Solow’s Productivity Paradox comes from Robert Solow’s quote,

“You can see the computer age everywhere but in the productivity statistics.”

That line has been repeated for decades because it describes a pattern most business owners recognise. Technology arrives, teams move faster and work feels easier, yet the numbers barely move. AI may be on the same path. The tools work. Time is saved. But revenue per employee and output per hour stay flat.

The gap comes from what happens to the time saved. In most small firms, nothing structural changes. The volume of enquiries stays the same, so staffing and pricing remain unchanged. A task that once took three hours now takes one, but the extra two hours do not convert into more work. They sit inside the business as spare capacity.

This is trapped capacity. It looks like efficiency, but it behaves like idle time. Estimators finish early and wait for the next job. Admin teams clear their backlog and then return to reactive work. Owners sense the release of pressure and don’t think to question the system. The gain stays contained within the existing flow of work rather than pushing the business forward.

Trapped capacity is breathing room rather than a business asset. Teams protect it after a busy period. The downtime is used to catch up with email, tidy up, or reduce stress. All of that has value, but none of it scales the business. The capacity is not used for growth or cost reduction.

Solow was writing in 1987. He wondered why the rapid adoption of computers in US manufacturing since the 1970s had not driven a golden age of productivity. Could AI have a similar lack of effect today?

One explanation of the paradox was that improvements in value are not captured by standard economic statistics. This is unlikely, because a company should sell more of a valuable product and hence boost profits.

A second explanation focused on timing. Productivity expanded by up to 3% a year from 1995 to 2004. Solow just had to wait. Yet productivity has since resumed a downward trend, suggesting that the smartphone, cloud computing and ecommerce have had little productive impact.

The third explanation was that computers were used for tasks that did not add value. For example, writing longer and more attractive reports. The modern equivalent would be AI slop that burns tokens and generates costs, but does not result in higher profits.

Modern theories suggest that trapped capacity shows up as intangible assets. It exists in human capital and scalable business processes. These only have value if they result in the business scaling, which requires deliberate action. Small businesses may benefit from taking the following three steps.

The first step is to recognise freed time as a resource that must be allocated. If a process takes half the time, someone must decide where the remaining capacity goes. That might mean increasing sales activity, shortening response times to win more work, or expanding the scope of what you offer. If no decision is made, the time will be lost to low-value activity.

The second step is to identify the constraint that limits growth. In many small businesses this is demand, pricing, or decision-making speed. If those areas remain unchanged, faster task completion achieves little.

As an example, a contractor may produce quotes in half the time, but if leads are sporadic, pricing lacks conviction, or approvals sit in an inbox for two days, the additional capacity never translates into revenue. The system absorbs the gain because the limiting factor sits elsewhere.

The third step is to link AI adoption to a commercial outcome from the start. Do not measure success by time saved. Decide whether the goal is more revenue per employee, faster turnaround to win work, or reduced reliance on hiring. Then redesign the workflow to deliver that outcome. This often means changing roles, expectations and targets. An AI tool can support the workflow, but it will not redesign it.

Trapped capacity is a management failure, rather than a technology problem. AI will create the conditions for it in almost every business. The firms that benefit will be the ones that force the gain out of the system and into the numbers. The rest will get another lesson in Solow’s Productivity Paradox.