French investors acquire a stake in Moonshot AI at a valuation of 30 billion US dollars: Kimi gets a bet from European capital before its IPO.

With Moonshot AI about to go public, a group of investors, predominantly French, have taken stakes in the company behind the Kimi model family. The funds were pooled through a special purpose vehicle (SPV) to inject approximately 10 million euros into the final pre-IPO funding round. One of the participants is Pyratz Corp. (MLPTZ), an investment company listed on the Access segment of Euronext Paris. Its CEO Bilal El Alamy explained to Trending Topics how this structure works and why the money is being invested in Beijing rather than Mistral.

The company stated that they invested at a pre-money valuation of 30 billion U.S. dollars for Moonshot AI. Among its publicly listed competitors, Z.ai is valued at 46.6 billion U.S. dollars, and MiniMax at 13 billion U.S. dollars. AI lab DeepSeek recently completed its first external financing round, raising around 7.4 billion U.S. dollars. Sources say its valuation exceeds 500 billion U.S. dollars.

What makes this bet reasonable is that Moonshot AI has significantly narrowed the gap with leading frontier models from the United States. Kimi K3 is the first open model to reach the 3-trillion-parameter level, matching Gemini 3.1 Pro on the Artificial Analysis Intelligence Index and outperforming Claude Opus 4.8, while its API price is only about one-thirtieth of that of Western frontier models. Among open-weight models, it ranks at the top, being the first model to reach the frontier tier instead of lagging behind by six to twelve months as in the past. The frontier-level quality that enterprises can download and host on their own is exactly what European buyers value.

A publicly listed company investing with its own balance sheet: Pyratz is not a traditional fund. "We are an investment company that draws some inspiration from the model of U.S. firms such as Pershing Square or Berkshire Hathaway," said El Alamy. "We are publicly listed ourselves, invest with our own balance sheet, cover all stages, but only invest in technology companies." The firm says it has built a portfolio of 29 companies since 2021, which have raised a total of more than 150 million euros in financing, including one unicorn, with six exits so far.

The path to investing in Moonshot AI originated from Pyratz's own product work. "We do a lot of R&D in-house, and collaborate with the technologies of the companies we invest in," El Alamy said. "We have been using Moonshot AI's models extensively, testing and deploying them, basically self-hosting, and they perform really well. So when the opportunity arose, we invested."

Investing 100,000 euros at a 30-billion-dollar valuation: Pyratz's official announcement did not disclose the amount. In the conversation, El Alamy was more specific: "We invested around 100,000 euros. We are an independent investment company using our own money, so we just contributed as much as we could." This financing round was completed in the summer, at a pre-money valuation of 30 billion U.S. dollars for Moonshot AI.

He used European comparisons to justify the price: "To be honest, in terms of the performance and revenue of their models, their revenue is twice that of Mistral, for example, this is a good investment."

How the SPV is structured: A single 100,000-euro investment cannot directly enter a financing round of this scale, so the investors joined forces. "We formed a syndicate with a group of other investors to purchase this stake," El Alamy said.

The vehicle was set up and managed by Techmind, a French investment intermediary operating under an investment studio model that packages deals for angel investors, family offices and enterprises. The SPV is named Techmind Moonshot AI. "The capital pool is around 10 million euros," El Alamy said. "Obviously, investing 10 million euros at a 30-billion-dollar valuation is negligible. The focus is on recognizing the performance of those Asian models."

He declined to disclose details of other participants, as this is a private SPV. He only said: "The participants are mainly French investors, mostly founders of French technology companies."

The shares themselves were purchased through Cathay Capital, a Sino-French investment company that is quite well-known in France.

In any case, Pyratz is in no hurry to sell. "We are basically long-term investors," El Alamy said. "We are publicly listed, which essentially allows us not to have to sell our shares when we don't want to. We are not the kind of fund that has to return capital at a certain point in time. People who hold shares in our company have liquidity themselves. So our strategy is to keep our best-performing investments."

The IPO is just around the corner: The timing of this investment was carefully considered. "Moonshot AI plans to list on the Hong Kong Stock Exchange in the first half of the year. That's why I call this a pre-IPO investment: the next financing round should be the IPO," El Alamy said.

According to the South China Morning Post, the company has secretly submitted its Hong Kong listing application, with the goal of going public as early as the first quarter of 2027. At the same time, the final financing round is underway, which could value Moonshot AI at up to 500 billion U.S. dollars. Recent reports also say the company is considering a dual listing, and plans to add a listing venue on the STAR Market of the Shanghai Stock Exchange, where many tech giants have listed recently. Trending Topics reported this summer that Moonshot AI targets to go public within six months after achieving the breakthrough of Kimi K3.

Why invest in Moonshot AI instead of Mistral: For French investors, the obvious question is why the money was not put into the local AI champion. El Alamy did not avoid the question: "From a pure model perspective, yes, I won't lie. I think if you ask ten people, nine of them would say the same."

He now believes the two companies have strengths in different areas: "What we have seen recently with Mistral is that since they cannot really compete on large models, they have started to diversify and move towards becoming a cloud service provider. They are deploying Chinese models on their own infrastructure, sacrificing their own models. What Mistral will be good at is the hosting part, the data centers. They are building a fleet of data centers where different models will be deployed."

He refused to frame this as a competition: "The strategy has never been to use only one model, but to use multiple models. So this is not against Mistral, it's basically complementary. We choose to deploy their models and Chinese models instead of American models."

Self-hosting is the real driving force: The strategic context is the shift to open weights. "63% of AI usage now goes through open-weight models, and this trend is definitely accelerating," El Alamy said, pointing out that regulated industries are the driving force: "Although providers like Anthropic and OpenAI say they won't absorb everything you do, they basically will. So for all regulated industries — finance, healthcare, insurance — people don't really use those models."

He dismisses claims that Chinese models pose security risks: "Some people might say: be careful, this is Chinese, not sovereign AI. For me, that's kind of nonsense. A model is formalized, it's a matrix of coefficients. There are no backdoors in it. Models can be biased, but all models have different biases. That's exactly why it's important to use multiple models, compare answers and challenge each other. From a cost-benefit perspective, Chinese models are really, really good."

Pyratz could run the publicly available Kimi model even without holding a stake, El Alamy admits. This deal is designed to open up the distribution side: "One feature of our structure is that when we invest in a company, we try to help it better deploy its solutions in France. We are building a joint product that allows people to deploy these models on their own."

"Open-weight strategies can compete at the frontier" Beyond that, Pyratz wants to combine Moonshot AI's open weights with the technology of portfolio company Zama. Zama works on fully homomorphic encryption, which allows computations to be performed on encrypted data without exposing the underlying information. In the official announcement, El Alamy put it this way: "Moonshot AI shows that a focused team, strong architecture and an open-weight strategy can compete at the frontier without matching the spending levels of the largest U.S. labs. This investment also supports a broader goal: to explore sovereign AI infrastructure where powerful open-weight models work in synergy with confidential computing technologies."

Founded in 2023, Moonshot AI has around 300 employees and has raised approximately 5 billion U.S. dollars since its establishment, far less than leading U.S. labs. Its latest flagship model Kimi K3 has 2.8 trillion parameters and a 1-million-token context window. According to Pyratz, around 70% of revenue comes from enterprise API usage, with the rest coming from paid consumer subscriptions.

This article is from the WeChat official account "Tech Business", and is republished by 36Kr with authorization.