The global capital race for China's AI large models has officially kicked off.

【AI Capital Race】

Within a mere 48 hours, the Hong Kong Stock Exchange witnessed the bell ring twice for China's AI large model firms.

On January 8th, Zhipu AI led the charge by listing on the Hong Kong Stock Exchange, earning the title of the "world's first publicly traded large model company." The following day, MiniMax followed suit, marking a dramatic "duel of the titans."

From subscription enthusiasm to post-listing performance, both firms ignited a surge of market excitement.

Prior to listing, both companies boasted IPO valuations around HK$50 billion, evenly matched in terms of market expectations.

In terms of fundraising, Zhipu AI made its debut at HK$116.2 per share, raising approximately HK$4.35 billion; MiniMax issued shares at the upper limit of HK$165, securing HK$5.54 billion.

This signals a consensus in the primary market regarding the value of domestic large model firms.

However, MiniMax and Zhipu's extraordinary subscription multiples of 1,837x and 1,159x, respectively, propelled this recognition to unprecedented heights.

Even within the tech sector, such intense subscription levels rank among the highest. Expectations for China's AI large model firms in the capital market have reached their zenith.

The real divergence becomes apparent in stock price trends.

On January 8th, Zhipu's stock briefly "broke issue" but closed with a modest 13% gain. Over the next two days, it surged 20.61% and 31.4%, with momentum building.

Meanwhile, MiniMax followed a vastly different trajectory, resembling a "rocket launch."

Debuting on January 9th, its stock doubled in a single day, soaring 109%; the next day, it paused briefly with a 15% "small gain."

By the market close on January 12th, Zhipu and MiniMax boasted total market capitalizations of HK$91.7 billion and HK$123.1 billion, respectively.

The HK$30 billion gap in market capitalization reflects the capital market's differentiation and evaluation of current business models.

Coincidentally, Zhipu and MiniMax have entirely distinct business models.

Zhipu AI focuses on the B2B market, emphasizing localized deployment for enterprise clients and adopting a "steady and deep" cultivation strategy.

MiniMax, on the other hand, targets the B2C market, reaching users directly through AI applications. Its products span 200 countries and regions, amassing over 200 million users. Remarkably, 73% of its revenue stems from overseas markets like Singapore and the U.S., earning it the moniker of "pioneer in global expansion."

The capital market is precisely inclined to bet on imagination.

MiniMax's B2C user base and global potential align perfectly with investors' excitement.

Thus, such performance comes as no surprise.

Most importantly, these two IPOs injected a substantial "booster shot" into China's large model industry.

The market has demonstrated with real money that investors are willing to pay for the technological, ecological, and commercial capabilities of large model firms.

Globally, the AI capital feast has just begun.

By late 2025, SoftBank confirmed the closing of an additional US$22.5 billion investment in OpenAI, setting the stage for OpenAI's potential IPO at a US$1 trillion (approximately RMB 6.9 trillion) valuation.

Recently, large model unicorn Anthropic planned to raise US$10 billion, pushing its valuation to US$350 billion (approximately RMB 2.4 trillion).

The domestic race is far from over either.

In early 2026, another major player, Kimi, officially announced the completion of its US$500 million Series C funding, with a post-money valuation reaching US$4.3 billion (approximately RMB 31 billion).

This means that even though the first movers have established valuation benchmarks, the competitive landscape remains fluid.

Will Kimi, entering later, disrupt market expectations?

【B2B vs. B2C: Who Will Dominate?】

The capital market's logic dictates that once the listing bell rings, evaluation dimensions transcend mere technological competition.

After all, technology forms the foundation, but commerce reigns supreme.

Investors prioritize whether a company can outline a clear and imaginative commercial vision.

Even amid the same AI wave, Zhipu, MiniMax, and Kimi have chosen divergent paths.

Consider Zhipu first.

As a "deep cultivator" of the B2B market, it relies on the MaaS (Model as a Service) business model, adopting a dual-engine approach of "localized deployment + cloud API services" to primarily serve enterprise clients' large model needs.

This focus has yielded tangible results.

In China, 12,000 enterprises across sectors like internet, public services, telecommunications, consumer electronics, retail, and media have chosen Zhipu.

More impressively, nine of China's top 10 internet companies are Zhipu users.

This "bind the leaders, radiate the industry" approach has solidified its "reliable" commercial capabilities.

In contrast, MiniMax is a B2C "trendsetter," with commercial potential hidden in its rich product matrix.

From the AI virtual social app "Talkie/Xingye" to the video generation tool "Hailuo AI" and MiniMax Voice, its practical products have amassed over 200 million global users.

Thus, in investors' eyes, these are two firms with vastly different financial "temperaments."

