The second-quarter earnings season for China's top five A-share game companies has concluded, revealing the most dramatic divergence in user acquisition strategies the industry has ever witnessed. On the evening of August 27, Century Huatong delivered its interim report, completing a week-long wave of filings that began with Perfect World on August 19, followed by 37 Interactive Entertainment on August 25, and both Kainet Network and Giant Network on August 26.
What makes these five reports most intriguing isn't who earned more, but rather how sharply attitudes toward paid user acquisition have fractured. Some companies are doubling down aggressively, while others are quietly retreating. Some declare that paid acquisition is dead, yet spend more recklessly than ever. This isn't mere disagreement—it's a fundamental shift in strategic direction.
Five Distinct Ledgers
Let's start with the hard numbers to see who's retreating and who's going all-in. The top three in sales expense growth—Perfect World at +213%, Giant Network at +122%, and Kainet at +117%—have all more than doubled their spending. Century Huatong's 42.3% growth looks moderate by comparison, but its absolute figure of 9.341 billion yuan exceeds the combined spending of the other four companies. Meanwhile, 37 Interactive Entertainment, formerly the king of user acquisition, is the only one of the five to post negative sales expense growth.
How aggressive was the cutback? DataEye's ADX platform shows that material投放 for 37's existing game "Time Grocery Store" dropped from a daily average of 4,391 sets in Q4 2024 to just 552 sets in Q2 2026—a staggering 87% reduction. Interestingly, 37's profit hasn't collapsed despite this. Net profit attributable to shareholders actually rose 26.14%. The secret lies elsewhere: investment income of 1.016 billion yuan accounted for 45.22% of total profit, with the equity revaluation from Zhipu's listing propping up half the income statement. However, non-recurring net profit stood at 1.016 billion yuan, down 26.74% year-on-year—meaning the core gaming business actually shrank by a quarter.
The old order of the user acquisition arena is beginning to crumble.
The Rules Have Changed
Why have all these companies suddenly changed their approach? Because two new measures have been introduced to the playing field. The first is what's called the "traffic tax." Starting in the second half of 2025, the pre-tax deduction threshold for advertising expenses has been tightened—traffic acquisition costs, creative production, and cloud tool usage are now all classified as advertising expenses, subject to a 15% annual revenue cap, with any excess not deductible before tax. In plain terms: previously, buying traffic was merely a cost issue; now it's become a tax burden. For 37 Interactive, with its 51.7% sales expense ratio, every yuan spent beyond the 15% cap also incurs an additional 0.25 yuan in tax.
The second measure involves stricter regulation of creative content. From May 2026, game user acquisition faces heightened scrutiny: suggestive art styles in anime, xianxia, and horror genres, startling visuals, and misleading copy are all being strictly regulated. Minor-directed channel placements must now include age-appropriate ratings and labels. Live streaming and short video content, including influencer oral promotion, trial plays, rebates, and gacha inducements, all count as advertising. Even soft-funnel tactics disguised as "reviews" or "walkthroughs" have been explicitly flagged as prohibited.
User acquisition used to be a medicine. Now the price has gone up, and the label carries three new contraindications. Add to this the invisible hand of deferred taxation. As software enterprise tax incentives phase out, some game companies' effective tax rates have risen from 10% back to 15% or even 25%. 37 Interactive's income tax expense for the first half reached 484 million yuan, up 132% year-on-year. Professional investors on Xueqiu have analyzed that the combination of taxable investment income and deferred tax liabilities has eaten up over 200 million yuan. More subtly, there's the mismatch effect of deferred revenue: when a new game launches, acquisition costs are immediately expensed, but player spending sits in a "deferred revenue" pool to be recognized gradually. Costs are front-loaded while revenue lags behind—meaning the hotter a game is, the worse the current period's financial statement looks.
Perfect World is a living example: "Anomaly" launched globally at the end of April, and by August 18 it had surpassed 2 billion yuan in cumulative revenue. Yet the company still posted a 118 million yuan loss in its interim report, with sales expenses up 213%. Revenue queues up in the deferred pool while spending hits the books immediately. This isn't a Perfect World-specific problem—it's a physical law across the entire industry's financial reporting.
Five Divergent Paths
With the rules transformed, the five companies have each handed in dramatically different answer sheets. Century Huatong has taken user acquisition global. Its 9.341 billion yuan in sales expenses tops the industry in absolute terms, but it's not buying anxious domestic traffic—it's securing overseas certainty: Century Games ranks second among global mobile game publishers by revenue, and "Whiteout Survival" once again claimed the global top spot in mobile game revenue in June, remaining hot three years after launch. The quality of this spending shows in cash flow: net operating cash inflow of 4.497 billion yuan, a near-100% cash conversion ratio from net profit, and 13.038 billion yuan in cash on hand. Globalized user acquisition is currently the highest-ROI gaming business in the world. That said, even as performance tops the charts, major shareholders have announced plans to reduce holdings worth over 3.5 billion yuan—the arena of user acquisition has never been governed solely by gaming rules.
