When we look back at the tech landscape in 2025, an intriguing paradox emerges:

While SaaS companies struggle to justify the need for an AI copilot to their clients, phone manufacturers continue to showcase trivial micro-innovations like 'one-click removal of passersby' at product launches. Meanwhile, certain so-called industry giants frequently hype up their AI strategies to inflate stock prices. In stark contrast, the gaming industry has quietly completed the commercial closed loop of AI.

For instance, while the outside world debates the timeline for AI implementation, Tencent's advertising system has already leveraged AI to elevate traffic monetization efficiency to unprecedented levels. As people continue to poke fun at chatbots, NetEase's NPCs (Non-Player Characters) have begun using emotional bonds to entice players into opening their wallets.

To put it bluntly, while other industries treat AI as a mere marketing gimmick or an 'internalized' tool for code writing, domestic game companies have taken the lead in transforming AI into a tangible growth engine, as evidenced by their profit statements.

This article aims to strip away the technological veneer and use a financial lens to uncover this silent revolution unfolding in China's gaming industry.

01 Tencent's 'Elephant Dance': AI Learns to Generate Revenue

Conventionally, Tencent's AI narrative may seem less glamorous compared to OpenAI's. However, financial data tells a different story: Tencent could well be the Chinese internet giant that has monetized AI the fastest and most aggressively in this wave.

The intriguing paradox lies in the fact that AI's greatest contribution to Tencent's gaming business does not stem directly from 'game development' but rather from the 'game selling' process.

Advertising has emerged as the primary driver of AI monetization.

According to financial reports, in the third quarter of 2025, Tencent's marketing service revenue soared to an astonishing 36.2 billion yuan, marking a 21% year-on-year increase. Against the backdrop of a moderate macroeconomic recovery and overall pressure on the advertising industry, this growth rate is nothing short of extraordinary.

The secret behind this success lies in Tencent's fully reconstructed 'Hunyuan + Advertising' full-link intelligent delivery system.

For game companies, acquiring traffic is an unavoidable challenge. Traditionally, optimizers had to manually monitor the market and adjust bids, targeting, and ad materials. Now, Tencent has infused its advertising system with the capabilities of the Hunyuan large model. AI can not only comprehend the content of advertising materials but also accurately predict which users will click, download, and make purchases after viewing.

The results are immediately evident.

Financial reports indicate that the number of operations required per 10,000 yuan of advertising spending by advertisers has plummeted by 80%, while eCPM (effective cost per thousand impressions) and CTR (click-through rate) have seen significant increases. In essence, without increasing ad inventory or disrupting the user experience, Tencent has managed to sell the same traffic at higher prices.

For Tencent's own games, this creates a perfect internal cycle: their games enjoy lower traffic acquisition costs and higher efficiency, while external games seeking to buy traffic from Tencent must pay a premium. It's a classic 'selling shovels' business, albeit one driven by AI-powered high-tech shovels.

Another data point that surprised Wall Street analysts is the efficiency improvement in capital expenditures (Capex).

In Silicon Valley, Microsoft and Google are locked in an endless arms race for AI dominance, with capital expenditures repeatedly reaching new heights. However, in China, Tencent has demonstrated a different approach to computing power. In Q3 2025, amid a limited supply of high-end computing power, Tencent avoided blind hardware accumulation, and its capital expenditure growth rate was significantly lower than its revenue growth rate.

Has Tencent halted its AI investments? Obviously not. The management's explanation is intriguing: optimization of existing computing power driven by chip supply constraints.

Due to the ongoing impact of export controls on chips like the H20, domestic giants cannot stockpile H100s as frantically as their overseas counterparts. This external 'force majeure' has compelled Tencent to forge a unique path of 'software for hardware.' Through techniques such as model distillation and quantization pruning, Tencent has significantly enhanced the inference efficiency of existing GPUs (including stockpiled A800s and domestic chips).

This reflects a compelling financial narrative: on the revenue side, AI drives high growth in advertising and gaming; on the cost side, software-hardware synergy leads to a substantial decrease in unit computing power costs. Squeezed from both ends, profit margins are naturally released.

