AI won’t copy the script of blockchain games because it’s genuinely reducing costs and improving efficiency, truly transforming gameplay. But the backlash it brings is equally real: trust crises, workforce disruption, and regulatory scrutiny.
Article author and source: 0x9999in1, ME News
- Web3 gaming is not merely “in a temporary slump”—it is structurally dead. According to the Caladan report, 93% of Web3 games are “effectively dead,” having burned through $15 billion, with token prices down 95% from their 2022 peaks.
- Capital isn’t foolish—the money has already moved. In 2022, gaming accounted for 62.5% of Web3 venture capital; by 2025, it has dropped to single digits, with funds flowing en masse into AI and RWA.
- The question isn’t whether to use AI, but to what extent it will be integrated. As of December 2025, over 10,000 games on Steam were labeled as using AI, accounting for one-fifth of all new games.
- Major companies show clear divergent attitudes: Tencent, NetEase, EA, Ubisoft, and Take-Two are actively advancing; Nintendo, CDPR, and Larian are cautious or even resistant; smaller studios are using AI to ignite a new cost war.
- Players and industry professionals have not welcomed AI gently. A GDC survey found that opposition to AI rose from 18% to 30% over two years, and a one-year strike by SAG-AFTRA directly led to the creation of the 2025 Video Game Agreement.
- AI won’t copy the script of blockchain games because it’s genuinely reducing costs and improving efficiency, truly transforming gameplay. But the backlash it brings is equally real: trust crises, workforce disruption, and regulatory scrutiny.
Blockchain games have truly died out—completely and thoroughly.
It’s not a “market correction,” not a “cycle downturn”—it’s a structural collapse. In a report released in April 2026, market-making and trading firm Caladan laid out the numbers clearly: after burning approximately $15 billion, about 93% of projects in the Web3 gaming sector have become “effectively dead.” Token values have plunged 95% from their 2022 highs. Venture capital funding flowing to game studios was halved, then halved again by 2025, resulting in a staggering 93% decline.
What are the specific names? There are countless of them. Hamster Kombat lost 96% of its users within six months of launch. Yield Guild Games’ token has dropped 99.6% from its all-time high in November 2021. Axie Infinity’s daily active users have fallen from a peak of 2.7 million to around 5,500. Pixelmon, which raised $70 million through NFTs, has yet to release a public game after four years. Ember Sword burned through $18 million over seven years and shut down outright in May 2025, refusing even to issue refunds. Gala Games is embroiled in a lawsuit involving its co-founder’s alleged embezzlement of $130 million in tokens. Even Japanese giant Square Enix quietly shut down its Web3 experiment, Symbiogenesis, in July 2024.
Why die? Because the pattern itself is the opposite.
Play-to-Earn blockchain games are essentially financial cycles: buy NFTs or tokens, earn the same assets, then cash out. The only condition under which this equation works is a constant influx of new players bringing in money. Once new funding slows down, token prices collapse, rewards shrink, players leave, and the entire in-game economy collapses.
Coda Labs has even more sobering data: at the height of the frenzy, only 12% of traditional gamers ever tried blockchain games. In plain terms—after all this effort, real players simply haven’t shown up. The ecosystem is filled with grinders, pumpers, and storytellers, but conspicuously missing is the person willing to pay for fun.
Capital is the most honest. In 2022, the gaming sector accounted for 62.5% of total Web3 venture investments; by 2025, it had dropped to single digits. Even Animoca Brands, the most enthusiastic player in this space, has reduced gaming’s share in its portfolio to around 25%, shifting its full focus to stablecoins, RWA, and AI.
The blockchain game told a story of “tokens driving the future,” but ended up as a joke. This wasn’t a conspiracy—it was financial engineering outpacing product-market fit. Developing a game takes three to five years, yet tokens fluctuate 24/7 on secondary markets—the timing was wrong from the start.
Blockchain games are stepping back; AI is taking over. But AI is not following the same path.
