Zhongjin: AI + Games evolve from tool iterations to ecosystem reconstruction. AI empowers Evergreen games.

CICC released a research report stating that recently, many gaming companies have announced progress in AI game tools, AI gameplay/native applications, and AI industry investments. In the industry, the 2025 Global Game Developers Conference (GDC2025, held on 3/17-21) also released relevant forum content, with many discussions on AI intelligent agents/AI game development. The exploration of AI tools and gameplay in the gaming industry is expected to deepen and upgrade further with the reduction in costs and efficiency improvements due to large-scale models and increased investment from leading companies. As for the target stocks, Kingnet Network (002517.SZ), Giant Network Group(002558.SZ), NetEase-W(09999), etc. are recommended. Key points from CICC: Research and Development: Multi-modal comprehensive experimental field, integrating audio-visual-textual interaction and other multiple applications; Leading companies are deepening the development of AI game production tools to optimize the development process. CICC believes that games, as a comprehensive experimental field for AI multi-modal applications, are gradually using generative AI to reconstruct the production process, playing a role in the entire process of game design, development, and testing to promote intelligent research and development processes. For example, in development, AI completes the automated generation of initial characters/scenarios, followed by optimization and iteration by developers. Currently, Tencent, NetEase, and other gaming companies have a high penetration rate of AI applications in various aspects of their internal gaming business, with companies like Kingnet Network and Giant Network Group using their own AI game tools to accelerate game content generation efficiency. Overseas, Microsoft released the game large model Muse on February 19, supporting the rapid generation of multiple subsequent game scenes through a game screenshot. Operation: AI empowers evergreen games, exploring AI NPC/Agent gameplay in the short term, and the future may achieve ecological restructuring to extend the lifecycle. 1) Diversification of AI NPC/Agent: After the recent popularization of DeepSeek, Tencent's "Peace Elite" has introduced the digital spokesperson "Gilly" to DeepSeek (started grayscale testing on 2/25, will be opened to all users in the short term). NetEase has incorporated the "Shen Qiu Suo" NPC character into the mobile game "Nishui Han," previously introducing a martial arts girl character integrated with five major models, opening up a new paradigm of AI interaction in the game. In the future, AI NPC may provide more emotional interaction, create a dynamic narrative network, and enhance playability. 2) Cost optimization drives AI gameplay exploration: With the lowering costs of base large models, AI gameplay within games like "Space Kill" is expected to further expand user coverage and deepen AI attributes. 3) UGC ecosystem upgrade: Platforms like "Egg Party" and "Roblox" are expected to stimulate user output through AI-assisted design tools. Focus on breakthroughs in "2D to 3D" technology, creative development of AI-native gameplay, and empowering reshaping of existing gameplay At the technical level, from text translation, story dialogue generation, to image painting, UI design, and audio-video animation applications, different levels of tool models have emerged and landed. The industry expects "2D to 3D" generation to achieve more accurate and convenient breakthroughs in vertical models. In terms of gameplay, one is to focus on empowering reshaping existing gameplay (such as companion games, puzzle games); the other is to focus on the emergence of creative AI-native gameplay. Risk factors Macroeconomic impact, industry policy changes, technology development falling short of expectations, and game performance falling short of expectations.