In Q3 2025, Tencent's gross profit increased by 22% year-on-year, and its operating profit margin rose to 38%.

This is Tencent's survival philosophy in the AI era: avoiding blind computing power competitions and instead focusing on extreme engineering implementation and commercial monetization.

02 NetEase's 'Human-Machine Romance': Selling AI to Players

If Tencent utilizes AI for 'infrastructure,' NetEase employs it for 'products.'

Unlike some companies that merely pay lip service to AI strategies, NetEase's actions represent a deep integration of 'AI + gaming,' rather than a simple substitution of tools. In 2025, this approach finally bore fruit, with the mobile game 'Justice Online' serving as a prime example.

In today's declining MMO (massively multiplayer online) gaming market, 'Justice Online' has emerged as an outlier, amassing over 8 million pre-registrations even before its international server officially launches. One of its core selling points is its 'sophisticated' AI NPCs.

Traditionally, NPCs were nothing more than ruthless task-giving machines, sporting question marks above their heads and reciting decades-old lines. However, in 'Justice Online,' NPCs possess memories, personalities, and even emotions. Players no longer engage in dialogue solely for leveling up but rather for 'socializing.'

Financial report data reveals that intelligent NPCs have significantly boosted user engagement time and retention rates. For a free-to-play game, retention is its lifeline—retention equates to LTV (lifetime value).

In simple terms, NetEase has excelled in transforming AI technology into an 'emotional commodity.' Players are willing to log in just to chat a bit more with their favorite NPCs and even pay to buy clothes for them.

The moat built through this emotional connection is far more stable than numerical stacking. In Q3 2025, despite industry-wide growth slowdowns, NetEase's gaming and related value-added services net revenue still achieved an 11.8% increase.

Even more intriguing is NetEase's foray into the UGC (user-generated content) field. 'Dan Zai Party' owes its success as a national-level hit to AI-assisted creation tools.

Previously, creating a game map required knowledge of editors, logic, and even some coding. Now, through AIGC (Artificial Intelligence Generated Content) tools, elementary school students can generate complex map components using natural language descriptions.

This has yielded an astonishing economic effect: NetEase has successfully shifted the cost of content production to users while enhancing their enjoyment in the process.

As long as the underlying AI toolchain is maintained, a vast number of users will continuously generate free, high-quality content for the game. This is the underlying logic behind 'Dan Zai Party's' ability to dominate the charts for an extended period and compete with Tencent's 'Yuan Meng Xi Zhi,' as well as NetEase's new moat in the AI era.

03 Mid-to-Lower Tier 'Battle Royale': AI or Extinction

If the giants are using AI to 'add icing to the cake,' for mid-to-lower tier companies like Perfect World and Giant Network, AI is 'providing charcoal in snowy weather'—a lifesaver.

2024 was the darkest hour for Perfect World, marked by layoffs, losses, and project cancellations. By the first half of 2025, however, the plot had reversed—the company expected net profits to reach 480 million to 520 million yuan, successfully turning losses into gains.

Who rescued Perfect World? The answer lies in 'Zhu Xian World' and the cost reductions and efficiency improvements brought by AI.

As a PC game developed over many years, 'Zhu Xian World's successful launch in 2025 proved that traditional MMOs still have a market, but only if they boast sufficient quality and reasonable cost control. Perfect World explicitly mentioned in its financial report that 'internal adjustments and optimization measures have begun to show results.'

Translating this 'jargon' into plain language: through the application of AI toolchains, the costs of art outsourcing and inefficient labor have been significantly reduced. In the development of 'Zhu Xian World,' AI was widely used for scene modeling and NPC behavioral logic design, enabling this product with 3A-level graphics to avoid runaway development costs.

Giant Network took an even more 'geeky' approach.

Instead of competing in general-purpose large models, Shi Yuzhu's team developed a vertical-domain 'GiantGPT.' This specialized model understands games, numerical values, and player psychology.

