Moonshot’s executives are currently engaged in negotiations to settle a high-profile legal dispute in which one of its shareholders is demanding compensation over claims of misconduct, sources close to the matter have revealed.

Specifically, Recurrent AI is accusing the executives, Yang Zhilin and Zhang Yutao, of misconduct. Sources have pointed out that, in a March 11th letter to Moonshot AI investors, Recurrent AI’s legal representative claimed that the pair misled investors and profited at the firm’s expense. Recurrent is now demanding roughly $700 million in compensation.

This legal tussle, however, is escalating at a crucial time. Moonshot is still preparing for its much-hyped Hong Kong IPO.

Recurrent accuses Yutao of concealing shares he received from Moonshot

The parties have been locked in the dispute since 2024, with the case now in arbitration in Hong Kong. Earlier, MinterEllison, the legal powerhouse representing Moonshot and its executive team, had signaled its intent to fight the lawsuit.

In November 2024, senior partner David Morrison told Chinese media that the case lacked any supporting facts or legal substance and vowed to mount a robust defense.

Recurrent’s March letter, nonetheless, further set out a series of allegations involving infrastructure and data handling. It claims that, in April 2023, servers hosting critical large-language-model data were transferred from Recurrent to a Moonshot subsidiary without authorization.

Furthermore, the letter alleges that the company’s Chief Technology Officer, Zhang Yutao, instructed the deletion of other critical information in September 2024. In addition to the above, the letter noted that Yutao received a substantial stake in Moonshot and failed to disclose it to Recurrent’s shareholders. 

Both Yang Zhilin and Zhang Yutao were partners in Recurrent AI, the company they founded back in 2016. However, they created Moonshot in 2023 and currently serve as directors at Recurrent.

Initially, Recurrent had acquired 9.5% of the newly formed firm’s total shares. But now, Recurrent and its investors seek compensation equal to 10.5% of the total shares, worth more than $700 million, given the company’s $30 billion valuation at the time of its June funding.

Moonshot targets a $50 billion valuation with potential IPO listing

As earlier reported by Cryptopolitan, Moonshot AI has set its sights on an initial public offering in early 2027. The startup, which filed initial paperwork confidentially in September, could see its valuation reach $50 billion as it structures a deal to raise up to $5 billion on the Hong Kong stock exchange.

Though the legal clash is unlikely to derail the listing entirely, an unresolved battle with Recurrent introduces distinct risks. The lingering conflict could trigger flags on compliance, corporate governance, and ownership transparency, ultimately complicating the regulatory vetting process for incoming public shareholders, one source said.

For a company preparing to go public, resolving disputes over ownership and corporate governance can be an important part of preparing for regulatory scrutiny. Prospective investors typically assess a company’s ownership structure, financial position, legal exposure, and management practices before committing capital.

An unresolved arbitration case could therefore prompt additional questions during the listing process, particularly if the outcome affects Moonshot’s shareholding structure or exposes the company to substantial financial liabilities. A settlement could reduce some of that uncertainty, but it would not automatically guarantee regulatory approval or investor demand for the offering.

The company’s IPO plans, nonetheless, remain contingent on regulatory approval, with the target proceeds and other financial details potentially changing depending on market conditions, according to insiders.

Moonshot initially built its reputation with Kimi, which gained attention for its ability to work through long documents. In July, Moonshot launched a new model, Kimi K3. So far, the 2.8 trillion-parameter system holds the crown as the world’s largest open-weight AI model and achieves performance metrics that approach the level of Anthropic’s proprietary models.

The model has drawn favorable reviews and significant demand. Sources say the startup is discussing revenue-sharing arrangements with Microsoft, Amazon, and Google that would enable the companies to host the model. Such an agreement could open a new chapter in business relations between Chinese AI firms and major U.S. cloud companies, with revenue sharing at its core.

If you're reading this, you’re already ahead. Stay there with our newsletter.