Core Judgment

On July 22, Tencent Holdings experienced a sharp decline of 7.05% with heavy volume. This was a drop driven by the resonance of three factors:a narrow-scope data misinterpretation, renewed concerns over AI capital expenditures, and fund rotation.The third-party revenue data that triggered the decline contains systemic biases at both the coverage and accounting recognition levels.

Verification will need to wait until the Q2 earnings report on August 12. If the year-on-year growth rate of local game revenue falls within the +6% to +8% range, this decline will be confirmed as an overreaction; if it is significantly lower than +5%, we need to re-examine the aging speed of evergreen games.

Before the earnings report is released,the current inclination is to understand this decline as a classic case where high-frequency alternative data was amplified by market sentiment, rather than a fundamental inflection point.

Why did it crash this time?

Tencent closed yesterday at HKD 440.6 with trading volume nearing HKD 29.8 billion, approaching 10% of the total Hong Kong stock market turnover. This turnover share itself indicates that the selling pressure included a large number of programmatic and passive following orders, rather than a simple repricing of fundamentals. Tencent's weight dragged down the Hang Seng Tech Index by 3.04%%.

The trigger was a research report from a foreign investment bank citing SensorTower data, which pointed out that Tencent's domestic iOS mobile game revenue declined by approximately 2.6% year-on-year in Q2, with the top three games declining by 13% year-on-year and evergreen products declining by 7%; subsequently, financial media quoted and spread this information,"game revenue decline" was simplified to "game income decline," and panic emotions were quickly transmitted during the trading session..

Additionally, the Q2 reports of public funds showed that Tencent was reduced by about RMB 18.7 billion, exiting the top ten heavy-weight stocks of public funds. This means that before the data shock arrived, the allocation willingness of mainland institutions was already very low. The crowding out of mid-US internet positions by AI hardware directions has been a continuous phenomenon since Q2.

With the chip structure already loose, any negative news at this time would trigger concentrated selling.

Bear Logic Restored: What is the Market Afraid Of

Let's look at the bear logic first.

Paragraph 1: SensorTower shows that iOS revenue turned negative year-on-year, and the decline in top games was even deeper (-13%), indicating that the problem lies not in the long tail but in pillar products: the evergreen engines like "Honor of Kings" and "PUBG Mobile" may be starting to age.

Paragraph 2: Revenue is a leading indicator for income. A decline in revenue means that the income growth rate will step down in the next one to two quarters. If the engine, which accounts for the highest proportion of group profits, stalls, the anchor of the entire SOTP valuation will loosen.

Paragraph 3: At the same time, AI capital expenditure enters an acceleration phase. Market participants generally believe that although AI investment will not immediately impact current profits, it will gradually reflect in medium-to-long-term performance through depreciation and amortization, raising the future cost curve. With revenue decelerating and costs rising, profit margins are squeezed from both sides.

After Data Scope Correction, How Much Bearishness Remains

I did some research yesterday. The bear logic actually has issues with data scope. It is broken down layer by layer below.

The Bear Data Only Sees a Corner of Revenue

The scope tracked by SensorTower is Domestic + iOS + Mobile, which is a triple-narrowed window.

First, look at the Android gap.The proportion of domestic Android channel revenue has long been higher than iOS, and the distribution capability of Tencent games on the Android side (including proprietary channels) is significantly stronger than the industry average.Whether iOS revenue -2.6% can represent the direction of all-channel revenue is itself an unverified assumption, because the payment behavior of Android users and the rhythm of channel activities are not synchronized with iOS.

Next, look at the PC gap. Industry data shows that in the first 5 months of 2026, the domestic PC game industry revenue grew by 32% year-on-year, with single-month growth rates in April and May being even higher. Tencent is precisely the absolute leader in domestic PC online games. The strong cycle of "Valorant" and "Delta Force" on PC is completely outside the field of view of mobile tracking data. Using a scope that misses PC to judge a company with a continuously increasing PC share is biased.

Revenue ≠ Income, Deferred Mechanism is a Cushion

This is where I think the bear logic stands the least ground.

When game companies sell virtual items, income is not fully recognized at the moment of player payment, but is deferred and recognized according to the expected consumption cycle of the items.This means that reported income is a weighted smoothed value of revenue from the past several quarters.

