Tencent lets go, AI has no leader

Over four years ago, Tencent significantly reduced its holdings in JD.com and Meituan. Now, Tencent is once again stepping back from its portfolio companies.

On the evening of July 6, Kuaishou announced in a public statement that Tencent would sell a 6.3% stake in Kuaishou to multiple buyers through an off-exchange transaction. After the deal is completed, Tencent's shareholding in Kuaishou will drop from 15.68% to 9.37%.

According to market sources, the price range for this transaction is HK$43.15 to HK$44.53 per share. Based on this calculation, Tencent can cash out up to HK$12.5 billion.

The day after the news broke, Kuaishou's Hong Kong stock price plummeted by more than 12%, closing at HK$40.46 with a market value of HK$175.1 billion. Compared with its 52-week high of HK$91.04, Kuaishou's stock price has fallen by more than half.

On the same day, Tencent's stock price rose by over 2%, closing at HK$461.2 with a market value of HK$4.19 trillion. However, compared with its 52-week high of HK$675.1, Tencent's stock price has also declined by more than 30%.

By reducing its stake in Kuaishou, Tencent has obtained the cash it urgently needs at the moment. What deserves more attention, however, is that Tencent is taking big steps back from its portfolio companies — it is not only giving up the position of Kuaishou's major shareholder, but also showing a stance of no longer being the largest external stakeholder of Kuaishou's spin-off AI business, Keling.

This is very "un-Tencent-like".

In the mobile internet era, Tencent was accustomed to placing heavy bets on star companies in key tracks, often becoming the largest shareholder with a stake of over 20% and securing a seat on the board of directors, giving it significant decision-making power.

But now, Tencent is not only no longer seeking controlling stakes in new investment projects, but is also massively exiting from old projects, even though it could indirectly profit from the AI "rising star" through those old holdings.

By loosening its grip on AI and not tightly controlling Keling, Tencent is once again conveying a major trend: In the AI era, tech giants no longer have the possibility to "monopolize" star startups.

Even a powerful company like Tencent, when building new alliances, can no longer pursue the "exclusive benefit" model of the mobile internet era. Instead, it must shift to an "open-door" approach, investing in AI companies together with others to share profits.

On the other hand, after the previous round of divestment from JD.com and Meituan, the number of companies in which Tencent still holds strategic stakes has dwindled, and now Kuaishou has been removed from that list. It is foreseeable that Kuaishou will not be the last portfolio company that Tencent "unties" itself from — the "Tencent ecosystem" born in the mobile internet era is rapidly becoming a thing of the past.

1

Tencent's this move to reduce its stake in Kuaishou is driven by both financial gains and, more importantly, strategic positioning.

The company is going all-in on AI, which requires massive capital investment across the board: training the Hunyuan large language model, developing, operating and promoting AI products such as WorkBuddy and Yuanbao, as well as the highly anticipated WeChat AI, all demand continuous and substantial financial support.

Tencent President Martin Lau previously stated that in 2025, the company will invest 18 billion yuan in new AI products, and the planned investment for 2026 will at least double that figure. In the first quarter of this year, Tencent's capital expenditure reached 31.9 billion yuan, a 16% increase over the previous year, almost equivalent to half of its operating profit for the quarter.

In addition, Tencent has restarted its investment engine, actively acquiring AI projects in the secondary market with a significantly higher transaction frequency. In the primary market, Tencent has been conducting large-scale share repurchases on a regular basis, buying back over HK$9.2 billion worth of shares just in June this year.

The more than 10 billion Hong Kong dollars obtained from selling Kuaishou shares has replenished Tencent's war chest of funds. However, if the goal were purely financial, Tencent would not have needed to reduce its stake right now.

For one thing, Kuaishou's stock performance has been mediocre recently, making this a suboptimal timing for cashing out. Worse still, the news of a major shareholder's stake reduction is bound to cause Kuaishou's stock price to plummet in the short term; since Tencent has not completely liquidated its position, it will suffer considerable book losses.

For another, although Tencent has many areas requiring capital expenditure, it does not lack this mere billion or so US dollars. As of the end of March, Tencent's cash reserves exceeded 530 billion yuan. The claim that "reducing the stake in Kuaishou is to increase the investment in Keling" does not hold up logically — that investment was small in scale, costing less than 1.4 billion yuan in total.

Tencent's own series of actions also indirectly indicates that this stake reduction is not a purely financial move.

The first point lies in the destination of the shares.

Four or five years ago, when Tencent reduced its holdings in JD.com and Meituan, it distributed the shares of the two companies to its own shareholders as a special dividend. In other words, the shares of those two companies remained within the broader "Tencent ecosystem", only transferring from Tencent the company to Tencent's shareholders.

