Organizational adjustment costs involve the learning and workflow changes firms need to use new technologies effectively.
These costs, like retraining workers, often appear after technology is adopted and can slow down productivity gains.
AI inference shows the fastest cost reduction in history, with a 1,000-fold price drop in just two years.
This rapid decline is much quicker than for hardware, energy, or other digital technologies.
Despite fast cost drops and widespread access, organizational adjustment remains a key hurdle for AI's economic potential.
Examples
- Electricity adoption took a generation for factories to reorganize layouts around new motors.
- Information technology productivity gains lagged adoption by over a decade due to workflow redesigns.
- AI's rapid cost reduction means access and usage costs have fallen faster than for comparable technologies.
- The existing digital infrastructure allows AI to diffuse much faster than past technologies like electricity.
- A 2026 survey found 43% of US workers use AI, but productive integration still faces challenges.
It's true that the raw cost of new tech like AI inference dropping so quickly is a big deal for budgets, as a fair reading of the claim suggests.
However, to treat that price drop as the dominant factor for organizational adjustment costs seems to miss the real work involved, especially when dealing with legacy systems or complex regulations.
Making a new AI tool actually work for people means retraining staff, redesigning workflows, and ensuring data migration is smooth, and that effort isn't cheaper just because the server time is.
For instance, getting a new AI system integrated into an existing French administrative service still means dealing with FranceConnect compatibility and strict data privacy rules, regardless of how much the AI itself costs.