AI-written C++ raises compute use up to 8%

AI-generated C++ increased compute consumption by 5% to 8% in a yearlong study of 3.52 million production code changes at an unnamed enterprise whose products serve billions of users. Researchers found more interface and coupling burdens, copy and allocation overhead, and explicit loops in AI-authored code; taxonomy-specific feedback then cut targeted static-analysis warnings by 11.1%. The finding puts a measurable operating-cost tail on the quality debt created by faster code generation: teams should track runtime resource use and review effort, not merge velocity alone.