Tokens Are the Cheap Part — Steve Jamison, Jellyfish
Developer salaries, not tokens, are the dominant AI cost organizations must optimize against
“these two numbers cannot tell the whole story. And even together they don't take into account the largest expense item.”
Jellyfish's Head of AI R&D presents platform data showing AI spending rose 60x year-over-year while engineer PR throughput increased 43%, but argues token costs are misleading because developer salaries dwarf them as the true expense. The core insight is that AI ROI frameworks focused on token spend miss the point — the real question is whether AI productivity gains justify the fully-loaded cost of engineering headcount. This reframes how engineering leaders should pitch AI budgets to CFOs and boards.