Steam’s data is the most straightforward. According to statistics from Totally Human Media, nearly 8,000 games on Steam labeled as using AI were available in July 2025, a nearly 700% year-over-year increase. By December 2025, this number rose to 10,258 games. The data firm further shows that approximately 8% of Steam games use generative AI, and these games alone generated about $660 million in revenue.
One in every five newly released games on Steam uses generative AI. This is now the norm, not an isolated occurrence.
Domestic adoption is more aggressive. According to Gamma Data’s 2024 report, over 99% of surveyed companies or departments have introduced AI tools through various channels. The Game Industry Association’s 2025 report, “Research on AI Technology Applications in Game Companies,” presents a staggering figure: AI adoption in game development reaches 86.36%, while its penetration in art production exceeds 84%. Even in areas most demanding of “creative coherence,” such as worldbuilding and storytelling, the penetration rate remains at 31.6%.
So asking “Will AI enter gaming?” in 2026 is already a bit outdated—it’s already in, and deeper than many people realize.
But the deeper the penetration, the greater the division. The gaming industry has never been so clearly split on a single technology as it is today.
Domestic tech giants: From strategic bets to “AI money printers”
Tencent has taken the most pragmatic approach. It hasn’t just stuffed AI into slogans—it has integrated it into its advertising system, in-game NPCs, and R&D processes. In Q3 2025, Tencent’s marketing services revenue reached RMB 36.2 billion, a 21% year-over-year increase; the company explicitly attributes this growth to its AI-driven precision targeting system. The Yuanbao large model has been deeply embedded in advertising, capable of predicting which users will click, download, or pay—allowing the same ad placements to be sold at higher prices. For Tencent, AI is not merely a “tool,” but a direct lever driving cash flow.
The gaming side isn’t standing still either. “Jili,” the digital ambassador for Peacekeeper Elite, has partnered with DeepSeek, and the AI teammates in Area Breakout can understand complex commands. Tencent has revealed nine major advancements in game AI applications, spanning from art assets to dialogue engines—with broad coverage.
NetEase chose a different path. The pre-registration count for “Yanyun Shisiliu Sheng” surpassed 80 million, making it one of the most anticipated domestic MMORPGs of 2025, featuring numerous NPCs powered by AI that players can freely converse with. During the 2025 earnings call, management made it clear: they would not blindly pursue general-purpose large models, but instead aim to become “the most game-savvy AI experts,” with integration capability as their core competitive advantage. In 2025, NetEase’s net revenue from online games reached approximately RMB 89.6 billion, with AI now serving as everyday infrastructure.
Juren Network went even further, developing a specialized large model called GiantGPT exclusively for gaming scenarios. Games like “Among Us” directly integrate three domestic large models—Alibaba’s Qwen, Tencent’s HunYuan, and ByteDance’s Doubao—with AI-driven agents participating as companions in every match. miHoYo’s “Whispers from the Star” has made real-time AI-driven dialogue the core gameplay mechanic, going far beyond simply adding AI agents.
Shijie Huatong’s “Art Assets and AIGC Creation Management Platform” claims a more than 60% increase in art asset production efficiency. Youzu’s YOOZOO.AI has reduced localization translation costs by 80%. These are real, tangible cost reductions—not just PowerPoint slides.
Global giants: Those who bet, bet big; those who hesitate, hesitate.
EA’s stance is aggressive. CEO Andrew Wilson has publicly stated multiple times that AI will be a core strategic focus, aiming to accelerate development with AI. However, internal tensions persist—employees at EA expressed clear concerns between 2025 and 2026 about potential job losses due to AI.
Ubisoft is another prime example. Ubisoft Ghostwriter was launched as early as 2023 to help writers generate draft dialogues for NPCs. Between 2025 and 2026, Ubisoft continuously hired for AI roles, publicly stating that AI was key to reducing costs and improving efficiency. But the outlook has been difficult: in fiscal year 2026, the company reported a massive loss of approximately $1.4 billion, and rumors emerged that AI experiments for Far Cry 7 had encountered setbacks. AI can save money, but it cannot rescue a project that is fundamentally flawed.
