Companies Are Spending Billions on AI. 80% See Zero Results.

Subtitle: The AI productivity revolution was supposed to transform every industry. A wave of new research says it hasn’t, and economists are dusting off a 40-year-old theory to explain why.

Key Takeaways

  • A study of 6,000 executives found 80%+ of companies report zero productivity gains from AI
  • Goldman Sachs found AI delivers a 30% productivity boost in exactly two areas: coding and customer service
  • 90% of firms say AI has had no impact on employment over three years
  • Economists are comparing this to the 1980s computer revolution, which also failed to boost productivity for decades
  • The average executive uses AI just 1.5 hours per week

The Billion-Dollar Question Nobody Wants to Answer

Here’s a number that should make every CEO uncomfortable: despite hundreds of billions of dollars poured into artificial intelligence, over 80% of companies report that AI has had absolutely no impact on their productivity.

That’s not from some anti-tech think tank. It comes from a study published by the National Bureau of Economic Research this month, surveying 6,000 CEOs, CFOs, and senior executives across the United States, United Kingdom, Germany, and Australia. These are the people making the buying decisions. And the overwhelming majority say the investment isn’t paying off.

Two-thirds of the executives surveyed said their companies are actively using AI. But when asked whether it had moved the needle on productivity or employment, 90% said no. Not “a little.” Not “we’re getting there.” Just… no.

Where AI Actually Works (It’s a Very Short List)

In the middle of all this skepticism, Goldman Sachs published its own analysis in early March. Senior economist Ronnie Walker examined Q4 earnings calls from S&P 500 companies and found something revealing.

Seventy percent of management teams mentioned AI on their earnings calls. But only 10% could actually quantify what AI was doing for specific tasks. And just 1% could point to a measurable impact on earnings.

The companies that did measure real productivity gains reported a median 30% improvement. But here’s the catch: those gains showed up in exactly two areas.

Software development. And customer service.

That’s it.

Coders using AI tools are writing more code, faster. Customer service reps using AI assistants are resolving tickets more efficiently. In those narrow lanes, the technology delivers. For the other 90% of the workforce? Goldman found “no meaningful relationship between productivity and AI adoption at the economy-wide level.”

You’ve Seen This Movie Before

If this sounds familiar, you might be thinking of the 1980s. That’s when businesses started buying personal computers at a furious pace. The future was digital. Productivity was about to skyrocket.

Except it didn’t.

In 1987, Nobel Prize-winning economist Robert Solow made a dry observation that became one of the most famous lines in economics: “You can see the computer age everywhere, except in the productivity statistics.”

What followed became known as the “Solow Paradox.” Despite massive investment in information technology, productivity growth actually slowed, dropping from 2.9% annually between 1948 and 1973 to just 1.1% after that. It took nearly two decades for computers to show up in the economic data, and even then, the productivity surge was brief (roughly 1995-2004) before flattening again.

Apollo chief economist Torsten Slok sees the same pattern today. “AI is everywhere except in the incoming macroeconomic data,” he recently stated.

The 90-Minute Problem

One detail buried in the research helps explain the gap between AI hype and AI reality.

The average executive in the study reported using AI for about 1.5 hours per week.

Ninety minutes. That’s less time than most people spend on social media in a single day. It’s hard to revolutionize your business with a tool you barely touch.

And that’s the people running the companies. Further down the org chart, a quarter of respondents don’t use AI at work at all. Not because they’ve tried it and it didn’t work. They just haven’t started.

This suggests that the “AI productivity gap” isn’t really about the technology. It’s about adoption, training, integration, and the messy reality of changing how millions of people do their jobs.

AI Doesn’t Reduce Work. It Intensifies It.

Harvard Business Review published research in February 2026 with an equally counterintuitive finding: AI tools don’t actually reduce work. They intensify it.

Employees with access to AI tend to work at a faster pace, take on a broader scope of tasks, and extend their work into more hours of the day. Harvard Business School professor Christopher Stanton estimates that AI can currently handle about 35% of typical office tasks. But instead of those workers going home early, they’re filling the freed-up time with more complex, harder-to-measure strategic work.

The result? Workload creep, cognitive fatigue, and in some cases, burnout. The productivity gains that look great on a quarterly report start to erode as workers hit a wall. Quality drops. Turnover increases. The math gets complicated fast.

Early AI adopters are even reporting weaker connections with co-workers and, paradoxically, lower overall productivity. When AI handles the easy repetitive tasks, what’s left is the hard stuff, and the effort-reward balance tips the wrong way.

What the Executives Predict

Despite the current numbers, the 6,000 executives in the NBER study aren’t giving up on AI. They forecast:

  • A 1.4% productivity increase over the next three years
  • A 0.8% increase in output
  • A 0.7% reduction in employment

That last number is the one that catches attention. It sounds small, but Goldman Sachs projects it differently: they estimate 6-7% of workers, roughly 11 million people, will eventually face AI-driven displacement.

Meanwhile, some companies are going in the opposite direction. IBM announced it’s tripling entry-level hires despite AI automation potential, citing the need to maintain its leadership pipeline. The logic: if you automate your way out of training junior employees, you eventually have nobody to promote.

What This Means for You

If you work in software development or customer service, AI is probably already making your day measurably different. The tools are mature enough in those fields to deliver real, quantifiable gains.

For everyone else, the honest answer is: we’re early. Really early. The technology exists, but the organizational change required to use it well barely has. Most companies are in the “buying the tool and hoping for the best” phase.

That doesn’t mean AI is overhyped. The Solow Paradox teaches us that transformative technologies often take 15-20 years to show up in productivity data. Computers eventually delivered. The internet eventually delivered. AI probably will too.

But right now, if your company is spending millions on AI and you haven’t noticed any difference in your daily work, you’re in the majority. Eighty percent of the majority, in fact.

And if someone tells you AI is about to change everything overnight, ask them: have they used it for more than 90 minutes this week?

Sources: Fortune, Goldman Sachs via Ronnie Walker, National Bureau of Economic Research (6,000-executive study), Harvard Business Review, Apollo Economics via Torsten Slok