After raising $500 million, how far is Dark Side of the Moon from the real "safe zone"?

Amid the hustle and bustle of the large model race, Yang Zhilin and the Moonshot AI he helms are at an extremely delicate crossroads. In early 2026, a financing news stirred up the long - dormant primary market again, pushing Moonshot AI into the public eye. According to public reports, Moonshot AI completed a Series C financing of $500 million, and the post - investment valuation of the company will reach $4.3 billion (approximately RMB 31 billion). Founder Yang Zhilin announced through an internal letter that the company's cash reserve on the books has exceeded RMB 10 billion, making it a veritable "cash king" in the field of large models in China.

However, beneath the surface of "never short of money", shadows are quietly spreading. The latest third - party data shows that Kimi, which once ranked second in the industry, has seen its monthly active users (MAU) plummet from the peak of 36 million in October 2024 to the current level of 10 - 15 million. Meanwhile, tech giants such as ByteDance (Doubao), Tencent (00700.HK), and Alibaba (09988.HK) are gradually seizing the C - end AI market with their ecological and traffic advantages, and the rising star DeepSeek has become a new industry star with its "ultra - high energy - efficiency ratio".

Caught in the contradiction between "stockpiling resources" and "falling behind", and as Zhipu AI and MiniMax are vying for IPOs, Moonshot AI is facing a difficult choice.

From "Child of Fortune" to "Arbitration Vortex": The Controversial Background of the Founder

The story of Moonshot AI began with the ultimate belief in AGI (Artificial General Intelligence) held by the top minds in China's AI circle, but its growth path has also been accompanied by unprecedented legal disputes.

Company founder Yang Zhilin has a typical "genius resume". He graduated from the Department of Computer Science at Tsinghua University and obtained a doctorate from Carnegie Mellon University. He has worked at Google Brain and Meta AI, and participated as a core contributor in research projects such as Transformer - XL and XLNet, which have changed the course of AI. In March 2023, after ChatGPT took the world by storm, Yang Zhilin returned to China and founded Moonshot AI.

Different from entrepreneurs who were eager to find industry application scenarios at that time, Yang Zhilin showed an extreme "technological pressure" logic from the very beginning. He didn't talk about commercialization or vertical tracks but focused on long - text processing capabilities (Long Context). This obsession with underlying technology gave Moonshot AI a high technological premium in the early days.

However, while the company's valuation soared like a rocket, an equity dispute from the past has become the company's biggest concern. Public reports show that at the end of 2024, five old shareholders of Yang Zhilin's previous startup project, "Circular Intelligence" (including GSR Ventures and Wanwu Capital), formally filed an arbitration at the Hong Kong International Arbitration Centre (HKIAC). The core of the dispute lies in the suspicion that Yang Zhilin "evaded his responsibilities".

The old shareholders pointed out that when Yang Zhilin launched Moonshot AI, he did not obtain a formal written exemption from the board of directors of Circular Intelligence, violating the non - competition and non - interference agreements. What really triggered public outrage was that Zhang Yutong, a former partner of GSR Ventures (Yang Zhilin's senior schoolmate and an early investor), was accused of privately holding approximately 14% of the shares in Moonshot AI during her tenure as a director of Circular Intelligence without disclosing it to the fund.

This incident raised doubts from Zhu Xiaohu, the managing partner of GSR Ventures, who believed that such actions seriously damaged the interests of early investors. Although Moonshot AI later tried to handle the situation quietly through methods such as issuing additional shares or providing cash compensation and stated that "the arbitration does not affect the company's operations", this game about "the founder's credibility" has undoubtedly cast a shadow over the company's compliance.

The "Throne" of Valuation and Cash: The Anti - Consensus Financing Path and Core Advantages

Among the "Six AI Tigers", Moonshot AI has shown unique strategic determination. It is not only the fastest to raise funds but also the only leading player that has clearly stated it is "not in a hurry to go public".

From a valuation of $300 million in the angel round in June 2023 to $4.3 billion in the Series C round in early 2026, Moonshot AI's value has skyrocketed nearly 14 times in two years. In the Series A +/B financing in February 2024, the company received an investment of over $1 billion led by Alibaba. Alibaba's move was not only for computing power output (Alibaba Cloud) but also to secure a ticket in the AGI era.

When the market was worried about the AI bubble and financing became colder, Moonshot AI received a strategic investment led by IDG Capital, with Tencent (00700.HK) and Meituan (03690.HK) following, completing a $500 million Series C financing.

Moonshot AI's ability to gain the favor of many institutions stems from its in - depth technological exploration. The company's highlight moment came in March 2024 when Kimi announced support for long - text processing of up to 200,000 or even 2 million characters. The entire A - share market even saw the emergence of "Kimi concept stocks". Its long - text processing is not just a technical parameter but a qualitative change in user experience. Kimi has enabled AI to directly read an entire professional book, analyze thousands of pages of financial reports, or process complex code libraries. This superiority in "intelligence" has quickly helped Moonshot AI build a technological moat.

