meituan (03690.HK), and its post-money valuation exceeded 200 billion US dollars. From June to July, during the new round of financing, Moonshot AI's pre-money valuation has reached 315 billion US dollars.
Market sources also pointed out that Moonshot AI plans to launch Pre-IPO round negotiations in August, with a target pre-money valuation as high as 500 billion US dollars (equivalent to about 3.4 trillion RMB). According to incomplete statistics, as of July, Moonshot AI's total cumulative financing amount within the year has exceeded 39 billion US dollars.
With large-scale financing, market discussions about Moonshot AI's planned IPO have gradually increased. It is reported that Moonshot AI plans to be listed on the Main Board of the Hong Kong Stock Exchange as soon as within 6 months, expected from the end of 2026 to the beginning of 2027, in accordance with Chapter 18C of the Hong Kong Stock Exchange's rules for specialist technology enterprises.
According to reports, to promote the listing process, Moonshot AI may have started preliminary consultations with China International Capital Corporation (03908.HK) and Goldman Sachs Group (NYSE: GS) on the potential sponsor role, and notified investors that it plans to basically complete the restructuring of the overseas red-chip VIE structure by the end of July.
Judging from Moonshot AI's coping strategies, at the end of 2025, Moonshot AI firmly denied the rumor of "backdoor listing", emphasizing that the company is in no hurry to go public in the short term, but in response to the recent listing rumors, Moonshot AI's official side has not yet responded.
Against the background of increasing computing power costs and peers entering the capital market one after another, taking IPO as a means of obtaining long-term financing in the next stage is in line with the rational choice of current large model enterprises on the capital side.
Business Restructuring and Competition Pattern: Commercialization Exploration Under Technological Iteration
The improvement of valuation essentially requires underlying business data as support. On July 17, 2026, Moonshot AI released the Kimi K3 model, which showed industry competitiveness in some technical indicators. The model adopts a highly sparse MoE architecture with 2.8 trillion parameters, which can efficiently activate 16 out of 896 experts.
In the intelligence index released by the evaluation agency Artificial Analysis, Kimi K3 scored 57 points in total, ranking third in the world, slightly lower than Claude Fable 5 (60 points) and GPT-5.6 Sol (59 points). In the demonstration of full independent semiconductor chip design capabilities, the model also triggered short-term fluctuations in the secondary market stock prices of global EDA enterprises such as Cadence (NASDAQ: CDNS) and Synopsys (NASDAQ: SNPS).
However, in terms of user structure, the strategic focus of Moonshot AI is shifting. QuestMobile data shows that as of the first quarter of 2026, the monthly active users of the Kimi client are about 8.338 million, ranking ninth among domestic native AI applications. By comparison, as of June 2026, ByteDance's Doubao had 383 million monthly active users in the same period, and Alibaba's Qwen also exceeded 167 million.
However, the change in Moonshot AI's user activity data is not purely a competitive disadvantage, but partly an objective reflection of the company's active reduction of traffic investment budget and change of C-end customer acquisition strategy.
While C-end traffic is shrinking, Moonshot AI's annualized recurring revenue (ARR) on the B-end is showing a relatively fast growth trend. Relevant data shows that Moonshot AI's ARR crossed the threshold of 100 million US dollars in March 2026, and reached about 300 million US dollars by mid-June.
At present, B-end API call revenue may account for more than 70% of Kimi's total revenue, and its business model is gradually transforming into productivity tools and large model infrastructure. In terms of pricing, the hybrid API pricing of Kimi K3 is about 2.3 US dollars per million tokens (3 US dollars for input / 15 US dollars for output). Compared with some domestic companies adopting low-cost open-source strategies such as DeepSeek, the company's pricing system shows a tendency to maintain technical premium.
Market Pressure and IPO Window Period: Secondary Market Value Revaluation Under Heavy Burden
Despite the expansion in technology and commercialization, Moonshot AI's long-term development may still be restricted by the macro supply chain and secondary market environment.
In terms of computing power demand, the inference business of the 2.8 trillion parameter model requires high marginal computing power support. Only three days after the release of Kimi K3, the company suspended new user subscriptions on the C-end due to tight computing power resources, reflecting the objective limitation of the company's underlying hardware reserves when facing sudden traffic.
Under the background of export control of overseas high-end chips, domestic enterprises are trying to accelerate the process of domestic substitution through solutions such as "super node". However, domestic chips are still in a long-term running-in stage in links such as CUDA ecosystem compatibility, advanced lithography, and packaging (CPO/HBM), and the mismatch between high parameters and weak computing power may still be the systematic challenge Moonshot AI is currently facing.
At the same time, competition in the large model field requires continuous capital investment. Tech giants such as Microsoft (NASDAQ: MSFT) and Google (NASDAQ: GOOGL) are expected to have a combined capital expenditure of more than 650 billion US dollars in 2026, and Alibaba's AI investment plan has also reached 53 billion US dollars. In the capital expenditure cycle dominated by giants, startups are facing great capital turnover pressure in the long-term computing power consumption war.
With the reduction of industry expectations, the valuation logic of the secondary market for unprofitable large model enterprises is also changing. Zhipu and MiniMax, which were previously listed on the Hong Kong Stock Exchange, both saw their stock prices correct after the lifting of restricted shares in July 2026.
Zhipu's market capitalization briefly rose at the initial stage of listing due to the impact of only 2.67% of the floating share, but after the lifting of the ban, superimposed on the impact of competing product releases, its market capitalization shrank sharply in a short period of time; MiniMax's stock price also plummeted, falling below 100 billion Hong Kong dollars and approaching the issue price. The stock price trend of the pioneers has reflected that high valuations without solid financial fundamentals show great vulnerability in a market with increasing supply and limited liquidity.
At the current stage when the overall valuation center of the large model sector in the Hong Kong stock market is facing revaluation, whether Moonshot AI's high primary market valuation expectation can obtain sufficient capital support in the secondary market may face a relatively harsh market test. In the current environment of hard supply chain constraints, capital suppression by overseas giants and intricate peer competition, how Moonshot AI balances valuation, financing and capital investment needs within the increasingly narrowing listing window period will be the key to determining whether the company can "reach the moon".
This article is from the WeChat official account "Investor Network - Thinking Finance", author: Wu Wei, authorized for release by 36Kr.