Intelligence from the financial app indicates that Orient Securities has published a research report stating that Google DeepMind has opened access to the experimental research prototype "Project Genie" for its world model Genie3, allowing users to create, edit, and explore virtual worlds. There is currently no need for excessive concern regarding overseas general world models like Genie3, as their potential for commercial application in gaming is extremely limited. Even with future improvements in duration support, Genie3 is judged to potentially only support limited story-driven games and cannot adequately support massively multiplayer online games that require infinite duration extension. AI game editors are beginning to deliver growth, and companies possessing large Daily Active User (DAU) platforms, especially those capable of integrating proprietary AI engines, are expected to secure a place in the future AIGC gaming platform landscape. The main viewpoints of Orient Securities are as follows:
Event: In the early hours of January 30, 2026, Beijing time, Google DeepMind opened access to the experimental research prototype "Project Genie" for its world model Genie3, allowing users to create, edit, and explore virtual worlds. Concurrently, on the evening of January 30, Huang Yimeng, founder of XD Inc. and the TapTap platform, launched simultaneous livestreams across major platforms to officially announce a new product named "Taptap制造" (TapTap Maker). The technological pathway for domestic AI game editors has already reached conditions suitable for commercial deployment.
Domestic manufacturers, represented by XD Inc. and TENCENT, focus their core technological implementation on deconstructing game production processes and using AI to simulate parts of programmers' work. By encapsulating AI-generated assets into universal hard-coded physics engines, they achieve the integration of AI 3D engine capabilities with traditional game engines, completing the technological transition through a gradual replacement approach. This model is already viable for commercial application at the current stage. The current application of this technology is more concentrated in UGC/PUGC production phases, where creators can use natural language to submit modification requests to the AI game engine, thereby refining more playable game products.
There is no need for excessive concern regarding current overseas general world models like Genie3, as their potential for commercial application in gaming is currently extremely low. Genie3 cannot support production for B2B game companies primarily because it does not output in the universal FBS game engine format and lacks essential game elements like specific numerical values and items; this fundamental limitation is unlikely to change without major underlying architectural revisions. Secondly, from an end-user (C2C) perspective, what Genie3 generates is an uncertain "game" that is not guaranteed to be playable. Thirdly, due to KV cache inflation caused by its autoregressive nature, Genie3 can currently only produce 60-second interactive video-like content; extending the duration involves non-linear cost growth, making cost control unfeasible for direct C2C player engagement at present.
AI game editors are beginning to deliver growth, and there is a long-term positive outlook on platform-based companies. AI game editors, represented by TENCENT's "Peacekeeper Elite: Oasis Genesis," have already surpassed 33 million DAU, signifying that game AI engines are progressively transitioning from empowering the B2B sector to empowering the C2C sector. Leading Chinese game companies, represented by TENCENT and NTES-S, are gradually upgrading their AI game tools and integrating them into games with large DAUs. AI's empowerment of natural language and drag-and-drop game development enables a massive influx of C2C players into the game production process, leading to a substantial increase in UGC game content. However, UGC is not the final destination but an intermediate state for accumulating platform-specific data and traffic, serving as the most crucial step before the future phase of de-manualization in the gaming industry.
Investment recommendations and targets: As AI applications in the gaming industry gradually shift from B2B to C2C, an increasing number of large-DAU games are utilizing natural language AI game engines to drive user co-creation, achieving high growth in both game content and DAU. Firstly, relevant targets include manufacturers with advantages in game distribution channels and large DAUs: TENCENT (00700), NTES-S (09999), XD Inc. (02400), Century Huatong (002602.SZ), Giant Network (002558.SZ), and Kingnet Network (002517.SZ). Secondly, it is advisable to monitor manufacturers with significant project pipelines this year: 37 Interactive Entertainment (002555.SZ) and Perfect World (002624.SZ).
Risk warnings include risks associated with project delays, product competition, policy changes, and game revenue falling short of expectations.
Comments