Lab & Nation Watch 2026-09-23

Lab Watch: China's AI Labs Are Racing to IPO and Train on Domestic Chips at the Same Time

Moonshot AI filed confidentially for a Hong Kong IPO targeting a $50 billion pre-IPO valuation, while Zhipu (Z.AI) shipped a 300-billion-parameter flagship trained entirely on 100,000-plus domestic chips -- the same week a US NSA/CISA/FBI advisory named six Chinese labs for large-scale model distillation.

Moonshot AI filed a confidential A1 with the Hong Kong Stock Exchange this week, seeking roughly $3 billion with a possible Q4 2026 IPO, on the back of a final pre-IPO funding round reportedly targeting a $50 billion valuation. Separately, Zhipu AI (Z.AI) launched a 300-billion-parameter flagship model trained entirely on more than 100,000 domestic chips -- a genuine milestone for a fully domestic training pipeline given ongoing semiconductor export restrictions.

The regulatory backdrop is sharper than the product news: a joint advisory from the NSA, CISA and FBI, published September 8, named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI for what it called "industrial-scale model distillation campaigns" against Claude, GPT, Gemini and Grok dating back to at least late 2024, alleging billions of tokens were extracted. China's own regulators are separately reported to be investigating DeepSeek and Moonshot.

Underneath both storylines sits a real compute crunch: Zhipu, Moonshot and Alibaba are all reportedly restricting user access amid GPU shortages, even as one industry estimate puts China's open-source core LLM revenue at $60-80 billion, with total annual recurring revenue in the $100-150 billion range across the sector.

Three simultaneous China AI storylines this week -- IPO fundraising, a fully domestic 300B-parameter training run, and a US government distillation advisory -- read together as an ecosystem racing to prove commercial scale and technical independence at the same time it's under formal accusation of extracting that scale from US labs.