AI and the Divide Between Code and Concrete: Software vs. Hardware

If you want to see where AI may affect jobs first, the stock market may already be pointing to the answer. It shows up in a simple metric: sales per employee.

AI adoption can increase productivity with fewer employees, which is one reason many companies are reducing headcount. Yet the market is selective about where it believes these headcount reductions will be beneficial. Technology companies are the epicenter of AI deployment, and the market’s reaction to job cuts differs sharply between software and hardware firms.

Using sales/employee as a proxy for productivity, the accompanying chart shows the long-short return for monthly-rebalanced portfolios of tech hardware and software company stocks within the top 1,000 U.S. stocks by market capitalization, sorted on sales/employee. The analysis was conducted using the Finsera platform.

Since early 2023, the sales/employee factor’s performance has risen steadily for software companies but declined for hardware companies. 

This divergence reflects the nature of work in these industries.

Software development is dominated by cognitive tasks, especially writing and maintaining code. These are precisely the activities that AI tools increasingly assist with or partially automate.

As AI coding assistants become more capable, software firms may be able to sustain or expand output with fewer engineers. That dynamic raises sales per employee, and the market appears to reward companies that demonstrate the strongest productivity gains.

Hardware production, by contrast, remains anchored in physical processes—manufacturing, supply chains, fabrication facilities, and logistics. AI can improve design and optimization, but it cannot replace the physical infrastructure required to produce chips, devices, and other hardware products.

As a result, sales/employee is less central to competitive advantage in hardware, which may explain why the factor has weakened there.

Taken together, these results suggest how AI may reshape employment.

The impact of AI is likely to be uneven across the economy. Knowledge-intensive industries may experience the greatest labor displacement—or slower hiring—while sectors tied to physical production may see more limited direct effects.

The emerging divide between code and concrete may ultimately determine where AI reshapes employment first.