From Biren Technology to MetaX Technologies and Insilico Medicine, and then to Iluvatar CoreX. Within 48 hours, Zhipu AI and MINIMAX successively made their debut on the Hong Kong Stock Exchange. A number of AI-related companies in China are going public one after another.

Globally, especially for AI large model companies, a capital extravaganza has just commenced.

By the end of 2025, SoftBank announced the confirmation of an additional $22.5 billion investment in OpenAI. This laid the foundation for OpenAI, which is valued at one trillion US dollars (approximately 6.9 trillion yuan), to work towards going public.

Recently, the large model unicorn Anthropic planned to raise $10 billion, also driving its valuation up to $350 billion (approximately 2.4 trillion yuan).

The domestic competition is far from over either.

In early 2026, another major player, Kimi, officially announced the completion of a $500 million Series C financing. Its post-investment valuation directly reached $4.3 billion (approximately 31 billion yuan).

This indicates that even though the pioneers in going public have completed their valuation anchoring, the competitive landscape is still far from being settled.

Will Kimi, which enters the market later, overturn market expectations?

The logic of the capital market is that once the listing bell rings, the dimensions of scrutiny will expand beyond just technological competition.

After all, technology is the foundation, but business success is paramount.

Investors are most concerned about whether a clear and imaginative business landscape can be outlined.

Even when facing the same wave of AI, Zhipu, MiniMax, and Kimi have all chosen different paths.

First, let's take a look at Zhipu.

As a "deep diver" in the B-end (business-to-business) market, it has established a dual-engine setup of "localized deployment + cloud API services" through the MaaS (Model as a Service) business model, primarily meeting the large model needs of enterprise clients.

The results of this focused approach are quite solid.

In China, 12,000 companies across various sectors, including the internet, public services, telecommunications, consumer electronics, retail, and media, have chosen Zhipu.

Even more remarkably, 9 out of the top 10 internet companies in China are Zhipu's users.

This approach of "partnering with top players and radiating across the industry" has established its "reliable" business capabilities.

On the other hand, MiniMax is a "trendsetter" in the C-end (consumer-to-consumer) market, with its business potential hidden in a rich product matrix.

From the AI virtual social application "Talkie/Xingye" to the video generation tool "Hailuo AI" and MiniMax Voice, these practical products have accumulated over 200 million users globally.

So, in the eyes of investors, these are two companies with completely different financial "profiles".

The majority of Zhipu's core revenue comes from "localized deployment," that is, B-end business.

In 2024, the localized deployment business contributed 84.5% of the revenue, with annual revenue growing by approximately 150% year-on-year to 310 million yuan. In the first half of 2025, revenue reached 190 million yuan, a year-on-year increase of 325%.

This growth is both rapid and steady.

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

In 2024, revenue reached $30.5 million (approximately 220 million yuan), a year-on-year surge of 782%. In the first nine months of 2025, revenue reached $53.4 million (approximately 380 million yuan), a year-on-year increase of 174.7%.

This explosive growth is remarkably similar to its soaring stock price after going public.

While Zhipu is making steady progress in the B-end market and MiniMax is charging ahead in the C-end market, Kimi has stepped out of the binary opposition of "either B or C" and is attempting to forge a differentiated path of "integration + upgrading," that is:

On one hand, it delves deep into technology, striving for excellence in the model and Agent tracks; on the other hand, it caters to both B-end and C-end markets, attempting to open up a "third path" for the commercialization of large models.

One of its core commercialization strategies is to focus on Agents.

In June 2025, Kimi launched its first Agent product - In-Depth Research.

This is a new-generation Agent model trained using end-to-end autonomous reinforcement learning technology and an Agent product specifically designed for in-depth research tasks.

An Agent can complete more complex tasks independently without step-by-step human guidance, delivering a well-structured research report within half an hour.

Previously, AI assistants were just talkative "consultants"; now, with the addition of Agents, they can directly execute tasks.

Kimi's PPT function also incorporates Agent capabilities, reaching almost a quasi-professional level.

Most importantly, the core advantage of Kimi Agent lies in its autonomous model. This provides technical support for its commercialization in both B-end and C-end markets.

In the C-end market, Kimi adopts a model of "free Chat mode + value-added pay-per-use for Agent functions": on one hand, the free chat mode attracts a large number of users; on the other hand, it offers professional users upgrade options for the number of uses of professional functions such as in-depth research and PPT.

In the B-end market, Kimi has launched a targeted API platform for enterprises and developers, providing not only higher-level data security guarantees and concurrency efficiency but also exclusive technical support and priority access to new capabilities, addressing the core pain points of enterprise clients.

Silicon Valley investor Chamath has openly stated that he has shifted a significant amount of work to the Kimi K2 model.

Kimi's founder and CEO, Yang Zhilin, mentioned in an internal letter that over the past year, the month-on-month growth rate of Kimi's overseas and domestic paid user numbers has exceeded 170%, and the overseas large model API revenue has achieved a fourfold increase compared to the period from September to November.

However, it is important to objectively view that all these achievements and data are merely interim results of market validation.

Whether it is Zhipu, MiniMax, or Kimi, they are all currently in an exploratory phase of continuous technological iteration and business model changes. The AI large model industry is constantly subject to disruptive impacts from new technologies and new models. Risks and opportunities are constantly intertwined, which is precisely the charm of the AI large model industry.

Against the backdrop of global technological competition, future competition will be a comprehensive contest of business models, ecosystem construction, and global resource integration.

For the large model industry, an infinite competition belonging to the intelligent era has already begun.

Disclaimer

The content related to listed companies in this article is based on the author's personal analysis and judgment made using information publicly disclosed by listed companies in accordance with their legal obligations (including but not limited to interim announcements, regular reports, and official interaction platforms). The information or opinions in the article do not constitute any investment or other business advice. Market Value Observer does not assume any responsibility for any actions taken as a result of adopting this article.

——END——