There's a number going around that should bother you more than it does: for every dollar companies spend on AI coding tokens, a large chunk goes straight back into fixing the bugs that same AI produced. The speedup is real — I feel it every day, I'm not here to tell you AI coding is fake. But "faster" and "cheaper" are not the same word, and 2026 is the year the bill started arriving.
TL;DR — AI doesn't give you a productivity gift, it gives you a loan: speed now, paid back later in debugging, review, and rewrites. Reporting around an Entelligence AI figure puts the "interest" at roughly 44 cents of every token dollar going to fixing AI-generated bugs. The loan is still worth taking — for the right tasks. The trap is spending borrowed time like it's income.
The number
The stat that kicked this off: a widely-shared claim from Entelligence AI, reported across tech press, that companies spend about 44% of their tokens fixing bugs their own AI generated. The fuller breakdown making the rounds is even starker — for every $1 of token spend, ~$0.44 goes to bug fixes, ~$0.27 to rewriting AI output, ~$0.11 to review and merge delays. The pitch version: spend $100k on tokens, ~$18k reaches stable production.
