A Google DeepMind developer used Anthropic's Claude Code to port "Command & Conquer: Generals Zero Hour," a 2003 real-time strategy game, to iPhone and iPad. The initial working build took 40 minutes to complete. The full source code is now available on GitHub.
This demonstrates Claude Code's capability to handle complex software engineering tasks at scale. The AI assistant managed to translate a two-decade-old PC game codebase to native iOS in hours rather than weeks, a timeline that would typically require multiple specialized developers. The speed reveals how AI-assisted coding accelerates porting workflows that traditionally involve manual refactoring, API rewriting, and platform-specific optimization.
The task involved substantial technical challenges. The original game ran on outdated Windows libraries and DirectX rendering pipelines. Converting it to Swift or Objective-C while maintaining gameplay logic and assets required understanding both the legacy codebase and modern iOS frameworks. Claude Code handled this by analyzing the source code structure, identifying platform-specific dependencies, and generating native replacements automatically.
This isn't theoretical. The developer published working code, meaning the port runs on actual iOS devices with functional gameplay. That level of completeness matters. Previous AI coding demos often produced partial or non-functional results.
The implications extend beyond gaming. This showcases AI's value for legacy software modernization, a persistent problem across industries. Banks, insurance companies, and enterprises maintain millions of lines of aging code that costs money to maintain but generates revenue. Accelerating porting and modernization could unlock significant business value.
However, speed doesn't equal perfection. A few-hour turnaround leaves questions about optimization, performance on different hardware, edge case handling, and testing depth. Mobile games have strict battery and performance budgets that rapid ports might violate.
The experiment signals a shift in how developers approach large codebases. Rather than treating