Days after Kimi K3 jolted semiconductor stocks, Moonshot AI is accelerating plans for a Hong Kong IPO that could value the firm at over $30 billion, while Alibaba is pushing ahead with an open-weight rollout of its Qwen model.
Moonshot is targeting a public listing within six months, moving quickly to tap capital markets after K3 challenged long-held views about the gap between China’s AI sector and global leaders. Bloomberg reported that the company has already circulated a shareholder resolution seeking approval for the Hong Kong listing — a clear signal that an IPO could arrive before year-end.
At the same time, Moonshot is closing a new funding round that may lift its valuation past $30 billion, up from $20 billion in a Meituan-backed round in May.
The company’s growth metrics underscore the shift. Annual recurring revenue rose to $300 million in June from $200 million in April, reflecting strong demand for its products.
That demand has been particularly evident with K3. The company briefly halted new subscriptions over the weekend after usage exceeded capacity, with daily sales reportedly jumping more than sixfold since the model’s launch.
K3 is at the core of Moonshot’s momentum. The open-weight model has outperformed nearly all rivals — trailing only Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.6 on select benchmarks — and topped a widely watched coding test. Its release triggered a selloff in chip stocks on Friday, with spillover effects hitting crypto markets.
Moonshot’s push comes amid broader moves in China’s AI landscape. Alibaba said its Qwen3.8 model will also adopt an open-weight framework. The 2.4 trillion-parameter system is positioned as one of the most advanced globally, with the company claiming it ranks just behind Fable 5 among frontier models. A preview version, Qwen3.8-Max, is already available through Alibaba’s developer platforms.
Open-weight models allow users to run AI systems without paying usage fees, potentially weakening the pricing power of U.S. providers that rely on token-based billing.
Parameters — the internal variables adjusted during training — are commonly used as a rough indicator of a model’s scale. Modern AI systems now operate with billions or even trillions of these parameters.
Meanwhile, Bitcoin has increasingly traded in line with the AI investment cycle. Crypto miners have been repositioning themselves as data center operators for AI workloads, making their revenue outlook dependent on sustained demand for computing power.
The next key signal will come from upcoming earnings reports by Alphabet, Tesla, and Intel, which are expected to show whether AI capital spending is still rising — and whether companies tied to that trend, including crypto miners, can maintain their footing.
