Chinese artificial-intelligence startups MiniMax and Zhipu AI are racing to launch Hong Kong initial public offerings as early as January, in a bid to secure capital and market visibility amid intensifying competition in China's large-model sector, people familiar with the matter said.
Shanghai-based MiniMax, backed by Alibaba Group and Tencent Holdings, aims to file for an IPO that could raise several hundred million dollars, the people said, declining to be identified because the information is private.
Zhipu, also funded by Alibaba and Tencent, is targeting a listing around the same time, the people added. Details of both deals remain under discussion and could change, and the companies still require approval from the China Securities Regulatory Commission. Representatives for both firms declined to comment.
The dual offerings highlight the urgency among China's AI developers to secure funding as they compete with domestic technology giants and U.S. leaders such as OpenAI. China's AI market has drawn record sums but also mounting investor caution, with concerns that heavy spending on training infrastructure is outpacing clear commercial returns.
Hong Kong has emerged as a preferred venue for China's loss-making tech firms. The city's IPO market has raised about US$35 billion so far in 2025, putting it on track for a four-year high, according to Bloomberg data. A surge of interest in Chinese equities following the rise of DeepSeek in January has helped revive listing activity.
Bankers and analysts say the first major model developer to go public will effectively set a valuation benchmark for the industry, shaping expectations for subsequent deals. For AI companies with high cash burn, public funding is seen as critical to maintaining training cycles and scaling operations.
But share-price weak performance among several newly listed Chinese tech names last year has also made investors more wary of high-valuation AI issuers, raising the bar for MiniMax, Zhipu and their peers.
Behind the two startups are China's most influential internet groups. Alibaba and Tencent, which are developing their own foundational models (Tongyi Qianwen and Hunyuan), have been investing across the ecosystem to hedge technological bets and broaden their cloud-AI offerings. Their backing helps populate their marketplaces with competitive third-party models such as Zhipu's GLM and MiniMax's products, part of a broader push to build an “AI-era app store” to attract developers.
Investments also serve as a defensive move, preventing promising startups from being acquired by rivals and potentially yielding strong returns when the sector consolidates.
Industry studies suggest China's large-model market is shifting from a focus on raw performance to scenario-specific application value. According to research by consultancy Sullivan, enterprise-level model usage reached a daily average of 10,186.5 billion tokens in the first half of 2025, up 363% from the second half of 2024 — evidence that the market is moving from proof-of-concept to deployment.
Companies now face deeper commercial hurdles, including weak monetization pathways and the difficulty of integrating AI into existing corporate systems. Analysts say resolving these issues will heavily influence valuation prospects as the companies prepare to disclose detailed financials in public markets.
Listing will amplify competitive pressures. Firms that secure ample capital are expected to accelerate technology development and potential acquisitions, while those unable to list or that underperform after listing could face a cycle of funding strain, talent losses and slower innovation.
Market share data reflects rapid consolidation. Sullivan estimates that Alibaba's Tongyi, ByteDance's Doubao and DeepSeek account for more than 40% of total model-usage volume, indicating a growing concentration of resources, talent and compute capacity among top players.
The IPO contenders are pursuing distinct commercialization paths.
• Zhipu has combined open-source releases with sharp price cuts to expand usage and turn AI into a productivity tool across smartphones and computing platforms.
• MiniMax is pushing deeper into content-generation tools for film and television, though the business has long payback cycles and content-related risks.
• Moonshot AI, another major model developer, is focused on productivity applications but has yet to convert strong user feedback into broad commercial adoption.
Analysts expect 2026 to mark a defining year in real-world AI deployment as investors shift from evaluating “technology stories” to assessing revenue quality, margins and cash-flow visibility.
For China's large-model firms, the upcoming listings will mark a transition from rapid early-stage expansion to a more disciplined, market-driven era. The outcome of the MiniMax and Zhipu IPOs is expected to influence not only fundraising conditions but also the strategic direction of China's AI industry as it enters a new phase of consolidation and commercialization.