Giant Network proves that blockbuster hits market themselves. "Supernatural Action Group" has surpassed 12 million DAU, 200 million cumulative registered users, and 5 billion yuan in cumulative revenue. Revenue grew 182% with a net margin of 45.7%—the highest among all five companies, bar none. The 122% surge in sales expenses looks alarming, but the expense ratio is only 28.2%—underneath it's still the Brain Gold-style brand play, relying on product-driven organic traffic. Its only blank spot: overseas revenue of just 15.45 million yuan. Giant Network hasn't even entered the overseas battle—which is also its biggest upside potential.
Kainet Network is maxing out its leverage in the domestic market. Its 117.26% increase in sales expenses represents the boldest bet on domestic user acquisition among the five. The confidence comes from the legendary IP genre, a market worth roughly 40 billion yuan—"Legend of Orcs," "Angel's Battle," and "Blaze Awakening" contributed around 1 billion yuan in revenue in Q1 alone, with user platform business growing 44.17%. Net profit contains another story: the arbitration settlement over the Legend IP brought in approximately 189 million yuan in non-recurring gains. Excluding that, non-recurring growth is 24.31%—still solid, but mobile game gross margin fell 6.29 percentage points. Nostalgia is a good business, but it's getting increasingly expensive.
Perfect World is hunkering down. The 118 million yuan loss looks ugly, but operating cash flow is 647 million yuan, up 28.89% year-on-year—receivables haven't deteriorated; it's the income statement that's paying for the future. The 2 billion yuan in revenue from "Anomaly" sits in the deferred pool, set to release gradually from Q3. This is the painful period of front-loaded R&D: survive it and it's a turning point; fail and it's just a story.
37 Interactive Entertainment is stepping back to wait for the wind. It's the only one among the five pursuing strategic contraction. Domestic revenue fell 26.18% as it proactively cut spending, while R&D investment actually rose 15.88%. Two variables are quietly building: first, AI-generated creative—according to DataEye, AI materials now account for 5% of WeChat mini-game platform content, and some of 37's titles have AI creative output ratios as high as 41%; second, a pipeline of 22 new game titles awaiting a fresh product cycle. With acquisition cuts, AI-driven cost reduction, and investment income as a cushion, 37 Interactive is positioning itself through retreat, waiting for the next card to be dealt.
What Are Investors Waiting For?
This is the year of cheap game stocks: the sector's expected PE for 2026 sits at 14-18x, at a five-year low and more than 60% below its 2021 peak. The CSI Animation and Gaming Index trades at a PE of 24.62x, only at the 20th percentile of the past year. Individual stocks are even more plainly inexpensive: 37 Interactive at a TTM PE of 12.82x, Kainet at roughly 13.2x forward, and Century Huatong at just over 11x.
Why is the market pricing them so low? It's applying a "discount for outdated models"—capital markets no longer believe long-term winners can emerge from the old user acquisition playbook. After a decade of traffic dividend stories, investors have wised up. But low valuations are also options. Century Huatong generates nearly 4.5 billion yuan in annual net profit with 13 billion in cash; Giant Network pays 0.80 yuan per 10 shares, totaling nearly 1.6 billion in dividends; 37 Interactive combined 300 million in buybacks and cancellations with 682 million in dividends. The cash on hand is more substantive than any narrative.
The market has essentially already voted between paths and valuations: overseas certainty (Century Huatong) outweighs blockbuster elasticity (Giant Network), which trumps nostalgia leverage (Kainet), ahead of transformation pain (Perfect World), with model transition (37 Interactive) last. One arena, five paths, five different PEs.
A New Arena
Is user acquisition dead? No—it has simply relocated. It's moved from domestic to overseas—Century Huatong spent 9.3 billion to secure a global publisher runner-up position. It's shifted from large-scale to precision—the competitive dimension in creative content has switched from quantity to efficiency. And it's migrating from human-produced materials to AI-generated ones—production costs are collapsing, even as regulators keep watch over AI content.
Industry data supports this migration: in the first half of 2026, daily direct spending on mini-game acquisition reached approximately 179 million yuan, up 26% year-on-year. The mini-game market is projected to reach 74.5 billion yuan for the full year, up 22.13%. The pie is still growing—only the playbook has changed.
Looking ahead, three certainties emerge. Overseas expansion is the only consensus—even Giant Network, the slowest to move abroad, has written overseas growth into its plans. Competition for domestic traffic has devolved into zero-sum cannibalization. AI is rewriting the cost curve: creative production shifts from labor-intensive to AI-intensive, pushing marginal acquisition costs lower—though compliance standards for materials will only grow stricter. The endgame of user acquisition is retention: the traffic tax and ad restrictions have raised customer acquisition costs, forcing everyone back to product quality. Whoever can convert "acquired users" into "retained users" secures a ticket to the next cycle.
At its core, user acquisition was never the original sin—ineffective user acquisition was. Five interim reports, five ways of operating. The arena hasn't fundamentally changed; it's just that the competition has shifted from who spends the most to who retains the most. What hasn't changed is this: those who make games and those who trade stocks have never lived on the same timeline.
---
Data sources: Company interim reports for 2026, DataEye Research Institute, Sensor Tower, Gamma Data; PE ratios as of each report's disclosure date. This article is an industry observation and does not constitute investment advice.
Comments