As the leading contenders in the AI field race to list on the Hong Kong Stock Exchange, the industry's competitive focus has quietly shifted. The allure of technological innovation is waning, and the ability to create tangible value has emerged as the new criterion for success.

Content/Huanlao

Editor/Yonge

Proofreading/Mangfu

On December 11, market insiders disclosed that two domestic AI unicorns, MiniMax and Zhipu AI, are planning to proceed with Hong Kong IPOs in the near future. Meanwhile, Yuezhi'an is exploring a backdoor listing on the Hong Kong Stock Exchange in a bid to become the first domestic large model stock.

Bloomberg provided further details, stating that MiniMax could go public as early as January of the following year, aiming to raise hundreds of millions of dollars. Zhipu AI is also targeting a similar timeframe, having shifted its focus from listing on the mainland China stock market to Hong Kong.

Although none of the companies involved have commented on the IPO rumors, a market consensus has formed that this battle for the first large model stock is more than just a race for a stock code. It represents a crucial turning point for the industry's transition from technological hype to practical value realization.

The first company to cross the finish line will set new valuation benchmarks and competitive rules for the entire sector.

Part.1

Evolution of Capital Logic

A Multidimensional Chess Game: From Financial Investment to Ecosystem Competition

The urgency to be the first to list stems from the competition for scarce capital market attention and pricing power during a critical window. For AI companies still in the heavy investment phase, public fundraising is not merely a necessity to stay in the technological race but also a vital arsenal for the next stage of scaled expansion.

However, while the Hong Kong Stock Exchange provides a listing avenue for unprofitable tech firms, the underperformance of several mainland tech stocks last year has made investors increasingly wary of high-valuation, high-loss models. The first company to succeed in listing will set a benchmark valuation anchor for the industry, directly influencing subsequent players' financing expectations and bargaining power.

What warrants deeper scrutiny is that behind MiniMax and Zhipu AI stand strategic investors Alibaba and Tencent—giants with their own powerful proprietary models (Tongyi Qianwen and Hunyuan). Their substantial investments are far from being mere financial bets; they represent a sophisticated blend of ecosystem competition and strategic hedging.

Large model technology is still in its early explosive phase, with future mainstream technical routes—such as architecture, training methods, and multimodal fusion—yet to be determined.

Proprietary models are the cornerstone and foundation for these giants, prioritizing stability, controllability, and strong alignment with their core businesses. However, their development processes may lean more toward industrialization, with high innovation trial-and-error costs and long cycles. Investing in external star teams is akin to placing bets on different, potentially more aggressive 'technical blind boxes' outside the mainstream track.

Startups in the large model sector are often led by top scientists, unconstrained by existing businesses, and may achieve breakthroughs in specific technical areas. Through investment, giants acquire options on potentially disruptive future technologies, ensuring they remain competitive regardless of which technical route succeeds.

Moreover, a healthy ecosystem cannot rely solely on 'domesticated' species but needs competitive and diverse wild forces. Introducing high-quality third-party models like Zhipu GLM and MiniMax significantly enriches Alibaba Cloud and Tencent Cloud's AI market offerings, enhancing their appeal to developers. Essentially, this builds an AI-era app store, leveraging ecosystem prosperity to counter single-product competition.

Investing in top startups is also a key defensive move to prevent them from falling into rival camps. Furthermore, AI is a capital-intensive industry, where early investments can yield substantial financial returns. More importantly, as the industry enters a consolidation phase, giants, as major shareholders, will occupy highly advantageous positions in potential business collaborations, data sharing, and even mergers and acquisitions.

Thus, giant-driven capital propels invested companies to act as both pathfinders and collaborative forces within their ecosystems. Even if proprietary development and external investments potentially compete, the gains from collaborative dominance far outweigh internal friction.

Part.2

Tightening IPO Window

How Will the Industry Landscape Diverge Post-IPO?

Frost & Sullivan's report provides crucial context for this IPO wave. In the first half of 2025, the daily average invocation volume of enterprise-grade large models in China reached 101,865 billion tokens, surging approximately 363% from the second half of 2024. This data not only confirms explosive market demand but also marks a fundamental shift: large models are moving from technological validation to scaled deployment.

The logic of industrial competition is being reshaped. Companies' focus has shifted from 'ultimate performance races' to 'balancing scenario fit and commercial value.' Decision-making now prioritizes 'finding optimal solutions for specific business scenarios' over 'pursuing the single strongest model.'

This transformation is particularly evident in changes to core industry pain points. The once-dominant issue of 'high training and inference costs' has been supplanted by more fundamental challenges: 'unclear application scenarios generating real value' and 'difficulty integrating with existing business systems.'

This clearly indicates that the industry has moved beyond technological novelty and entered the deep waters of commercialization. Success in addressing these deeper issues will directly determine the ultimate outcome of IPO narratives.

If successfully listed, AI unicorns will enter a radically different arena with transparent financial disclosures, regular performance reviews, and global investor scrutiny, forcing them to shift from telling technological stories to proving commercial capabilities.

Going public will accelerate industry differentiation. Companies that secure sufficient resources can further expand technological advantages, pursue strategic acquisitions, and refine industrial layouts. Those that fail to list or underperform post-IPO risk falling into a vicious cycle of 'financing difficulties - talent loss - technological stagnation.'

Frost & Sullivan's data reveals a harsh reality of concentration: in model invocation volume, Alibaba Tongyi (17.7%), ByteDance Doubao (14.1%), and DeepSeek (10.3%) collectively account for over 40%, with industry concentration accelerating.

This 'Matthew effect' manifests simultaneously across capital, talent, data, and computing power. Giants' ecosystem investment strategies further solidify this top-heavy pattern, incorporating the most promising external innovations into their growth trajectories through capital ties.

For numerous small and medium-sized AI firms, survival space is rapidly shrinking. Focusing on vertical domains like healthcare, law, and finance to build deep industry knowledge barriers has become a pragmatic—though capped—survival path.

Part.3

Commercialization Capabilities

The Ultimate Watershed Determining the Future

Successful listing merely grants entry to the next competition phase. The true test lies in converting financing and valuation advantages into sustainable, scaled commercial strengths.

Future winners will need to simultaneously possess multiple capabilities: deep and sustained technological innovation, clear and replicable commercialization paths, strong ecosystem integration abilities, and profound understanding of specific industry processes.

Currently, IPO contenders exhibit differentiated explorations. Zhipu AI employs an open-source strategy combined with significant price reductions, demonstrating cost control and technological confidence while advancing AI toward genuine productivity tools capable of operating smartphones.

MiniMax dives deep into film and television content creation, attempting to build a commercial closed loop from foundational technology to content output, though it faces dual challenges of return cycles and content risks.

Yuezhi'an continuously optimizes product experiences, focusing on efficiency gains in office and learning scenarios. However, its market share indicates that the path from excellent product to broad commercial success still requires breakthroughs.

As AI deployment reaches a critical inflection point, 2026 may well become the industry's true watershed. The technological aura is fading rapidly, and tangible value creation will become the ultimate yardstick for measuring enterprises.

For IPO-bound unicorns, the test is twofold: short-term, they must prove commercialization potential and valuation rationality to capital markets; long-term, they must find irreplaceable strategic niches within ecosystems dominated by giants—whether as absolute experts in vertical domains or key components of universal platforms.

For China's large model industry, the bell ringing at the Hong Kong Stock Exchange will mark a highly symbolic moment. It signals the end of the industry's era of maverick heroes and the beginning of its market-driven, capitalized coming-of-age.

This IPO race is not just about seizing a time window but defining qualifications for the next industry phase.