Alibaba Group has received confirmation from the Hong Kong Stock Exchange that it may proceed with a proposed spin-off and separate listing of its chip subsidiary T-Head, moving one of China’s leading in-house semiconductor developers closer to the public market.

T-Head confidentially submitted a listing application form to the exchange on September 11, Alibaba said in a November 10 filing. The planned transaction would involve a global offering of T-Head H shares, including a Hong Kong public offer and an institutional placement.

The filing did not specify the size, valuation or timing of the offering. Alibaba said the deal remains subject to regulatory and listing approvals, market conditions and final decisions by the companies. T-Head would remain an Alibaba subsidiary after the proposed listing, although Alibaba’s final ownership stake has not been determined.

Founded in 2018, T-Head develops processors and other semiconductor intellectual property for cloud computing, artificial intelligence and connected devices. Its products include RISC-V-based processors and the Zhenwu family of AI accelerators used within Alibaba’s infrastructure.

Alibaba has made chips a central part of its effort to build a more self-contained AI technology stack. In September, the company outlined a roadmap that pairs its Qwen models with T-Head processors and large-scale data-center expansion. A separate listing could give the chip business access to public capital while preserving its operational links with Alibaba’s cloud and AI units.

The proposal also reflects efforts by Chinese technology companies to expand domestic computing capacity amid limits on access to some advanced foreign chips and manufacturing equipment. T-Head designs processors but relies on external foundries and supply-chain partners for production.

Alibaba said it expects the spin-off to improve T-Head’s profile, governance and access to capital, while allowing investors to value the semiconductor business separately.

Alibaba targets 20GW global data center capacity by 2032, maps next Qwen and AI chip roadmap