Nitrode Thinks AI Can Make Game Development Less Miserable

The Macro: Making Games Is Still Brutally Hard

The game development industry has a dirty secret that everyone inside it knows and nobody outside it talks about enough: making games is extraordinarily expensive, painfully slow, and the failure rate is staggering. The average AAA title takes three to five years and costs $50 to $200 million to produce. Indie games take less money but often more relative time, with solo developers and small teams grinding for years on titles that may never find an audience.

The tools haven’t kept pace with the ambition. Unity and Unreal Engine dominate the engine market, and both are powerful, but “powerful” is doing a lot of heavy lifting in that sentence. Learning Unreal’s Blueprint system or Unity’s scripting architecture is a multi-month commitment. Asset creation is still largely manual. Level design is iterative in the most tedious sense of the word. QA is a slog.

AI has started creeping into game development from the edges. Scenario generates concept art. Promethean AI helps with asset placement. Ludo.ai does market research and game concept generation. But these are point solutions. They handle one piece of the pipeline and leave the rest untouched. Nobody has taken a serious run at making AI native to the engine itself.

The opportunity is real, but so is the resistance. Game developers are a protective bunch, and rightfully so. Their craft involves deeply creative work, from narrative design to visual aesthetics to mechanical feel. The suggestion that AI can handle any of that tends to generate pushback. The companies that succeed here will be the ones that position AI as a tool that handles the tedious parts, not the creative ones.

The Micro: Four Friends, Two Universities, One Engine

Nitrode is building an AI-powered game development engine. The specifics of what that means in practice are still emerging, but the pitch is transforming the gaming experience through AI-native development tools.

The founding team is four people who’ve known each other since their teens. Ben Kim and Brian La were college roommates at Cornell, where Ben studied Information Science and Brian studied Computer Science. Sejoon Chang comes from Stanford, where he studied Computer Science and Product Design. Richard Gu rounds out the team, also from Cornell’s Information Science program.

The backgrounds are interesting for what they include and what they don’t. Brian La was previously a Data Scientist at Nexon, which is one of the largest game publishers in Asia (they’re behind MapleStory, KartRider, and Dungeon Fighter Online). That’s the only direct game industry experience on the founding team. Ben’s prior experience was at Wilson Sonsini in data and VC-facing roles. Sejoon’s product design background from Stanford adds a UX dimension.

They came through Y Combinator’s Winter 2025 batch and are based in San Francisco. Sejoon describes himself as a “Civilization and Sim City BuildIt strategist,” which at minimum confirms these are people who play games, not just people who think games are a large market.

The competitive dynamics here are layered. At the engine level, Unity and Unreal are the incumbents, and both are integrating AI features into their existing platforms. Unity acquired Ziva Dynamics for character animation and has been adding AI-assisted tools. Epic Games (Unreal) has MetaHuman for character creation and has been experimenting with generative AI for environments. Any new engine has to contend with the fact that millions of developers already know Unity or Unreal and have years of muscle memory invested.

At the AI tooling level, companies like Rosebud AI and Scenario are building specialized tools that work alongside existing engines rather than replacing them. That’s a less ambitious but potentially more practical approach.

The Verdict

Nitrode is swinging for the fences, and I respect that. Building a new game engine is one of the hardest things you can do in software. Building one that’s AI-native is even harder because you have to get both the engine fundamentals and the AI integration right simultaneously.

The team has strong technical credentials and the Nexon experience gives them at least one person who understands how game studios actually work. But the gap between “AI-powered game engine” as a concept and a product that developers actually adopt over Unity or Unreal is enormous. Game developers don’t switch engines casually. The switching costs are measured in months and years, not days.

In 30 days, I want to see what the product actually does. A working demo that shows a concrete workflow improvement over existing engines. Not a pitch deck, not a vision statement. Working software. In 60 days, have any indie developers used it to build something real? Early adoption signals from actual game makers matter more than anything else. In 90 days, the strategic question is whether Nitrode is building a full engine replacement or an AI layer that could integrate with existing engines. The latter might be a faster path to adoption, even if the former is a bigger long-term opportunity.