MiHoYo’s $15 billion bet to become the first AI-native gameco

MiHoYo’s reported plan to invest up to RMB 100 billion, or almost $15 billion, to build a full AI stack over three years only sounds absurd if you still think of the company as the developer of global hit Genshin Impact.

That is not how miHoYo sees itself. Since the launch of HoYoverse, its broader ambition has been to build immersive virtual worlds, not simply to operate a portfolio of anime-style gacha games.

In that context, it doesn't see AI as a bolt-on production tool. Instead, it's the next operating layer for the company’s future – handling everything from NPCs to personalized quests, live-service operations, localization, animation, content pipelines, customer support, creator tools, and eventually more dynamic character interaction.

The strategic logic is clear. MiHoYo’s original moat was its ability to invest massive capex and industrialize high-budget anime games at global scale. Genshin Impact proved that a Chinese studio could build a world-class cross-platform blockbuster. Honkai: Star Rail then proved this was not a one-off. Zenless Zone Zero extended the machine again.

But that advantage is now being copied. The anime open-world and anime RPG market is no longer blue water. Chinese and Korean rivals have studied miHoYo’s production model, monetization structure, art direction and update cadence. In other words, miHoYo’s advantage is becoming an industry template.

AI offers a possible second leap. If miHoYo can build the best internal AI stack for anime-style interactive worlds, it does not merely make development cheaper. It changes what its games can be. Characters could remember, react, improvise and scale beyond scripted content. Live operations could become more personalized. Asset production could accelerate. Testing and balancing could become more automated. The long-term prize is not “AI-generated art” but more responsive, denser, more adaptive virtual worlds.

The question is how miHoYo can discuss deploying this level of capital.

The simple answer is that it is one of the most profitable private games companies in the world. Genshin Impact and Honkai: Star Rail turned miHoYo into a cash machine.

Reported 2022 figures had the company generating RMB 27 billion ($4 billion) in revenue and RMB 16 billion ($2.4 billion) in net profit — an extraordinary margin by the standards of most entertainment businesses. Later revenue estimates vary, but the direction is obvious: miHoYo has had years of massive global cashflow from high-margin digital live-service games.

Its ownership structure also matters. miHoYo is private and founder-controlled. It does not have public-market investors demanding buybacks, dividends or quarterly margin discipline. It has historically resisted becoming someone else’s strategic asset. That gives its founders unusual freedom to recycle game profits into a long-range technological bet.

Still, the apparently flippant tone matters. Chairman Liu Wei’s reported remark that failure could be treated as “setting off a big firework” sounds almost reckless. But that may be the point. Founder-controlled companies can talk this way because they are not asking the market for permission. It is extravagant, but it is not random.

MiHoYo is rich enough to fund the experiment, strategically exposed enough to need a new moat, and culturally ambitious enough to treat AI not as cost-saving software but as the next foundation of its world-building project.