Capcom is slowly turning its in-house RE Engine into what it calls an “AI-generation game engine.” The plan came out at the Capcom Open Conference RE: 2026, where programmer Satoshi Ishida laid out a long-term goal: “a future where we create games together with AI.” The Verge AI reports on the talk, citing IGN’s translation, and notes the irony: Capcom’s upcoming game Pragmata is all about the horrors of AI.
This matters because RE Engine isn’t a side project. It runs Capcom’s biggest franchises, Resident Evil included. When a studio at that level says AI belongs at the core of its engine, other studios notice.
What Capcom Actually Said
Ishida’s talk had a long title: “The Outlook and Future of the REX Project, Further Evolving the RE Engine for the Next Generation.” The main point was easier to follow. Games at Resident Evil’s scale are so big that even simple tasks can eat a huge amount of time.
His answer is “successfully integrating AI technology into development workflows.” The rollout he described is slow and step by step. Capcom isn’t swapping its pipeline overnight. It plans to evolve RE Engine bit by bit toward a tool where AI plays an active part in making games.
Here’s what we know so far:
- The project: REX, Capcom’s next-generation upgrade of RE Engine
- The goal: Gradually turn RE Engine into an “AI-generation game engine”
- The stated problem: Huge AAA projects make routine tasks painfully slow
- The existing policy: Capcom has said it won’t use AI-generated assets in its games
The Asset Promise and the Open Door
That last point is where things get interesting. Capcom has said before that it won’t ship AI-generated assets and that it’ll use AI to make development more efficient instead. Ishida’s comments mostly fit that position.
Still, “create games together with AI” goes further than “use AI to speed up bug fixing.” As The Verge AI points out, the wording leaves the door open to much wider use. If AI becomes a real co-creator inside the engine, the line between an AI-assisted workflow and an AI-generated asset gets hard to draw.
I think that gap is the real story. Studios have been making careful public promises about AI content because players and creative workers have pushed back hard. Video game voice actors went on strike over AI protections not long ago. Capcom’s engineers seem to be planning for a future the company’s public policy doesn’t fully cover yet.
Why Big Studios Are Moving This Way
Capcom has plenty of company. AAA budgets and timelines keep growing, and some games now take five years or more to make. Publishers across the industry are trying AI for QA testing, code help, animation cleanup, localization, and level prototyping. Square Enix, for one, has talked openly about automating a large share of its QA work.
Usually, the first targets are the boring jobs that pile up quietly:
- Testing and bug triage, where AI agents can play builds and flag problems
- Pipeline chores, like asset tagging, file conversion, and build management
- Code assistance, which speeds up engine and tool work
- Prototyping, so designers can test ideas before artists commit hours to them
Capcom’s advantage is that it owns its engine. Studios on Unreal or Unity wait for Epic or Unity Technologies to ship AI features. Capcom can build AI right into the tools its teams already use every day.
The Future Cast: 2027 to 2029
Over the next one to three years, expect AI to arrive in RE Engine quietly. It’ll show up in tooling and automation first, not in anything players can see. Capcom’s “slowly and incrementally” wording points to a long rollout tied to its next generation of releases.
The bigger question is policy. As the tools get better, pressure will grow to let AI draft more of the actual content, such as placeholder art that becomes final art, generated variations, or procedural dialogue. Watch whether Capcom’s no-AI-assets promise holds, gets sharper, or gets quietly reworded.
For game developers, the takeaway is practical. Engine-level AI is becoming standard at the top of the industry. Learning to direct, check, and fix AI output inside production tools will probably matter more than learning any single AI app. Studios that own their engines will set the pace, and everyone else will follow.
You can find more details on Ishida’s presentation in the original report from The Verge AI.