Technology and Innovation Community

You probably saw Anthropic post on 18th Feb highlighting AI agents penetration per subindustry/tasks based on Anthropic's API usage:

That said, GDP alone is an incomplete benchmark since large industries dominated by physical labour, such as construction, should not command the same AI focus as knowledge-intensive sectors like finance. A more meaningful measure is the share of automatable work: GDP share × automation potential

• On this basis, healthcare accounts for roughly 19.2% of automatable work but only 1% of AI usage - an 18.2% gap.
• Marketing represents about 1.2% of automatable work yet captures 4.4% of AI usage - a 3.2% overshoot.
  

The divergence suggests AI attention is not yet aligned with underlying economic opportunity.

------------------------------
Carlos Salas
Portfolio Manager & Freelance Investment Research Consultant
------------------------------

Very good point Carlos.

Healthcare should be a major area of focus for automation going forward. Below is a quick snapshot of the price changes in consumer goods and services in the US for the past 30 years. The data is CPI adjusted and Medical care is one of the areas with the highest %-age increase +140%.

Original Message:
Sent: 23-02-2026 07:57
From: Carlos Salas
Subject: AI Adoption vs Economic Reality

You probably saw Anthropic post on 18th Feb highlighting AI agents penetration per subindustry/tasks based on Anthropic's API usage:

Anthropic's API usage data offers an incomplete picture when viewed in isolation:

• Marketing and copywriting account for approximately 4.4% of observed AI usage. While this may initially appear to signal untapped opportunity, the broader economic context suggests otherwise: marketing represents only about 1.2% of total economic output. In relative terms, AI attention directed toward marketing exceeds its share of the economy by roughly 3.7 times.
• By contrast, healthcare comprises nearly 18% of the economy yet accounts for only about 1% of AI usage. The disparity here is substantially more pronounced.

The most compelling opportunities are therefore unlikely to be found in sectors already saturated with AI experimentation. Rather, they exist in industries that command significant economic weight but remain comparatively underpenetrated by AI adoption.

That said, GDP alone is an incomplete benchmark since large industries dominated by physical labour, such as construction, should not command the same AI focus as knowledge-intensive sectors like finance. A more meaningful measure is the share of automatable work: GDP share × automation potential

• On this basis, healthcare accounts for roughly 19.2% of automatable work but only 1% of AI usage - an 18.2% gap.
• Marketing represents about 1.2% of automatable work yet captures 4.4% of AI usage - a 3.2% overshoot.
  

The divergence suggests AI attention is not yet aligned with underlying economic opportunity.

------------------------------
Carlos Salas
Portfolio Manager & Freelance Investment Research Consultant
------------------------------