This is the commentary column of Pencil News. I am Wang Fang, the founder of Pencil News. Born after 1989, a Virgo, and an INTJ, I have been committed to delivering truthful content for a decade, with the ambition to build a media outlet that stands the test of time.

Kimi recently made headlines with a high-profile financing announcement: it aims to raise $1 billion at a valuation of 120 billion (roughly 70 billion yuan).

Why do we describe it as "high-profile"? Typically, financing news is disclosed only after the funds are secured. However, Kimi released the announcement before fully securing the investment.

While I am optimistic about Kimi’s future, this move raises questions: Is this round of financing not progressing as smoothly as expected?

- 01 -

What Message Does It Convey?

My concerns stem from two key factors.

First, prematurely revealing financing needs. The more a company insists it is not short of funds, the more likely it is seeking investment.

Few institutions can afford a valuation of 120 billion—likely fewer than 200 globally.

Such high-stakes financing should ideally be conducted through small-scale, targeted, one-on-one negotiations.

Yet, the current approach resembles: public speculation and widespread market discussion.

What does this resemble? Rather than signaling financial strength, it hints at urgency: "We still need more funds."

Second, the peak financing window for large AI models has closed.

This point is even more critical.

Several leading AI model companies have already gone public, such as Zhipu and Minimax, with market capitalizations ranging from 250 billion to 300 billion yuan.

Early investors are now eyeing exits, while those who missed out are shifting focus to other sectors, such as embodied AI.

These emerging fields promise faster IPOs and better capital returns, enabling quicker recouping of investments.

So, why would investors pursue large AI models at this stage? While there are reasons, the criteria have become stricter.

Condition 1: Valuation must offer compelling value.

The logic is straightforward. If competitors have already gone public and you are still in the funding stage, can your valuation be higher? Otherwise, the math doesn’t add up.

For instance, Zhipu and MiniMax were valued at 25-30 billion yuan pre-IPO and reached market caps of 250-300 billion yuan post-listing (a 10x return). This is a proven trajectory.

What about Kimi? With a current valuation of 120 billion yuan, can it reach 1.2 trillion yuan post-IPO? That will be challenging.

Thus, the "value" path is not feasible for Kimi.

Condition 2: A new narrative must justify the valuation.

Kimi has indeed introduced a new story: AI agents.

This narrative has fueled its valuation surge. In February, it secured $700 million in financing at a valuation of about 70 billion yuan. By March, its valuation exceeded 120 billion yuan—a 71% increase in two months.

Where does this new story originate?

Kimi has developed a product akin to OpenClaw: Kimi Claw. Since late January, the company claims its cumulative revenue over 20 days has surpassed last year’s total.

How is this revenue generated? Broadly, from three sources:

1. Selling computing power: Users run tasks, consume resources, and pay accordingly.

2. Interface fees: Users call APIs for tasks and incur charges.

3. Hosting fees: Users host their AI agents on Kimi’s platform and pay for the service.

- 02 - Is the New Narrative Sustainable?

Is this new story easy for Kimi to sustain? It’s uncertain. From a profitability standpoint, I believe that for large AI models, the more realistic monetization path may not be AI agents but advertising.

Many dismiss advertising as a "low-class" revenue model. Yet, the reality is: China’s most profitable tech companies rely heavily on advertising.

Take ByteDance as an example: its 2024 revenue exceeded $150 billion, with profits surpassing 240 billion yuan—driven primarily by advertising.

Why is advertising a faster monetization route? The logic is simple: a superior user experience attracts traffic, which in turn attracts advertisers, generating revenue.

Why is monetization through AI agents slower? Because current products prioritize form over function. Once users’ initial enthusiasm fades, they revert to practical metrics: ROI.

Kimi Claw’s short-term revenue surge does not guarantee long-term viability.

Thus, I believe large AI model companies will also explore advertising as a revenue stream.

Step 1: Deliver an exceptional user experience.

Step 2: Secure a top-2 market position.

Step 3: Monetize through advertising.

The potential for this path is substantial: even if annual revenue cannot reach 500 billion yuan, surpassing 100 billion yuan is highly plausible. Take Baidu as an example: at its peak, its annual advertising revenue exceeded 80 billion yuan.

However, achieving this requires a critical milestone: securing a top-1 or top-2 market share.

The question is: Has Kimi reached this milestone? No. By mid-2025, Kimi faced growth challenges, dropping to 9th place domestically (QuestMobile). By early 2026, it had rebounded slightly, hovering around 3rd to 5th place.

What does this imply? Kimi has not yet secured its "ticket" to advertising monetization. This applies not just to Kimi but to all domestic large AI model companies.

- 03 - Conclusion

Compared to advertising, AI agents represent a longer, more arduous path: requiring more financing, higher cash burn, and the cultivation of mature payment habits. Is it an opportunity? Absolutely. But it may not necessarily be Kimi’s opportunity.

I even posit: the leader in AI agents and the leader in large AI model companies may not be the same entity.

Why? History provides answers: the leader in operating systems is not necessarily the leader in applications; the leader in cloud providers is not necessarily the leader in SaaS.

Large AI models and AI agents are fundamentally different: the former focuses on building engines, while the latter focuses on driving and transporting passengers. One tests foundational technological capabilities, while the other tests application capabilities—two entirely distinct systems.

Kimi’s challenge is not the absence of a narrative but that the narrative is outpacing reality.

This article does not constitute investment advice.