When AI Costs More Than the Engineer
tomtunguz.com
July 6, 2026
4 min read
🔥🔥🔥🔥🔥
51/100
Summary
Anthropic allocates 2.3 times its payroll to compute costs, amounting to approximately $2 million per employee annually, with a projected $10 billion spend on inference and training in 2026. In contrast, the top 1% of companies spend $89,000 per engineer on AI, while the median spend is $137, highlighting a significant disparity in AI investment across the software market.
Key Takeaways
- Anthropic spends 2.3 times its payroll on compute, amounting to approximately $2 million per employee per year against an all-in compensation of over $500,000.
- The top 1% of companies spend $89,000 per engineer per year on AI, while the median company spends only $137, highlighting a significant spending gap.
- By 2029, AI spending per engineer could reach $596,000 in a bullish scenario, matching the revenue contribution of a median SaaS employee.
- Goldman Sachs projects a 24-fold increase in token consumption by 2030 due to the rise of agentic AI workloads.
Community Sentiment
Positives
- Open-weight models are poised to disrupt the cost landscape, allowing users to achieve similar results for a fraction of the price — a game changer for cost-conscious developers.
- Some users are finding that local models meet their needs effectively, suggesting a shift towards more accessible AI solutions that can handle real-world tasks competently.
- The conversation hints at a growing recognition that AI can be used for rapid prototyping, with some users advocating for leveraging AI to kickstart projects before refining them manually.
Concerns
- There’s a palpable frustration with the last-mile problem in AI, where it can handle most tasks quickly but then stalls on the final touches, leading to wasted time and effort.
- Critics argue that comparing training costs of LLM developers to companies using the models is misleading, undermining clear understanding of AI expenses and value.
- The sentiment that tech leadership is out of touch is strong, with many feeling that AI has exposed the incompetence of high-level decision-makers in the industry.
