January 8 and 9, 2026, are destined to be written into the history of China’s artificial intelligence development. In just two days, Zhipu AI and MiniMax successively listed on the Hong Kong Stock Exchange, not only creating the capital-market myths of the “world’s first listed large-model stock” and a “trillion-level market capitalization,” but also marking China’s large-model industry’s formal entry—from the era of wild-growth technical competition—into the deep-water zone of commercialization validated by capital. As the two companies stood under the spotlight with market capitalizations of HKD 69.8 billion (approx. USD 8.95 billion) and HKD 106.7 billion (approx. USD 13.68 billion), the entire industry has been thinking: behind this capital feast, what kind of industrial upheaval does it foreshadow?
The Divergence of Technical Routes: A Contest Between the “Academic School” and the “Product School”
The listing paths of Zhipu AI and MiniMax reflect two typical archetypes in China’s large-model ecosystem. Originating from Tsinghua University’s Knowledge Engineering Laboratory, Zhipu AI carries a distinctive “academic-school” and “national-team” flavor; its chairman, Liu Debing, reiterated in his listing remarks the long-term pursuit of Artificial General Intelligence (AGI). The company adopts the MaaS (Model-as-a-Service) model, outputting intelligent capabilities to developers and enterprises through API calls, and has already empowered 12,000 enterprise customers worldwide, more than 80 million terminal user devices, and over 45 million developers.
In sharp contrast, MiniMax—founded only four years ago—shows a more engineering-oriented and product-driven character. Founder and CEO Yan Junjie comes from the front lines of industry; from the day it was established, the company has adhered to a dual-track strategy of “foundation model + self-developed applications,” emphasizing rapid validation of technology in real scenarios and global expansion. With its self-developed MoE (Mixture of Experts) + Linear Attention architecture, MiniMax has built a fully multimodal model system with global competitiveness; its consumer products already cover more than 200 countries and regions worldwide, and overseas revenue accounts for more than 70%.
The divergence of these two paths, in essence, reflects different choices Chinese large-model companies make between technological ideals and commercial realities. Zhipu chose a steady B2B route, striving to become the operating system of the AI era by building a computing-power adaptation system that spans “cloud–edge–endpoint”; MiniMax chose a more adventurous C-end globalization route, seeking a breakthrough in the consumer market through hit products such as Talkie and Hailuo AI.
A Cool-Headed Reflection Under the Capital Frenzy: The Paradox of High Valuation Coexisting With Losses
Although both companies performed brightly in the capital market, the financial data revealed in their prospectuses shows a harsh reality: China’s large-model industry is still in an early stage characterized by “high growth coexisting with high investment.”
From 2022 to 2024, Zhipu AI’s revenue grew from RMB 57.40 million (approx. USD 8.24 million) to RMB 312 million (approx. USD 44.77 million), with a compound annual growth rate as high as 130%, yet its net loss expanded from RMB 144 million (approx. USD 20.66 million) to RMB 2.958 billion (approx. USD 424.41 million). In the first half of 2025, the proportion of R&D expenses to revenue was as high as 835.4%, of which computing-power procurement accounted for 71.8%. This means that for every RMB 1 of revenue Zhipu earns (approx. USD 0.14), it needs to more than RMB 8 in R&D (approx. USD 1.15+).
MiniMax’s situation is likewise not optimistic. The company’s cumulative losses from 2022 to 2024 were about RMB 9.3 billion (approx. USD 1.33 billion); its net loss in 2024 reached RMB 3.27 billion (approx. USD 469.18 million); and its monthly cash burn in 2025 is estimated at nearly RMB 2.0 billion per month (approx. USD 286.96 million per month). Although its revenue in the first three quarters of 2025 increased by more than 170% year-on-year, high-intensity R&D investment remains the main reason for its losses.
This paradox of high valuation coexisting with huge losses reflects a fundamental shift in the capital market’s valuation logic for large-model companies. Traditional PE and PB valuation tools have already failed when faced with large-model companies that sustain losses while having an extremely high share of intangible assets; investors are more willing to pay for technological assets and ecosystem potential. Zhipu’s IPO corresponds to a price-to-sales ratio of about 147x; MiniMax, by virtue of its globalization narrative, obtained an even higher premium. This reconstruction of the valuation framework marks the maturation of the capital market’s understanding of AI companies.
