Moonshot AI, a three-year-old Beijing frontier model lab, is preparing for a Hong Kong IPO that could raise around $3 billion. The company confidentially filed for the listing in September, with reports suggesting the debut could come as early as this year. The Chinese firm is also reportedly considering a dual listing on Shanghai’s STAR market.
Moonshot’s IPO plans in Asia add to a wave of AI-related fundraising globally. However, major U.S. AI companies have recently delayed their own debuts due to market uncertainty and concerns around AI safety. Moonshot’s listing will be a key market test for whether capital-market-funded Chinese labs can sustain frontier-tier valuations without an internet-giant parent behind them. Unlike Alibaba’s Qwen and ByteDance’s Doubao, which draw on parent-company balance sheets, Moonshot funds expensive operations through private equity and public markets.
Find a detailed analysis of Moonshot AI’s upcoming IPO below, drawing on insights from AlphaSense’s SuperAnalyst.
Related Reading: Top APAC IPOs to Watch in 2026
Company Overview
Moonshot was founded by former Carnegie Mellon doctoral researcher Yang Zhilin in 2023. The company reached global recognition in July 2026 with the launch of Kimi K3, an open-source LLM. Moonshot earns revenue by monetizing its Kimi model family, which spans a consumer assistant, a developer and enterprise API, and coding products. Analysts say that Moonshot’s valuation rests on Kimi K3, the largest open-weight model released to date, having closed the gap with the U.S. frontier labs while running cheaper. The company is now targeting a valuation of up to $50 billion in a final pre-IPO round ahead of the Hong Kong listing. Major investors in Moonshot include China’s two dominant internet strategics, Alibaba and Tencent.
Financial Performance
The company’s valuation has skyrocketed in the past year, from $4.3 billion in late 2025 to around $35 billion in a Series F funding round that closed in early July 2026. However, the company is not yet profitable, and its cost base is compute-heavy even as revenue grows.
According to media reports, the company’s annual recurring revenue topped $1 billion in August, up from $300 million in June. That jump was driven by the release of Kimi K3 in July 2026, which reached top performance benchmarks while costing significantly less than models from its U.S. peers. The company is targeting annualized revenue to reach $2 billion by the end of the year.
Market Competition
Moonshot competes against frontier labs globally and has closed much of the capability gap with U.S. leaders. However, it operates within a crowded field of Chinese open-source models that is driving down token prices. Moonshot is considered both a beneficiary and a cause of that commoditization, which threatens its own long-term pricing power.
Analysts bullish on Moonshot say that broker testing has placed Kimi K3’s capability at or near Claude Fable 5 and above GPT-5.6 Sol on coding benchmarks. The bear case is that Moonshot is not profitable and that heavy server and power costs plus low pricing will compress margins. The same Kimi K3 launch that drove Moonshot’s valuation higher also forced the company to pause subscriptions when demand exceeded capacity. In September, the FBI, NSA, and CISA released a joint advisory stating that Moonshot and other Chinese AI companies were using distillation campaigns to copy U.S. frontier models. The U.S. agencies accused the Chinese firms of exploiting American AI models to train their own systems.
What Experts Are Saying
From our experience, we saw that the Chinese model was cheap, first of all. It gave almost the same result as we were using [with] Claude, second of all, and students loved it. That's the second part. Another pattern that we noticed was that most students tend to have more conversation with the Chinese LLM than the Claude model.
There are some companies there that we are aware of such as Moonshot AI, and there's also DeepSeek, which is a Chinese company…I don't think they pose a real threat today because a lot of the work that we do with Liquid AI, the systems that are developed, the frontier models, and the neural networks from Liquid AI are deeply integrated in with the hardware. It would be rather difficult to replace that with another model, once the design is done and in production. Going forward into 2027 and 2028, it's always a threat, and it could impact Liquid AI.
We can get into it, but you can argue all these open-source type of models, like the DeepSeeks and recently this Moonshot AI and all that stuff. The Chinese model companies are trying to compete, but there are a lot of questions on can they even actually compete in the U.S.? Will they be allowed? They're not allowed in many use cases. I think China is probably the dark horse on the frontier model side because potentially they're probably the only other real competitor to Anthropic and OpenAI on the model side.
With AlphaSense's SuperAnalyst, an always-on AI agent purpose-built for financial and strategic workflows, you can run an in-depth analysis for any upcoming IPO. Start your free 2-week trial of AlphaSense today.