In contrast, Tencent's reduction of its Kuaishou stake is a direct sale, with the shares flowing into the open market. This time, Tencent does not seem to intend to create another "internal cycle".

Secondly, the magnitude of the stake reduction is also significantly different.

Back then, Tencent almost "liquidated" its positions in JD.com and Meituan, with its shareholding plummeting from around 17% to about 2%. For Kuaishou, however, Tencent precisely reduced its stake to 9.37%, slightly lower than the 9.83% held by Kuaishou's co-founder Su Hua. It is no longer the largest shareholder, but still remains a major stakeholder.

Finally, there is the special timing of the stake reduction.

In early July, more than 30 companies invested in Keling, including all three BAT giants, with Tencent holding approximately a 1.12% stake. But Kuaishou is the controlling major shareholder of Keling, retaining a 68.33% stake after the spin-off; this means that Tencent, as a Kuaishou shareholder, indirectly holds over 10% of the equity interests in Keling.

In this financing round, Keling only offered about 16.7% of its total shares for sale, yet dozens of companies are splitting this share, with most investors obtaining less than a 1% stake.

Tencent, which effectively holds more than 10% of the equity after equity penetration, stands out conspicuously.

Thus, a few days later, Tencent attempted to voluntarily step down from the position of "Keling's largest external stakeholder" by reducing its stake in Kuaishou. After this operation, Tencent's equity interest in Keling dropped to 6.4%, which, although still higher than that of other investors, at least narrowed the gap.

These three unusual actions convey a clear message to the outside world: Tencent is not bearish on Kuaishou, but rather does not want to be Kuaishou's largest shareholder, nor does it want to be the largest beneficiary of Keling.

In other words, neither Kuaishou nor Keling is Tencent's "private property" that no one else can touch; others who want to invest in these companies will not encounter obstacles from a controlling shareholder.

2

Tencent's decision not to be Kuaishou's largest shareholder and its reluctance to become Keling's largest external beneficiary is consistent with its previous investment strategy for the "five domestic AI dragons" and DeepSeek.

Companies such as Zhipu AI, MiniMax, and DeepSeek are the most worthy investment targets in China's AI industry. Tencent is willing to place bets on them, but the size of its shareholdings remains very limited.

Tencent participated in Zhipu AI's Series B4 financing, investing 200 million yuan in exchange for approximately a 2% stake. It also invested 50 million US dollars (about 340 million yuan) in MiniMax around 2023, holding a 2.58% share.

As for Moonshot AI, Baichuan Intelligence, and StepFun, Tencent has not disclosed its shareholding ratios. But referring to the previous two companies, the stakes Tencent obtained are estimated to be not very high either.

Regarding DeepSeek, Tencent squeezed into its first round of financing in early June, injecting 10 billion yuan and becoming one of its largest external investors. Compared with its investments in companies like Zhipu AI, Tencent raised its investment scale by two orders of magnitude.

However, DeepSeek's valuation has reached around 340 billion yuan, and Tencent's shareholding ratio is only about 3%.

Such a "restrained" investment approach is markedly different from Tencent's practices in the mobile internet era.

Tencent has a large number of portfolio companies, represented by JD.com, Meituan, Kuaishou, and Pinduoduo. In these major companies, Tencent's shareholding ratio at its peak was around 20%. In slightly smaller companies like Bilibili and Xiaohongshu, Tencent also holds about 13% of the shares.

This figure is several times, or even more than ten times, higher than Tencent's shareholding ratios in AI companies.

Tencent spent huge sums of money to take large stakes in emerging leaders across various tracks, forming a massive "Tencent ecosystem". The integration of traffic and complementary business operations between Tencent and its portfolio companies was one of the key pillars of its huge success.

But in the AI era, Tencent has quietly changed its approach, with its shareholding ratio in many star projects dropping sharply to the low single-digit percentage range, no longer insisting on becoming a major shareholder with overwhelming decision-making power.

The reason is not difficult to understand: the external environment has undergone profound changes.

After 2021, the "interconnection and openness" of the internet industry has gradually become widely accepted, making it increasingly unworkable for tech giants to build their own "fiefdoms" through exclusive investments and partnerships. Tencent itself has actively dismantled the "Tencent ecosystem" by significantly reducing its holdings.

In fact, even if Tencent wanted to provide more capital, star AI companies might not be willing to accept it.

In the primary market, as long as AI companies achieve certain results, a large amount of capital will flock to them with checks in hand.