Founder Yang Zhilin has a clear development plan for Moonshot AI. AGI is a long - term, high - investment marathon. The current RMB 1 billion in cash on the company's books is not for dividends but for defense. It is estimated that the computing power investment required to train the next - generation model K3 will be several times that of Kimi. The billion - yuan reserve ensures that even if the primary market financing dries up, Moonshot AI will still have at least 24 months of "pure R & D cycle".

In the AI field, the annual salary of top algorithm engineers has soared to the million - dollar level. Moonshot AI's sufficient cash will be the only guarantee to maintain the morale of this "special force".

Ecosystem Encirclement and Traffic "Roller Coaster": The Fatal Ceiling of Independent Applications

However, the support of capital and the technological moat cannot hide Moonshot AI's traffic anxiety. As the technological dividend levels out, Moonshot AI is experiencing a cruel squeeze as the competition shifts from a "technology war" to an "ecosystem war".

Kimi's initial popularity was due to the scarcity of C - end AI products. But as of 2026, this advantage has disappeared. Alibaba's Tongyi Qianwen, Baidu's (09888.HK) Wenxin Yiyan, and the open - source leader DeepSeek have all achieved coverage of millions of tokens and have formed their own barriers in logical reasoning and mathematical ability through reinforcement learning. When "long - text" is no longer the only selling point, the attractiveness of Kimi as an independent entry point has begun to waver.

According to authoritative data from QuestMobile and others, Kimi's MAU (monthly active users) has dropped from a peak of 36 million to the current level of tens of millions, a decline of more than 50%. This phenomenon reveals a harsh reality: without an ecosystem, Kimi is finding it increasingly difficult to retain users in today's increasingly homogeneous competition.

In addition, Kimi's explosion in 2024 largely relied on monthly advertising investments of tens of millions or even hundreds of millions (including short - video information flows and offline advertising on Focus Media). However, as large companies enter the market and start a "price war", its customer acquisition cost (CAC) may have soared from the initial RMB 10 to over RMB 50. To preserve the "billion - yuan cash" for R & D, Moonshot AI has had to cut its marketing budget. After the budget cut, Kimi, which lacks social and office scenarios, has naturally seen its traffic stagnate.

It should be noted that compared with large companies, the disadvantage of Kimi lies not only in funds and technology but also in the ecosystem. Doubao, under ByteDance, with the unlimited traffic diversion from Douyin, has a stable MAU of over 160 million, has achieved a closed - loop of search, entertainment, and AI, and is trying new ventures in the hardware sector. Tencent Yuanbao is directly embedded in the WeChat and QQ ecosystems, and through functions such as "Search" and chat summarization, it has fragmented the usage scenarios that originally belonged to Kimi.

Among its competitors in the same track, Zhipu AI has officially applied for an IPO on the Hong Kong Stock Exchange and plans to use secondary - market funds to build a large - scale B - end ecosystem; MiniMax has focused on the overseas social market and achieved considerable cash flow, and its listing on the Hong Kong Stock Exchange is a foregone conclusion.

In contrast, Moonshot AI lacks distribution channels like DingTalk (Alibaba) or Enterprise WeChat (Tencent) on the B - end and traffic entry points like Douyin and WeChat on the C - end. Even with a billion - yuan cash reserve, Moonshot AI still has to rent computing power and buy traffic from large companies.

This $500 million Series C financing is not a luxury addition for Moonshot AI but more like a "strategic lifeline". The completion of the Series C financing shows that the capital market is still willing to bet on Yang Zhilin's technological intuition, but it also increases the pressure for an exit. If the K3 model fails to achieve a "generational - level" lead in the second half of 2026, then this RMB 1 billion in cash may only prolong its lonely journey before becoming an accessory to a large company.

Currently, Moonshot AI is caught in the contradiction between "extremely abundant cash" and "extremely anxious about traffic". For Yang Zhilin, avoiding an IPO is to avoid the short - term performance scrutiny of the capital market and pursue a pure technological singularity. However, the reality is that the competitive environment in China is far more fragmented and cruel than that in the United States where OpenAI operates.

In the long history of the Internet, there has never been a shortage of precedents of "technological pioneers" who eventually became "accessories to large companies". Will the billion - yuan - worth of resources piled up by Moonshot AI be the heavy hammer to break open the door to AGI, or will it gradually become a sunken island under the ecosystem encirclement of large companies? The answer does not lie in the RMB 1 billion on the books but in whether Kimi can find a reason for users to stay during the next technological iteration.

(This article is for reference only and does not constitute investment advice. The market is risky, and investment should be made with caution.)

This article is from the WeChat official account "FUSE", written by Wu Wei, and is published by 36Kr with permission.