Is it certain that one of the valuations, 20 billion for KeLing and 9 billion for the main Kuaishou platform, is wrong?

According to LatePost, Kuaishou is planning to spin off Keling for a new round of financing, and the market has given it a valuation as high as $20 billion.

Looking at this figure alone, it doesn't seem particularly exaggerated in the current AI financing frenzy. However, when you put it back into Kuaishou's entire ecosystem, you'll immediately feel a strong sense of dissonance.

What does $20 billion mean?

As of today's close of the Hong Kong stock market, the entire market value of Kuaishou is still less than $30 billion.

In other words, excluding Keling's $20 billion valuation, the implied valuation left by the market for Kuaishou's main - site business may be only about $9 billion.

It should be noted that this is a short - video platform with 700 million monthly active users, quarterly revenue close to 40 billion yuan, and adjusted annual net profit reaching 20.6 billion yuan.

On one hand, there is a mature Internet platform that has achieved stable profitability and has a large user base, but its valuation is pushed down to less than $10 billion.

On the other hand, there is an AI business that is still in a high - investment period, but the primary market has given it a separate valuation of $20 billion.

Behind this strong contrast actually means one thing: Chinese AI assets are undergoing a systematic revaluation.

And the fuse for this round of revaluation is the sky - rocketing prices of Zhipu and MiniMax after their listings.

The rise in the secondary market not only allows Zhipu and others to enjoy higher valuations but also directly raises the pricing anchor for AI large - model companies in the entire primary market.

Thus, an extremely delicate situation has emerged:

Zhipu and others have gained the "face" and are enjoying exaggerated valuations in the secondary market; while companies like Yuezhianmian and Jieyuexingchen have gained the "substance" and raised a considerable amount of money in the primary market. Everyone is a big winner.

More importantly, with this rapid increase in valuation, the long - standing valuation discount between Chinese and American AI model assets is also rapidly narrowing.

Today, let's talk about this large - scale revaluation of Chinese AI assets, which was ignited in the secondary market and quickly transmitted to the primary market.

The listing of Zhipu has triggered a "bull market" for domestic AI assets

The starting point of this revaluation is the sky - rocketing prices of Zhipu and MiniMax after their listings.

Calculated at the issue price, MiniMax's post - listing increase reached 330%; Zhipu's increase was even higher, reaching 657%.

More important than the stock price is that they have redefined the market's valuation method for AI companies.

Based on an ARR of about $150 million in February 2026, the market's valuation multiple for MiniMax is about 200 times ARR.

Zhipu is even more exaggerated.

As of March 31, 2026, the ARR of the company's open - platform API has reached about 1.7 billion yuan, approximately $250 million; and the current valuation has reached $58.8 billion, corresponding to a valuation/ARR multiple of about 235 times.

This also provides the most direct valuation anchor for AI large - model companies in the primary market, and this pricing logic quickly reverberated back to the primary market.

The most typical example is Yuezhianmian.

At the end of 2025, Yuezhianmian's valuation was only about $4.3 billion. But after entering 2026, the company's financing clearly accelerated: it completed three rounds of financing in January and February, with amounts of $500 million, $700 million, and $700 million respectively; in May, it completed a new round of financing of about $2 billion, and its valuation directly exceeded $20 billion.

In terms of commercialization, its ARR in April 2026 has exceeded $200 million. Correspondingly, the valuation/ARR multiple is about 100 times.

Jieyuexingchen is also entering a similar stage.

In January 2026, the company completed a Series B+ financing of over 5 billion yuan, setting a new record for a single - round financing in China's large - model track in the past year; in May, it was close to completing a new round of financing of nearly $2.5 billion, and the market's corresponding IPO valuation is about $10 billion.

This change in the valuation logic also applies to other video - model companies.

LatePost exclusively learned that Keling's current ARR (Annual Recurring Revenue) has reached $500 million. Calculated at a $20 - billion valuation, it corresponds to about 40 times ARR.

Looking at it alone, 40 times is not cheap.

But if placed in today's valuation framework of China's AI primary market, Keling doesn't even seem expensive.

The post - investment valuation of Vidu/Shengshu Technology exceeds $2 billion, while the previously publicly disclosed ARR is still only about $20 million, corresponding to a valuation/ARR multiple of about 100 times.

In contrast, Keling's valuation of $20 billion based on an ARR of $500 million even starts to look quite "attractive".

The valuation discount of Chinese AI assets is being eliminated

The most interesting thing about this round of AI asset revaluation is:

Zhipu and others have gained the "face", while Kimi, Jieyue and others have gained the "substance".

The former enjoy high valuations in the secondary market, while the latter have raised real large - scale financing in the primary market.