Zhipu's core revenue primarily comes from "localized deployment," i.e., B2B business.

In 2024, localized deployment contributed 84.5% of revenue, with annual revenue growing approximately 150% to RMB 310 million; in the first half of 2025, revenue reached RMB 190 million, up 325% year-on-year.

This growth is both rapid and stable.

Although MiniMax's absolute revenue is smaller than Zhipu's, its growth is explosive:

In 2024, revenue reached US$30.5 million (approximately RMB 220 million), soaring 782% year-on-year; in the first nine months of 2025, revenue hit US$53.4 million (approximately RMB 380 million), up 174.7% year-on-year.

This explosive growth mirrors its post-listing stock surge.

While Zhipu steadily advances in B2B and MiniMax charges ahead in B2C, Kimi has stepped out of the "B2B or B2C" binary opposition, attempting a "fusion + upgrade" differentiated path:

On one hand, it deepens technological development, racing in the model and Agent tracks; on the other, it balances B2B and B2C, seeking to pioneer a "third path" for large model commercialization.

One of its core commercialization strategies focuses on Agents.

In June 2025, Kimi launched its first Agent product—Deep Research.

This next-generation Agent model, trained via end-to-end autonomous reinforcement learning, is designed specifically for deep research tasks.

Agents can complete complex tasks independently without step-by-step human guidance, delivering a rigorously structured research report within half an hour.

Previously, AI assistants were merely talkative "advisors"; now, with Agent capabilities, they can directly execute tasks.

Kimi's PPT function also incorporates Agent abilities, nearly reaching quasi-professional standards.

Most crucially, Kimi Agent's core strength lies in its proprietary model, providing technological support for its B2B and B2C commercialization.

In B2C, Kimi adopts a "free Chat mode + paid Agent function upgrades" model: on one hand, the free chat mode attracts massive users; on the other, it offers professional users upgrade options for deep research, PPT, and other professional functions based on usage counts.

In B2B, Kimi launched a targeted API platform for enterprises and developers, offering higher-level data security, concurrency efficiency, exclusive technical support, and priority access to new capabilities, addressing enterprise clients' core pain points.

Silicon Valley investor Chamath Palihapitiya stated he has migrated substantial workloads to Kimi's K2 model.

Kimi founder and CEO Yang Zhilin revealed in an internal letter: Over the past year, Kimi's monthly ring growth in overseas and domestic paid users exceeded 170%, while overseas large model API revenue surged 4x from September to November.

However, it must be objectively acknowledged that all these achievements and data represent only phased market validation results.

Whether Zhipu, MiniMax, or Kimi, all are currently in an exploratory phase of continuous technological iteration and business model evolution. The AI large model industry remains susceptible to disruptive shocks from new technologies and models at any time.

Risks and opportunities continually intertwine—this is precisely the allure of the AI large model industry.

【The Infinite War】

In Kimi's recently completed Series C funding, existing investors such as Alibaba, Tencent, and Wang Huimin chose to increase their stakes.

This has pushed Kimi's current cash reserves beyond RMB 10 billion, with its valuation rising to US$4.3 billion.

However, the AI industry has always been a "high-investment, slow-return" sector. For companies still scaling commercially, even with substantial cash reserves, they must continue renting computing power from industry giants and purchasing traffic, burning money endlessly.

No one knows where this money-burning ends.

Thus, both confidence and capital are crucial.

In response, Yang Zhilin stated, "We plan to use listing as a means to accelerate AGI, timing our move proactively—the initiative remains in our hands."

This adds a layer of industry exploration and testing to Zhipu and MiniMax's listings, probing how the capital market understands and prices these rapidly evolving AI large model firms.

The capital market's pricing logic will, in turn, influence the future evolution of the AI industry, reshaping its landscape.

Currently, the industry consensus is: The Chat war has concluded; the Agent "battle" is inevitable.

However, in Yang Zhilin's view, future model competition will transcend "hard metrics" like parameters and computing power, ultimately hinging on taste, worldview, and values:

"What you inject into the model determines what kind of intelligence it will manifest."

This highlights AI's essence: Technology is the skeleton; values are the soul.

Against the backdrop of global tech rivalry, future competition will extend beyond technical rankings to encompass business models, ecosystem construction, and global resource integration.

For the large model industry, the listing of the "world's first publicly traded company" is not an endpoint but a new beginning.

An infinite war belonging to the intelligent era has commenced.

Disclaimer

The content regarding listed companies in this article reflects the author's personal analysis and judgment based on publicly disclosed information (including but not limited to interim announcements, periodic reports, and official interaction platforms) as required by legal obligations. The information or opinions herein do not constitute any investment or other business advice. Market Capitalization Observer assumes no responsibility for any actions taken based on this article.

——END——