We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Hacker News·Jul 6·5 min read
When AI Costs More Than the Engineer
TL;DR
A significant gap exists in AI spending between leading companies and the broader software market, with Anthropic spending 2.3 times its payroll on compute resources. The analysis explores three future scenarios for how this disparity might close by 2029, focusing on AI cost per engineer.
✦ Why It Matters
Engineers should assess their AI spending strategies to align with industry leaders and remain competitive.
Key Takeaways
- Evaluate your company's AI spending relative to industry leaders.
- Consider future scenarios for AI investment growth.
- Explore ways to optimize compute resource allocation.
- Stay informed about trends in AI cost management.
Full Summary
Anthropic, a leading AI company, allocates 2.3 times its payroll on compute resources, translating to approximately $2 million per employee annually, compared to a fully-loaded salary of around $500k. In contrast, the top 1% of software companies spend only $89k per engineer per year on AI, which is about 40% of a senior engineer's salary.
The analysis presents three scenarios for 2029: Bear, where token deflation prevails; Base, where the top companies' growth slows; and Bull, where the rest of the market matches Anthropic's spending ratio. Each scenario predicts different annual AI costs per engineer, highlighting the potential for significant shifts in AI investment across the industry.
The findings underscore the stark disparity in AI resource allocation and suggest that the broader market may need to increase its investment to remain competitive.