The institution spun off two Kuaishou entities

On the evening of August 19, Kuaishou held its Q2 earnings launch event.

I went through the financial report over and over: revenue rose by 1.4%, book profit dropped by 36%, the live streaming business continued to shrink, and the advertising growth rate missed the guidance it previously issued. There is nothing remarkable about these figures.

The real story behind this earnings report may have little to do with profits, because there is a key point: Kuaishou is splitting its business to sell off parts separately.

A month ago, Kuaishou announced that it would spin off its self-developed video generation large model Kling for independent financing. More than 30 institutions rushed to submit bids, valuing the business at 18 billion USD.

Tencent invested 1.363 billion USD, Alibaba Cloud invested 1.363 billion USD, and Baidu invested 341 million USD.

No one can remember the last time these three companies appeared on the same investment list after years of competition. The list also includes Beijing State-owned Assets Investment, and Huace Media that focuses on film and television production, which looks like everyone is chipping in for a share.

What is Kling?

It is a video generation large model independently developed by Kuaishou in June 2024, which supports text-to-video and image-to-video generation. It is the first in the industry to realize native 4K direct output.

It has gained huge popularity in the past six months, with over 100 million global users covering 224 countries and regions, nearly 50,000 enterprise clients, and it has won awards at the Cannes Lions International Festival of Creativity.

A few months ago, two creators made an AI short film *Paper Phone* with Kling, which was completed in three days and gained over 100 million views across all platforms. Its popularity is fully backed by real high-quality works.

However, here comes the contrast: it is not profitable at all.

It lost 500 million USD in 2024, and 1.9 billion USD in 2025. With a revenue of 1.1 billion USD and a loss of 1.9 billion USD, it loses 1.7 USD for every 1 USD it earns. By the end of last year, Kling had a negative net asset of 9 million USD, with more liabilities than assets.

Kuaishou made a real net profit of 20.6 billion USD in the whole year of 2025, while Kling lost 1.9 billion USD in the same year. For every 10 USD Kuaishou earns, almost 1 USD is burned by Kling.

This kind of business is a cash-eating monster for any listed company.

Raising a cash-eating monster costs a lot of money. Kuaishou's total capital expenditure for 2026 is about 26 billion USD, 11 billion USD more than last year, and almost all the incremental funds are invested in AI computing power.

The R&D expenditure in the second quarter reached 4.6 billion USD, up 34.7% year on year. More than half of the 36% profit drop in the income statement is caused by this huge spending.

The reason for the spin-off is that Kuaishou can no longer afford to support it. This is what Kuaishou's employees told the media: they have no choice but to spin it off.

Apart from the unaffordable cost, there is a more realistic reason: Kling is far less valuable if it stays within Kuaishou's system.

The market values Kuaishou as a "content company" at 1.5 times its annual revenue; while Kling, after being spun off, is priced as an "AI company" at 36 times its revenue.

I mentioned before that the same business, with two separate accounting systems, has a valuation gap of 24 times. If it is kept in the parent company, it can only be priced at a low, cabbage-level value.

Therefore, Kuaishou spun it out. For the 3 billion USD equity offering, nearly 2.8 billion USD has been signed, and Kuaishou retains 68.33% of the equity, remaining the major shareholder and consolidating it into its financial statements.

Note that even after the spin-off, the accounts are still connected.

The 1.9 billion USD loss of Kling in 2025 is still recorded in Kuaishou's income statement, and all its future losses will still be counted as Kuaishou's losses.

The funds raised go into Kling's account, and have nothing to do with Kuaishou's profits.

What Kuaishou gets is a 18 billion USD valuation figure and a 68.33% equity certificate; the market thinks it has got rid of a burden, but in fact it hasn't, it just gets another supporter to share the pressure.

Four days after the spin-off, another move was made on this business layout.

......

The one who made the move is Tencent.

On July 6, Kuaishou issued an announcement that Tencent sold 273 million Kuaishou shares through over-the-counter block trade, cashing out about 12.5 billion HKD; its shareholding ratio dropped from 15.68% to 9.37%, and it is no longer a major shareholder.

Tencent also left a sentence in the announcement: it is confident in the long-term development prospects of Kuaishou.

