Capcom Plans to Turn RE Engine Into an AI-Assisted Game Engine
Capcom programmer Satoshi Ishida said on October 2 that Capcom plans to gradually turn its RE Engine into an "AI-generation game engine."
Ishida spoke at the Capcom Open Conference RE:2026 in a session titled "The Outlook and Future of the REX Project, Further Evolving the RE Engine for the Next Generation." The Verge The talk laid out the outlook for REX and the next generation of the engine, according to a report published October 3. Gamereactor
The problem, Ishida said, is scale. Building games at the size of Resident Evil can make even simple tasks take a very long time. His proposed answer was "successfully integrating AI technology into development workflows."
He described the plan as a slow update to RE Engine tooling, not a full replacement. The stated goal is "a future where we create games together with AI." IGN reported the same wording on October 3, 2026.
Capcom has previously said it will not use AI-generated assets in its games and will instead use the technology to make development more efficient. The focus in Ishida's account is faster workflows, faster iteration, and less repetitive work, not shipping AI-made art, audio, or code directly to players.
REX stands for RE neXt Engine. Capcom is developing REX as the next generation of its engine by improving RE ENGINE in stages rather than replacing it, according to company materials. Capcom The company has said it plans to add functions to RE ENGINE in stages, wording that dates to a session called "RE ENGINE's Past and Future" at Capcom Open Conference Professional RE:2023. Capcom Capcom also said there are currently no plans to release RE ENGINE to outside developers.
AI assistance inside the engine is not new. The 2023 conference included a presentation on Navigation AI technologies in RE ENGINE. Capcom That earlier work covered pathfinding, or how characters move through levels, agent behavior, and editor support. The REX plan extends that tooling to wider development work.
The broader context here will be familiar to teams running large projects. Generative AI is rapidly becoming part of game development, as Reuters reported on July 22, 2026. A separate Reuters report on April 22, 2026 said advanced AI tools could cut development costs by nearly half, possibly unlocking about $22 billion in profit for the industry.
In my view, the technical question is narrower than those cost forecasts suggest. An AI-generation engine for experts does not mean typing a prompt and getting a full game. It means tighter loops inside existing systems: search across large collections of game files, automatic creation of simpler model versions (LOD) and simplified collision shapes for human review, test coverage for scripted events, draft text for translations, and checks on game builds. For teams working at Resident Evil scale, with dense interiors, lighting passes, and long testing periods, small time savings add up fast.
Looking at what this means for engine architects, the hard parts are integration, not models. Version control for AI-assisted edits, records of where generated drafts came from, builds that run the same way each time, and clear approval steps in the editor will decide whether the tools save time or create extra work. Keeping generated material out of shipped games while allowing it in drafts and automation simplifies rights review and art direction, but it needs strict separation so temporary content never slips into final files.
The long arc here remains constructive. Better tooling has repeatedly let smaller teams attempt larger worlds, and let large teams spend more time on design instead of manual cleanup. If Capcom carries out this staged change inside its private engine as described, the near-term result would be faster internal iteration rather than visibly different games.
