AI BriefingThe point at which AI costs exceed engineer salaries
Key point
Analyzes the structural shift in which AI infrastructure costs surpass engineer labor costs, along with market scenarios.
Details
Frontier AI companies like Anthropic show an infrastructure-centric cost structure, with compute costs reaching about 2.3x salary. In contrast, typical software companies spend only 0.4% of engineer salary on AI, with the median figure being close to zero.
Three scenarios are projected for future market changes:
- Bear (Decline): Token prices continue to fall, with AI cost per engineer staying at around $106k by 2029
- Base (Baseline): Following the trend of the top 1% of companies, AI cost per engineer rises to about $363k by 2029
- Bull (Rise): Following Anthropic's model, AI cost per engineer reaches about $596k (230% of salary) by 2029
Key drivers of the Bull Case include a sharp increase in token consumption from Agentic workflows and demand outpacing supply. On the other hand, variables for the Bear Case include token prices falling 10x every year and the rapid growth of Open-weight models.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.