In products like 'Space Action,' Giant Network utilized GiantGPT to achieve low-cost AI gameplay deployment. Compared to calling expensive GPT-4 APIs, the inference costs of their self-developed vertical model are extremely low, and response speeds are faster. Ultimately, Giant Network not only defended its core business in 2025 but also achieved high net profit growth.

37 Interactive Entertainment and Century Huatong, on the other hand, applied AI in their areas of expertise: traffic acquisition and reskinning.

The data disclosed by 37 Interactive Entertainment is staggering: with AI assistance, 2D art drawing efficiency increased by 60%-80%, producing over 280,000 AI original artworks per month. What does this mean? It means 37 Interactive Entertainment can test different advertising materials and game art styles at twice the speed of its competitors. In the SLG (Strategy Game) and card game sectors, where 'traffic acquisition reigns supreme,' this is akin to a dimensional strike.

DianDian Interactive, a subsidiary of Century Huatong, achieved a miracle with 'Whiteout Survival,' generating over 200 million USD in monthly revenue in 2025 through AI-driven data analysis. AI helped them accurately predict the LTV of users in different global regions, realizing a 'buy high, sell high' arbitrage model.

04 Shadows Behind the Prosperity: The Jevons Paradox and Compliance Challenges

Of course, while we revel in the impressive financial report data, we must not overlook the other side of the coin.

First is the manifestation of the 'Jevons Paradox' in the gaming industry.

Economics tells us that when technological progress improves resource utilization efficiency, the total consumption of that resource tends to increase.

In the gaming industry, AI has indeed driven the cost of a single original artwork to rock-bottom prices. But what happened? Companies did not save money to distribute bonuses but instead demanded 10 times, 100 times the content volume.

To compete for users' limited attention, game install packages have grown larger, skins are released faster, and events are more frequent. The industry has shifted from 'competing on headcount' to 'competing on computing power' and 'production capacity.' This intense competition has resulted in a situation where, despite lower unit costs, the overall competitive threshold has been raised infinitely. Small and medium-sized teams that still rely on hand-drawn art to compete with AI-driven production lines are akin to throwing eggs against rocks.

Second is the dramatic oscillation in talent structure.

The lightly described 'personnel structure optimization' in financial reports conceals the departure of countless junior artists, translators, and copywriters. In their place are highly paid AI engineers, technical artists (TAs), and data scientists.

Game companies are increasingly resembling tech companies. Previously, their core assets were 'creative talent,' but now they have become 'computing power + data + models.' This shift in asset attributes poses significant challenges to organizational management.

And then there is the ever-present Sword of Damocles—compliance.

On September 1, 2025, the 'Administrative Measures for the Identification of Artificial Intelligence-Generated Content' officially took effect, with far-reaching implications for the gaming industry.

The regulations require that all AI-generated content must include explicit or implicit identification. For giants like Tencent and NetEase, which have self-developed engines and middleware, this merely requires adding a line of code. However, for small studios relying on third-party black-box models, embedding compliant identification in millisecond-level dynamic rendering and ensuring AI NPCs do not utter non-compliant remarks pose significant technical hurdles.

The rising compliance costs are, in fact, accelerating industry consolidation, further concentrating resources toward the top players.

05 Epilogue: 2026, From 'Tools' to 'Species'

Looking back from the beginning of 2026, it is clear that AI's impact on China's gaming industry has undergone a transformation from 'quantitative change' to 'qualitative change.'

For Tencent, AI serves as operational leverage, reducing expense ratios while driving revenue growth; for NetEase, AI enhances experience value, making existing users willingly spend more; for mid-to-lower tier companies, AI is a survival ticket—those who obtain it live on, while those who do not are eliminated.

Looking ahead to 2026, the competitive landscape will once again evolve.

We may witness the emergence of true 'AI-native games'—not games with a bit of AI added but games entirely generated by AI in real-time. Agents will replace scripts as the new driving engines of the game world.

In the new, AI-empowered world, the sole constant might well be an age-old commercial tenet: technology merely serves as a tool, while products represent the ultimate arena where businesses vie for supremacy. This is a trajectory that has been validated by the gaming sector and stands as a classic case study that other industries should take the time to delve into and learn from.