Take Zhipu as an example. Although its stock price has risen astonishingly after listing, the actual fundraising scale is not as exaggerated as expected. Its IPO fundraising amount is about HK$4.3 billion, equivalent to about 3.7 billion yuan.

However, leading model companies in the primary market have raised far more funds than through an IPO while their valuations are rising.

For example, Yuezhianmian has completed 4 rounds of financing this year alone, with a cumulative financing amount of $3.9 billion, equivalent to about 26.5 billion yuan; Jieyuexingchen has also raised more than 22 billion yuan in two rounds of financing.

To some extent, the sky - rocketing prices in the secondary market have instead become a financing "amplifier" for model companies in the primary market.

Another notable phenomenon is that with the rapid increase in valuation, the long - standing valuation discount between Chinese and American AI model assets is being quickly eliminated.

Take Anthropic for example.

Based on the latest round of financing, Anthropic's valuation is approaching $1 trillion, and its ARR has reached about $45 billion, corresponding to a valuation/ARR multiple of about 22 times.

After OpenAI's latest round of financing, its post - investment valuation is about $852 billion, and its latest annualized revenue is about $25 billion, corresponding to a valuation/ARR multiple of about 34 times.

In the field of AI video, Runway's latest valuation is about $5.3 billion, and its ARR in mid - 2025 was about $90 million, corresponding to a valuation/ARR multiple of about 58 times.

Just looking at the price - to - sales ratio, this valuation is far lower than that of domestic large models. Of course, there is an important premise here:

The commercialization of domestic models is still in a very early stage, with a lower revenue base, and high - growth expectations will naturally amplify the valuation multiple. Therefore, the valuation/ARR itself cannot be simply compared horizontally.

But one thing is certain: Over the past few years, the huge valuation gap between Chinese and American AI assets is rapidly narrowing.

Currently, the combined valuation of Anthropic and OpenAI, two model companies, is about $2 trillion; while the combined valuation of core model assets such as Zhipu, MiniMax, Yuezhianmian, Jieyuexingchen, and DeepSeek in China is about $158.9 billion.

If we further consider the value of ByteDance's AI business, the situation will be even more interesting.

The market generally believes that ByteDance's overall valuation is about $550 billion. Referring to its profit of about $33 billion in 2024, if the mature business is estimated at a 10 - times PE ratio, then the implied valuation of ByteDance's AI segment has probably reached about $200 billion.

That is to say, the combined valuation of unlisted Chinese AI model assets has reached about $358.9 billion.

Correspondingly, the valuation gap between Chinese and American AI core assets is about 5.67 times.

And this figure is very close to the valuation gap between Chinese and American Internet core assets.

For example, in the e - commerce sector, the combined market value of Alibaba, Pinduoduo, and JD.com is about $510.9 billion; while Amazon alone has a market value of about $2.89 trillion, with a nearly 5 - fold gap between the two.

Interestingly, in this round of AI revaluation market, one sector has become an exception, that is, autonomous driving.

In terms of valuation, Waymo's latest valuation is $126 billion, while the combined market value of Pony.ai and WeRide is about $8 billion, which means a 16 - fold valuation gap between the two. In contrast, the market value gap between Uber and Didi is only 7.77 times.

If we look at the price - to - sales ratio, there are also obvious differences between Chinese and American autonomous - driving companies.

Waymo has not separately disclosed its revenue, but according to Sacra's estimate, Waymo's annualized revenue reached $355 million in February 2026, with a price - to - sales ratio of 354 times.

In 2025, the Robotaxi revenues of Pony.ai and WeRide were 116 million yuan and 150 million yuan respectively, with a combined total of 266 million yuan, equivalent to $39 million, and a price - to - sales ratio of 205 times.

This means that after the crazy revaluation of AI model assets, autonomous driving may instead become the "value depression" in the next stage of AI assets.

Summary

On the surface, this round of AI asset revaluation is a valuation game, but in essence, it is actually a process of re - pricing China's AI competitiveness.

Undoubtedly, Chinese AI companies are gradually getting rid of their past position as "followers". Even in some sub - fields such as AI video, Chinese companies have begun to achieve partial overtaking.

More importantly, different from the Internet era back then, this round of AI companies naturally have stronger global opportunities.

Whether it is models, video generation, or Agent products, many capabilities are designed for the global market from the very beginning.

Of course, there will definitely be emotions, bubbles, chip games, and temporary over - valuations.

But behind these constantly refreshing valuations, financings, and stock prices, what is more worthy of attention is actually another thing: A group of China's top - notch technology companies are, for the first time, truly approaching the core stage of the global technology industry with the help of this AI wave.

And from a longer - term perspective, this may be the most important part of this AI asset revaluation.

This article is from the WeChat official account "Silicon - based Observation Pro", author: A Qi, published by 36Kr with permission.