Q1 this year coincided with the Spring Festival season, with strong revenue growth recorded (Q1 local game revenue was RMB 45.4 billion, +6% year-on-year, and due to the late Spring Festival, income growth lagged behind revenue growth, meaning the numbers on revenue were stronger than the report). The deferred balance formed by this excess revenue is being rolled out into income in Q2 and Q3.

So revenue looks at seasons, income looks at deferral.

Revenue dropping by about 20% from the Spring Festival peak quarter-on-quarter is a seasonal normalization that happens every year, and does not mean there is a demand problem; while reported income, because it still has Q1 deferred income recognition, the quarter-on-quarter decline will be much smaller than the revenue decline. Some sell-side forecasts for Q2 local game revenue expectations roughly fall around quarter-on-quarter -4% and year-on-year +8%.

Is the Worst Case Worth a 7% Drop

Let's do a stress test.

In Q2 2025, local game revenue decreased by 5.8% quarter-on-quarter from Q1. Assuming Q2 this year replicates the same quarter-on-quarter decline, i.e., assuming the deferred buffer is not thicker than last year and PC incremental gains do not exist at all, local game revenue would be approximately RMB 42.7 billion, +6% year-on-year; compared to the sell-side baseline scenario of RMB 43.6 billion, +8% year-on-year, the difference is about 2 percentage points, less than RMB 1 billion in revenue.

A quarterly revenue swing of RMB 1 billion corresponds to a single-day market value evaporation of about HKD 280 billion. Even if this RMB 1 billion in revenue is converted to profit at the high profit margin of the gaming business, and then given a 30x valuation, it still cannot explain a fraction of the drop.

There is a magnitude mismatch between the worst case and the stock price reaction, which is the main reason for currently defining the sharp stock price drop as an overreaction.

Risks That Truly Deserve Attention

Refuting data misinterpretation does not mean denying all concerns. As mentioned at the beginning, whether it can be confirmed as an overreaction requires verification from the secondary report. Here are the key verification indicators:

First, the delayed effect of AI capital expenditure depreciation is real.

The fact that current profits do not see an impact does not mean the impact does not exist. The key variable is whether AI investment can be realized as improved ad recommendation efficiency and cloud revenue increments in the same cycle, forming a capex→income closed loop. This is a long-term issue much more important than single-quarter game revenue.

Second, if the reading of -13% for the top three games continues in Q3, the nature will change.

Single-quarter volatility is noise, and consecutive two quarters in the same direction is a signal. The natural aging of evergreen games and the ability to take over with new products are the decisive variables for the valuation multiples of the gaming business.

Third, the retreat of public funds may not be a short-term behavior.

If the configuration migration of mainland institutions selling internet and buying AI hardware is structural, Tencent's valuation center will be repriced by foreign capital and Southbound capital, and the volatility center will rise systematically.

Conclusion

Tencent rebounded by almost 20% from this low point. Yesterday's drop directly narrowed the gain to single digits. The implied upside space for Tencent's target price from mainstream foreign banks is generally above 50%. Such a large pricing divergence means that the market's narrative framework for Tencent has not yet converged.

This wave of recovery in the Hong Kong stock market is a fund behavior driven by multiple resonances such as short covering and fund high-low switching, rather than changes in fundamentals. Funds remain in a wait-and-see state regarding increasing holdings in the Hong Kong stock market.

Before the overall market logic and individual stock narratives complete convergence, any wind or grass movement could trigger fluctuations like those seen yesterday.

$NVIDIA (NVDA.US)$Tesla (TSLA.US)$Meta(META.US)$Amazon (AMZN.US)$Gold ETF - SPDR(GLD.US)$Silver Trust ETF - iShares(SLV.US)$Meta(META.US)$Taiwan Semiconductor (TSM.US)$Grayscale Bitcoin Mini Trust ETF(BTC.US)$Circle(CRCL.US)$Coinbase(COIN.US)$Intel (INTC.US)$AMD(AMD.US)$Google-A(GOOGL.US)$Russell 2000 Index ETF - iShares(IWM.US)$Nasdaq 100 ETF - Invesco(QQQ.US)$Micron Technology (MU.US)$Broadcom (AVGO.US)$Eli Lilly (LLY.US)$Novo Nordisk (NVO.US)$IREN(IREN.US)$Applied Digital(APLD.US)