Take-Two’s stance is the most interesting. CEO Strauss Zelnick has repeatedly confirmed that not a single element of GTA 6 was created using generative AI. Yet he hasn’t taken an anti-AI position either; instead, he has praised Google’s Genie world model. His exact words are worth reflecting on: AI can generate game assets, but it cannot create a blockbuster like GTA. This is the ultimate expression of the “tool theory”—AI can lay the foundation, but it cannot provide the soul.
Co-CEOs of CDPR publicly expressed concerns about AI. Former executives at Take-Two also dampened enthusiasm for “AI-made games” in several talks in 2026. Larian Studios faced backlash from players after its CEO stated the studio was using AI; CEO Sven Vincke was forced to clarify that the new Deus Ex game would include no AI-generated content, and he made individual pledges on Reddit regarding art, voice acting, and copyright.
The most resolute is Nintendo. President Shuntaro Furukawa has publicly stated multiple times that Nintendo has no plans to use generative AI in its own games, preferring instead to rely on years of accumulated expertise to deliver unique experiences. In October 2025, a politician claimed Nintendo was lobbying the government to restrict AI; Nintendo swiftly issued a statement denying this, while emphasizing it would “take appropriate action against infringement.” The stance is clear—no AI, but strict protection of IP. Multiple games on the Switch 2 platform have explicitly refused to integrate AI tools.
Microsoft has taken a middle path, primarily promoting assistant-type products like Copilot; Razer is also developing in-game intelligent assistants. These tools do not directly generate game content, face less controversy, and function more like selling shovels to the ecosystem.
Small and medium-sized enterprises: The shovel of AI digs up more than just gold
Small teams most need AI—and are most willing to use it. “No artists required; everything created with Midjourney; only two full-time staff”—this kind of team is no longer unusual in 2025.
But the backlash is also the strongest. AI has lowered the unit cost, but the equilibrium price across the entire industry has followed suit. Previously, small teams could carve out a niche through handmade quality; now, everyone is competing on speed, the number of skins, and asset throughput. Cost reduction is not a moat—it’s the new floor.
Who are the winners? Those with data, distribution, and player mindshare. AI has lowered the barriers to production, making control over distribution even more valuable. This is the true power shift in the gaming industry from 2025 to 2026.
Don’t just look at “penetration rate”—look at resilience.
In the GDC 2024 report, 23% of respondents said they were “not at all interested” in AI; in 2025, this figure rose to 27%. The percentage of those who believe AI will have negative impacts increased from 18% to 30%. These are the attitudes of industry professionals themselves—not the sentiments of casual players.
Players reacted more directly. “Six-Fingered Santa” appeared in Call of Duty 21: Black Ops 6. Battlefield 6 skins were exposed as疑似 plagiarizing assets from Call of Duty 10, years old, yet the developers had explicitly promised before launch that they wouldn’t use AI. In 11 bit’s new title, Many Lives, the “Captain’s Log” directly includes lines like, “Of course, this is a revised version focused solely on scientific and astronomical data”—a classic AI-generated phrasing. The Portuguese subtitles even begin with, “Of course! Here is the text translated into Portuguese.” This forced the developer to issue a public apology.
The most dramatic event was the 2025 TGA, where “Light and Shadow: Expedition 33” won nine awards, setting a record. Ten days later, The Indie Game Awards, which had also named it “Game of the Year,” issued an emergency statement revoking its eligibility due to the game’s use of generative AI. Sandfall Studio explained that AI was only used briefly in the early stages for placeholder textures, which were later replaced, but the award was still rescinded.
The labor side’s resistance was stronger. In July 2024, SAG-AFTRA organized a strike involving over 2,600 voice actors targeting major studios such as Activision Blizzard, Warner Bros., EA, and Epic, impacting more than 160,000 people. The strike lasted nearly a year, concluding in 2025 with the signing of the 2025 Video Game Agreement, which became the model for future AI voiceover provisions.
In early 2026, Capcom’s “Mega Man: Double Overload” was issued a direct work ban by SAG-AFTRA due to AI voice acting issues, prohibiting union members from participating in the project. This is one of the strongest tools in the union’s arsenal; invoking it signals that labor disputes over AI in the gaming industry have moved from the negotiating table to the supply chain.