Saying you are confident after selling the shares sounds familiar, which is just like saying "we are still friends" when breaking up with someone.

The announcement did not disclose who took the shares, only eight words were written: several independent third parties.

Market sources said the transaction price was between 43 HKD and 44.5 HKD, a few percentage points discount to the closing price of the day; someone was willing to take all the 12.5 billion HKD shares at this price. Who is the buyer became the biggest gossip in the Hong Kong stock market in those days.

However, the interesting part of this incident comes later.

Four days ago, on July 2, Tencent's name just appeared on Kling's investment list. On one hand, it sold 12.5 billion HKD of old Kuaishou shares, on the other hand it paid 1.36 billion USD for new equity, selling old shares at 9 HKD per share and buying new shares at 1 HKD per share. This is a position adjustment.

Don't underestimate this 1 HKD per share move. Kuaishou spun off this high-growth business, and Tencent is one of the first batch of investors to grab the shares.

Where did the money from selling the shares go?

Some analysis says Tencent has been almost repurchasing its own shares every day recently, with nearly 10 billion HKD spent on repurchases in a month. The 12.5 billion HKD from selling Kuaishou's shares is enough for it to keep repurchasing for another month.

Converting the unwanted assets into cash, then converting the cash into its own shares. This move is very clear.

What's more surprising is that on the day of the reduction, July 6, Kuaishou's Hong Kong stock price rose by nearly 8% at close; the major shareholder sold 12.5 billion HKD worth of shares, but the stock price still went up. The market understood the implication:

Tencent is not bearish on Kuaishou, it only wants the most valuable part of Kuaishou.

The market learns extremely fast. Kuaishou itself started the spin-off, Tencent followed to split its position, and securities firms also began to evaluate the business separately. Previously, Kuaishou's valuation was calculated in one single table with a multiple based on profit.

Now it is different, Kling is counted as one part, the main station as another part, cash as the third part, and the sum of separate parts is the total valuation. This valuation method has a name called SOTP, which means sum of the parts valuation.

This summer, securities firms are queuing up to revise their valuation models.

CICC gave a target price of 72.8 HKD, Morgan Stanley 65 HKD, CLSA 57 HKD. The average target price of 34 institutions is 66.4 HKD.

At that time, Kuaishou's stock price was 37.8 HKD. The most pessimistic institution gave a target price of 39.4 HKD, only 1.6 HKD higher than the current price. The most optimistic one gave 93.99 HKD. For the same Kuaishou, the target price gap between the most optimistic and the most pessimistic institution is nearly 1.5 times.

The valuation calculated by sum of the parts method is nearly double that of the consolidated valuation.

Everyone has learned to calculate separately, but their calculation methods are different, and the two most professional institutions got two completely different valuation results for Kuaishou.

......

Let's lay out the accounts. At the close of August 19, Kuaishou's stock price was 37.8 HKD, with a market value of 163.5 billion HKD.

What does this figure mean? When Kuaishou was listed in 2021, its market value was at the trillion HKD level. In less than six years, it shrank to 160 billion HKD, less than a fraction of its peak value.

The available cash on the account is 121.3 billion HKD, which is the figure at the end of June. The sum of cash, time deposits and wealth management products, plus the equity of Kling calculated at the financing price, Kuaishou's share is about 12 billion USD. This is the valuation, not cash, and it is not listed and circulated yet. But it is clearly written on paper and recognized by investors.

Converted all to HKD, the sum of these two parts is 50-60 billion HKD more than Kuaishou's total market value.

For the remaining main businesses, including advertising, live streaming and e-commerce, the market gives a negative valuation. In the main businesses, e-commerce even stopped disclosing its GMV data, and it will no longer publish this indicator from 2026. A company that voluntarily hides its core operating indicators will only make the market think the worst of it.

The main business is not unprofitable. Kuaishou's operating cash flow in the second quarter reached 5.9 billion HKD, which is real money coming in. But the market does not trust other things: advertising revenue is declining fast, live streaming business is shrinking year by year, which is very disappointing.

I specifically checked the stock forum, some people said:

The stock price is trending towards bankruptcy. Some others murmured that the stock price has fallen so much that they thought the company was going to run away, and some others shouted while looking at the income statement: it is still making huge profits, exceeding expectations.

