Why This French Investor Is Betting on China’s Moonshot AI at $30 Bn Valuation

Pyratz Corp, an Euronext-listed AI-native venture builder, has invested in Moonshot AI, the Chinese AI lab behind Kimi AI models, at a $30 billion valuation.

In an exclusive interview with AIM, Pyratz CEO Bilal El Alamy says the company “federated a syndicate of $10 million” for the investment. He adds that the investment reflects a broader bet on open-weight AI, self-hosted models, and a more decentralised AI ecosystem.

Unlike a traditional venture capital fund, Pyratz invests from its own balance sheet rather than managing capital on behalf of third-party investors, giving it greater flexibility over where it invests. This gives it greater flexibility over where it invests, including in companies and regions that may fall outside the usual constraints of a traditional fund.

El Alamy did not disclose Pyratz’s exact stake, saying the investment thesis matters more than the size of the ticket.

Moonshot AI has raised $7.27 billion to date, according to Tracxn. The company’s largest fundraise was $3.5 billion in July this year.

Why Pyratz Chose Moonshot

Pyratz was already a user of Kimi internally. El Alamy says the company has been building products to help entrepreneurs test and grow ideas faster and has integrated Kimi into those products.

The biggest attraction, however, is the economics of running the model.

“What we like about Moonshot is that it’s the cheapest model to self-host,” El Alamy says.

He compares the infrastructure cost of self-hosting an OpenAI model with Kimi. According to him, running an OpenAI model could cost around $30,000 a month in infrastructure, while Kimi can cost less than $5,000.

“Models like Kimi that are open-weight, efficient in terms of size and token consumption are very good models to start experiencing with AI,” he notes.

The company’s latest model, Kimi K3, has also made Moonshot a more serious commercial contender. Released in July, the 2.8-trillion-parameter open-weight model posted competitive results against leading proprietary models in third-party evaluations.

Moonshot is also reportedly in talks with Microsoft, Amazon and Google to host K3 on their cloud platforms, potentially creating a new distribution channel for a Chinese open-weight model outside China.

The investment also builds on Pyratz’s existing AI portfolio. Pyratz is also a shareholder in Paris-based AI startup Mistral AI, but El Alamy described the Moonshot investment as complementary rather than competitive.

“Mistral is developing a neocloud business where they’re open to basically hosting any type of open-weight model,” he says. He sees the Moonshot investment as complementary to that business.

Kimi Has $1B ARR

El Alamy argues that Moonshot is undervalued relative to its commercial performance.

“Kimi is undervalued because Kimi has really amazing revenue KPIs,” he says, adding that Moonshot generates more revenue than Mistral while having a similar valuation.

El Alamy shares that Kimi has reached $1 billion in annual recurring revenue. “The thesis is that open-weight models are going to become more and more mainstream,” he adds.

He believes open-weight models will bring greater decentralisation to AI power because companies can run them on their own infrastructure rather than depending entirely on providers such as OpenAI and Anthropic.

El Alamy also points to growing usage of open-weight models, claiming that open-weight models accounted for 63% of AI tokens in the previous month.

Moonshot is also continuing with plans for a public listing in Hong Kong. El Alamy says the company is targeting the first half of 2027 for the IPO.

Moonshot confidentially filed for a Hong Kong IPO in September, targeting a potential $3 billion raise. The company was valued at around $50 billion at the time, after raising more than $5.5 billion, Reuters reported.

For Pyratz, the Moonshot investment combines a public-market opportunity with a longer-term thesis around open-weight models, self-hosted AI and a more decentralised AI ecosystem.

El Alamy Pushes Back on Calls to Slow AI

El Alamy dismisses calls from AI leaders including Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman for a slowdown in AI development, describing them as a “marketing gimmick”.

He argues that the industry should prioritise expanding AI adoption rather than spending heavily to push model performance further. According to El Alamy, only around 5–7% of people globally are actively using AI.

He also acknowledges the risks associated with open-weight models, including the possibility of malicious actors using them to develop biological weapons. But he says that concentrating AI capabilities among a small number of companies presents its own risks.

“We don’t want to have one entity or two entities that centralise all the AI power in the world,” he notes.

How Chinese AI Labs are Dealing With Compute Constraints

The investment also gives Pyratz exposure to China’s AI ecosystem at a time when access to advanced NVIDIA GPUs remains constrained by US export controls.

El Alamy says Chinese and other Asian AI companies have responded by becoming more efficient in how they train models.

He also says Chinese AI companies are renting compute from neocloud providers, allowing them to access infrastructure without directly owning the chips.

“They cannot buy chips, but they have access to compute,” he says.

El Alamy expects compute availability to increase as countries invest in energy infrastructure and data centres. He suggests Chinese AI companies could increasingly work with infrastructure providers across Japan, Singapore, Malaysia, Indonesia and India.