Zhipu AI completed 9 rounds of financing in the seven years before its planned listing, and MiniMax also raised funds 7 times. Keling's first financing round attracted more than 30 institutions of various types, including industrial capital like BAT, market-oriented PE firms such as CPE Yuanfeng and CITIC Goldstone, as well as industrial capital, local state-owned assets, and science and technology innovation funds.

This scenario would have been unimaginable in the mobile internet era.

At the same time, AI entrepreneurs no longer need to take capital from a certain giant to "pick a side" in advance.

Choosing between Giant A and Giant T was always a dilemma that plagued mobile internet entrepreneurs. No matter which choice they made, it meant a break or even confrontation with the other giant and its "affiliated companies".

But in the AI era, this problem has disappeared.

AI companies either directly sell tokens through APIs, engage in B2B business via localized deployment, or develop their own C-end products. They do not need the traffic pool support of tech giants, nor do they treat the giants' super apps as essential entry points.

For them, tech giants are just one type of potential investor, not entities that they must attach to or "side with".

On the other hand, AI companies will not act as the "pawns" of tech giants the way mobile internet companies did.

One of the defining features of the mobile internet era was the "proxy wars" manipulated by giants, with fierce battles breaking out in every niche track. For entrepreneurs, taking money from Giant A or Giant T meant being prepared to act as a "fighter" for the investor.

But in the AI era, the competition among AI startups is relatively healthy, focusing mainly on technology and product competition. After being invested in by BAT, Keling is often compared with ByteDance's Seedance, but the two are far from being at daggers drawn.

In this context, tech giants can buy a ticket to the AI arena by investing in AI companies; but they can no longer hold the expectation of "betting heavily on one company to harvest an entire track".

Of course, Tencent's "capital strength" remains its strong suit. Previously, it frequently made investments in tracks such as large language models, AI chips, underlying computing power, and AI applications, almost buying up half of the AI industry. A "new Tencent ecosystem" is poised to emerge.

However, this time Tencent not only did not seek exclusivity, but also did not insist on taking the largest equity stake to become the biggest shareholder. It even explicitly signaled to the outside world through directly reducing its stake in Kuaishou (and thus indirectly reducing its interest in Keling) that Keling is not under Tencent's "control".

Perhaps only in this way can Tencent's new alliance maintain lasting appeal and continuously attract more star AI companies. The "new Tencent ecosystem" is destined to be much looser than the old one, but it will also have greater inclusiveness, which can not only reduce resistance, but also create broader room for growth.

3

After this stake reduction in Kuaishou, among Tencent's remaining investment portfolio, the most important company is undoubtedly Pinduoduo.

Tencent once held approximately 18.5% of Pinduoduo's shares at its maximum, and this figure has now dropped to 13.8%. Tencent remains the second-largest shareholder after Pinduoduo's founder Huang Zheng, as well as the largest external institutional shareholder.

Estimated at current prices, this stake is worth about 130 billion yuan, which can bring generous book returns to Tencent. At the same time, Tencent can also maintain its crucial e-commerce "bridgehead" to continue competing with Alibaba and ByteDance.

However, reducing its stake in Pinduoduo may be a more worthwhile path for Tencent to take.

Tencent's top priority is to quickly establish a dominant position in the AI era. At the all-hands meeting earlier this year, Pony Ma declared, "Among Tencent's business lines, the only one that spends heavily and is still worthy of vigorous investment right now is AI."

This statement was addressed to internal staff, but it also carried a connotation of a public announcement — companies whose core business has little to do with AI have gradually moved out of the scope and focus of Tencent's investment. The fact that Tencent announced its stake reduction in Kuaishou shortly after Kuaishou spun off Keling is a direct reflection of this new investment philosophy.

Compared with Kuaishou, Pinduoduo has an even lower "AI content".

It has not made high-profile efforts to develop large models, nor has it participated in the competition for AI applications. Its management almost never talks about AI during earnings calls, forming a sharp contrast with other internet companies that "cannot open their mouths without mentioning AI".

Pinduoduo's application of AI is mainly focused on internal cost reduction and efficiency improvement. Last year, it publicly recruited AI-related talents, hoping to apply AI to scenarios such as e-commerce scenario implementation, product understanding, and attribute recognition.

Pinduoduo may have its own considerations and strategies for not rushing into AI development. But for Tencent, Pinduoduo, which has not fully embraced AI, does not fall into the category of projects "worthy of vigorous investment".

On the other hand, the trend of "interconnection and openness" has not subsided.

Tencent has been acting in compliance with this trend. Starting with JD.com and Meituan, it has gradually dismantled most of its "ecosystem"; with this stake reduction in Kuaishou, Tencent's "large-scale dismantling" is nearing completion.

In Pinduoduo, Tencent's