The words are crude, but the accounts make perfect sense.

For a company with more than 120 billion HKD cash on its account, whose market value is even less than the sum of cash and equity, there is only one explanation in the market: no one believes its remaining business is still valuable.

The market prices for future returns. The cash on account represents present value, Kling represents the near future, and the main business represents the distant future. The value of present and near future is recognized by investors, but no one believes in the value of the distant future, so the stock price cannot go up.

JPMorgan Chase did a calculation in June.

After deducting net cash and Kling's value, the transaction value of Kuaishou's remaining business is negative, but this business is expected to make a profit of about 3 billion USD in 2026.

A business that can make 3 billion USD in profit is marked as negative in its model. To put it in another way: a chicken that can lay eggs is calculated as having negative value.

Morningstar is even more extreme.

After its separate calculation, it said that almost all of Kuaishou's valuable assets are in the short video main business, and Kling is given almost no valuation. The 3 billion USD financing is only enough for it to burn for one year.

ByteDance and Google spend more than this whole round of financing on video generation business in a single quarter.

This is the original words from Morningstar. The implication is very direct: You only take 3 billion USD to fight this war, but your opponent's military expenditure in a single quarter is more than yours. Who is the opponent? It's ByteDance.

In 2026, its expenditure on AI infrastructure is expected to exceed 200 billion USD, and the financing will not change Kling's long-term development trend.

The same Kling gets two completely different valuations in the two models, one is extremely high and the other is extremely low. The huge gap lies in the "spin-off" and the "consolidated accounts".

To put it bluntly, you spun off Kling, but you still keep it in your consolidated financial statements, splitting the business but not sharing separate accounts, which almost treats investors as fools. But institutional investors are not fools.

......

When they invested, there is a special clause hidden in the investment agreement: Kling must be listed before October 30, 2031, otherwise Kling and Kuaishou will repurchase the shares at an annualized return of 8%.

In plain language:

If Kling is not listed within five years, investors have the right to sell their equity back to Kling and Kuaishou at the original cost plus 8% annualized interest. Kuaishou is the joint and several obligor.

This transaction is half an investment and half a loan. It bets on whether Kling can complete its IPO within five years.

With this clause, the institutions' money is not at risk. If the listing is successful, they can earn profits from the secondary market. If the listing fails, they can get a guaranteed 8% annualized return. They will not lose in either case.

There is another clause in the investment agreement:

Within 5 years, Kuaishou cannot develop any other video generation large model. It is agreed at the time of spin-off that Kuaishou cannot start a new similar business. Kuaishou blocked this path, which is equivalent to telling the market: Kling is my only AI video business, there is no plan B.

The funds are in place, and the clauses are clearly written. The next step depends on Kling itself, because after being split out, it has to face the market independently.

In the domestic market, Seedance accounts for more than 80% share. Calculated by the consumed computing power, its penetration rate in the short drama industry is close to 95%.

For the professional market that Kling is proud of, H3 was open-sourced at the end of July, with an official price of 0.8 RMB per second, and the lowest price in third-party distribution channels is 0.09 RMB per second, less than one eighth of Kling's official price.

H3 is also the first open-sourced first-tier video model in the industry. After open-sourcing, other developers can modify the model, and their products are even cheaper than yours.

At the end of July, Seedance 2.5 was also launched, which can generate 30-second long shots at one time, while Kling can only generate 15-second clips. Kling is faced with the specification competition from Seedance and the price competition from H3 at the same time.

The opponents are not fighting alone. Seedance is backed by Douyin and TikTok's more than 2 billion global monthly active users. It is never just an independent product, but a key puzzle in ByteDance's ecosystem and content flywheel.

The more troublesome problem is the loss of core talents.

Zhang Di, the key figure who built Kling from scratch, left in August 2025. Five months later, he led the team to develop Alibaba's Happy Horse. On August 6 this year, senior researcher Wang Xintao also left.

Who is Wang Xintao? He is the author of two open-source super-resolution projects Real-ESRGAN and GFPGAN, the important foundation of Kling's image quality performance.