The platform is also tightening its rules. Starting in 2024, Steam required developers to disclose their use of AI. After being publicly criticized by Epic’s CEO in January 2026, it further updated its policies: generative AI-generated content must be disclosed and is subject to player monitoring and reporting; AI-assisted programming does not require disclosure. This is a smart compromise—targeting content, not tools.
At this point, a judgment can be made.
Blockchain games and AI are not playing the same script. Blockchain games failed due to the absence of product-market fit and an overemphasis on financial logic—they essentially packaged a pseudo-demand with tokens. AI is different—it reduces real costs, transforms actual workflows, and enables genuine gameplay. Tencent’s ad targeting, NetEase’s NPCs, miHoYo’s narrative dialogues, and Shijihuatong’s asset pipelines aren’t stories—they’re numbers already proven in financial reports.
But “having potential” doesn’t mean “everyone wins.” This wave of AI’s impact on the gaming industry will unfold along at least three fronts:
The first issue is the truth of the proposition regarding “AI-native games.” Musk has stated that xAI will launch an “excellent AI-generated game” by the end of 2026; Huang Renxun says such games will emerge within 5 to 10 years; and Google’s Genie 3 can already generate coherent virtual worlds at 720p for several minutes. While world models are indeed evolving, claims that AI will “replace traditional engines and AAA production pipelines” still fall within the realm of speculative promises for 2026. Take-Two’s Zelnick is correct—AI can generate assets, but it cannot create a game like GTA.
The second point is the double-edged sword of cost reduction and homogenization. AI enables small and medium-sized teams to access part of AAA-level productivity—“creation equity” is real; but when the entire industry uses AI to cut costs, competitive moats also shrink. In the end, those who remain are still the ones with IP, data, and distribution capabilities. This hasn’t changed much over the past decade.
The third point is the chain reaction of compliance, ethics, and employment. SAG-AFTRA, Steam disclosure rules, the Interim Measures for the Administration of Deep Synthesis of Internet Information Services, and the EU AI Act—these are institutional variables, not technical ones. China’s new regulation on identifying AI-generated content, effective September 1, 2025, mandates that AI-generated content must carry explicit or implicit identifiers, creating a structural impact on the gaming industry. While AI can do many things, the number of things it can do while remaining compliant may be far smaller.
Blockchain games are dead because they are fundamentally financial games. AI won’t die because it is fundamentally a productivity tool. But tools don’t decide everything. What decides everything is whether people are willing to pay for what’s created with them—this is a challenge blockchain games have failed, and AI cannot avoid either.
The hot trends keep changing. Those who last until the end are never the ones who tell the best stories, but the ones who do the math right and make the game truly enjoyable. It’s a simple saying, but in 2026, it matters more than any concept.
- Caladan, “Web3 Gaming Reality Check: Why 93% of Crypto Games Failed,” DocSend Report, April 2026.
- BlockTempo, “Web3 Gaming Is Dead! Caladan Investigation: 93% of Web3 Games Are Effectively Defunct,” April 2026.
- Huxiu, “By 2026, How Deeply Has AI渗透ed the Gaming Industry?” January 2026.
- OFweek AI, “AI Begins Generating Revenue for Game Publishers: Tencent’s Strategy, NetEase’s Obsession, and the Make-or-Break Moment for Mid-Tier Developers,” January 2026.
- GameLook, “Steam New Rule! AI-Generated Game Art, Audio, and Video Must Be Disclosed,” January 2026; “AI Usage Surges Eightfold! 8,000 Steam Games Utilize AI in Development,” July 2025.
- The Hollywood Reporter / Variety, “SAG-AFTRA Video Game Strike Over: Deal Ratified,” 2025.
- Zhongguancun Online, “Ubisoft Reports $1.4 Billion Loss in Fiscal Year 2026; R&D Efficiency and AI Transformation Spark Industry Debate,” 2026.
- Dongdian Technology, “Can AI Create the Next GTA? Take-Two CEO Supports AI but Throws Cold Water,” May 2026.