The Reshaping of the Industry Landscape: From “Hundred-Model War” to Concentration at the Top
The successful listings of Zhipu and MiniMax are not only milestones for the two companies, but also a watershed for the entire Chinese large-model industry. Since 2025, the “Matthew effect” in the foundation-model field has basically become apparent; the once-feverish competition of “a hundred models blooming” has gradually come to an end, and resources, users, and capital are further concentrating toward leading enterprises.
According to statistics from public data, in 2025, domestic “model-layer” AI companies completed a total of 22 investment events, with a total disclosed amount of RMB 9.416 billion (approx. USD 1.35 billion), down 8.3% and 52.9% respectively compared with 2024. The number of investment events and the amount as a share of the overall AI industry fell significantly: the former dropped from 4.3% to 2.2%, and the latter from 51.0% to 14.4%. Behind this cooling is the basic stabilization of the competitive landscape among the top tier, as well as the forceful entry of internet giants such as ByteDance, Alibaba, and Tencent.
The success of listing provides Zhipu and MiniMax with strong capital advantages, but it also accelerates industry differentiation. Listed companies will form a Matthew effect in business expansion, financing costs, and talent acquisition, thereby widening the gap with unlisted players. As industry insiders put it: “In the past, the large-model industry was often driven by marketing; but as leading companies land in the secondary market, the window of opportunity left for primary-market players is narrowing.”
This differentiation trend is also pushing other companies to rethink their positioning. Moonshot AI (Yuezhianmian) chooses to continue raising funds in the primary market, with cash reserves on its books exceeding RMB 10.0 billion (approx. USD 1.43 billion); 01.AI (Lingyi Wuwu) has fully focused on B2B business, seeking survival space through differentiated competition. It can be foreseen that China’s large-model industry will form a multi-layered landscape of “listed leaders + unicorns + vertical specialists” in the future.
Implications and Outlook: The Survival Rules in the Deep-Water Zone of Commercialization
The listings of Zhipu and MiniMax provide valuable commercialization samples for China’s large-model companies. The two companies’ different paths reveal one core fact: at the critical juncture where the AI industry shifts from a “technology race” to an “efficiency race,” whoever can be the first to run through a closed commercial loop will gain recognition in the capital market.
For Zhipu, the success of its MaaS model verifies the enormous potential of the B2B market. The company’s MaaS ARR (annualized recurring revenue) increased from RMB 20.0 million (approx. USD 2.87 million) to more than RMB 500 million (approx. USD 71.74 million), a 25x surge in 10 months, proving the advantages of standardized products in lowering marginal costs and enhancing economies of scale. But how to balance the stable income from localized deployment with the rapid growth of cloud business remains a problem it must solve.
MiniMax’s C-end globalization route demonstrates another possibility. Through a “China technology + global market” model, the company successfully avoided fierce domestic competition and found user groups overseas that are willing to pay for models. In the first three quarters of 2025, the number of paying users of its AI-native products had reached 1.7716 million, proving the feasibility of a productized route.
However, listing is not the end point, but the beginning of an even more severe test. Both companies face the challenge of how to convert technological advantages into a sustainable profitable model. Under capital’s scrutiny, they need to find a balance between R&D and profit pressure, seek breakthroughs between global competition and localized implementation, and build bridges between technological innovation and commercial application.
Looking back from the starting point of 2026, the successive listings of Zhipu and MiniMax are like pressing an accelerator for China’s large-model industry. This capital feast is not only a reward for the past four years of technological accumulation, but also a test of future commercialization capability. When the spotlight gradually fades, the market will ultimately use cold financial statements and user data to measure the true value of a large-model enterprise. In this new era full of opportunities and challenges, only those companies that can find a perfect balance among technological depth, commercial breadth, and development speed can laugh to the end in this marathon with no finish line. This is not only a milestone in the development of China’s large-model industry, but also an important signal of the evolution of the global AI industry landscape. Their success proves China’s strength in the field of artificial intelligence and also points the way for those who come after. However, after the capital frenzy, the real test is only just beginning. In the future, whoever can build a solid moat in technological barriers, whoever can run through a profitable closed loop on the commercialization path, will be able to stand firm in this efficiency race. This protracted war about technology, capital, and business has only just opened its curtain.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.