Capcom announced plans on October 2 to evolve its RE Engine into an AI-generation game engine
through a multi-step improvement project called REX. Revealed at a technology conference in Japan, the initiative aims to streamline massive AAA developments while keeping artificial intelligence focused on development workflows rather than art assets.
The Capcom Open Conference RE: 2026 brought together developers to inspect the future roadmap for the company’s foundational technology. Programmer Satoshi Ishida took the stage to outline the multi-step improvement plan, designated as the REX initiative, which will incrementally upgrade the engine rather than replacing it with a ground-up rewrite.
Satoshi Ishida Details REX Project Challenges at Capcom Tech Event
Speaking at the conference, Ishida pointed to three distinct pressures that have accumulated over the engine’s lifespan. Having debuted alongside Resident Evil 7 in 2017, the proprietary technology has powered nearly 30 titles and is used daily by more than 2,000 developers across international teams.
Modern video games have grown so massive that routine processes take excessive amounts of time, creating severe iteration bottlenecks. Furthermore, larger and more diverse development teams require a system that remains accessible to everyone. To solve these friction points, Capcom determined that integrating AI into workflows has become essential for streamlining future production.
The REX multi-step improvement plan specifically targets efficiency, performance, tools, and quality assurance to counteract the rising costs and time requirements associated with modern AAA production.
Unifying Codebases Around Common Programming Languages for AI Understanding
Rather than building a brand-new foundation, the REX project will layer upgrades onto the existing framework. A core pillar of this evolution involves unifying the engine’s codebase under one of the world’s most common programming languages. This standardization makes the architecture easier for human developers to navigate while simultaneously allowing artificial intelligence systems to read and interpret the underlying code.
By making the engine intelligible to machine learning models, Capcom expects the technology to eventually assist human programmers by writing new code, running automated playtests, and checking for bugs during quality assurance. To support this ongoing training, the publisher also intends to make some of its technologies publicly available on the internet.
The technical roadmap includes faster data handling, lighter development tools, a rebuilt logging and networking setup, easier gameplay prototyping, and support for scenes featuring significantly more characters on screen at once.
Capcom Maintains Strict Boundaries Against AI-Generated Game Assets
Despite the push toward machine learning integration, studio executives have drawn a sharp line regarding how artificial intelligence will be applied in production pipelines. Company leadership reiterated that automated systems will serve strictly as productivity multipliers rather than creative replacements.
“We will not implement assets generated by AI into our games.”
Capcom, March 2026 shareholders meeting
Instead of generating graphics, audio, or character models wholesale, the technology will be deployed across graphics, sound design, and programming to optimize efficiency. Shinichi Inoue, vice president of game development platform and AI solutions, previously emphasized that the publisher views artificial intelligence not for creating art, but for unleashing the potential of creators by restoring person-to-person communication and clearing away development bloat.
Inoue explained that the overarching vision is to return to an environment where team members can discuss concepts, build consensus, and bring ideas to life with a shared understanding of intent. This collaborative medium relies on better contextual communication rather than wholesale asset generation.