56% of CEOs got nothing from AI: the $600B Solow Paradox returns

Most executives believe their AI investments are about to pay off. The data says otherwise: 56% of CEOs across 95 countries report getting "nothing out of" their AI spending, according to PwC’s 29th Global CEO Survey of 4,454 leaders. Only 12% saw both revenue growth and cost reduction.

This is not a temporary lag. It is a pattern economists have seen before, and it cost the global economy a decade of stalled growth.

The $600 billion version of a 40-year-old mistake

In 1987, economist Robert Solow observed something that baffled his peers: "You can see the computer age everywhere but in the productivity statistics." Companies had spent billions on mainframes and PCs. Productivity barely moved. It took nearly 15 years of organizational redesign, worker retraining, and process reengineering before IT investments actually showed up in GDP numbers.

Today, Big Tech capex has doubled in two years, hitting $427 billion in 2025 with projections exceeding $560 billion in 2026. J.P. Morgan estimates that $650 billion in annual revenue would be needed "into perpetuity" just to deliver a 10% return on current AI infrastructure. Meanwhile, roughly 90% of firms say AI has had zero measurable impact on either employment or productivity over the past three years.

The Solow Paradox is back. And this time the price tag has six more zeros.

The layoff trap nobody warned about

Companies didn’t just invest poorly. Many fired people based on what AI might do, not what it actually does. Forrester’s Predictions 2026 report found that 55% of employers who laid off workers for AI already regret it. Challenger, Gray & Christmas tracked 54,694 AI-attributed layoffs through November 2025 alone.

The problem: most of those roles weren’t replaced by AI. They were eliminated based on speculation about future capabilities. Now companies are scrambling to rehire, often offshore and at lower salaries, creating a workforce that’s cheaper but less capable.

This pattern hit only 6% of companies that actually profit from AI particularly hard. The gap between "we adopted AI" and "AI generates returns" turns out to be organizational, not technological.

IBM just proved the skeptics right

While most tech companies doubled down on headcount reduction, IBM did the opposite. In February 2026, IBM announced it would triple entry-level hiring in the United States.

The reasoning was blunt. IBM’s Chief HR Officer Nickle LaMoreaux warned that cutting junior roles creates a leadership vacuum five years down the road. "The companies that are going to be the most successful are those that doubled down on entry-level hiring in this environment," she said.

IBM CEO Arvind Krishna went further, rejecting the idea that AI should mean fewer opportunities for graduates. Junior developers now spend less time on standard coding (which AI handles) and more time on customer interaction and problem-solving, skills that AI time savings often vanish into rework if humans aren’t there to manage them.

Why the paradox persists (and when it might break)

General-purpose technologies follow a predictable arc. The technology arrives fast. The organizational changes required to use it arrive slowly. Electric motors took 30 years to reshape factory floors. PCs took 15 years to show up in productivity data.

AI is following the same script. The PwC survey found that CEOs still forecast AI will boost productivity by 1.4% and output by 0.8% over the next three years. But those gains require complementary investments most companies haven’t made: process redesign, worker training, management restructuring.

The companies treating AI as a cost-cutting shortcut (using it as an excuse for layoffs, not the cause) are the ones most likely to see zero returns. The companies treating it as a catalyst for organizational transformation are the ones IBM is betting will win.

What this means for the next three years

The $600 billion question isn’t whether AI works. It does, in specific, well-integrated applications. The question is whether companies will repeat the Solow Paradox playbook: overspend on technology, underspend on the humans and processes that make technology useful, then blame the technology when results don’t materialize.

If 56% of CEOs are getting nothing from AI today, and 55% of those who fired workers for it already regret the decision, the answer is becoming uncomfortable. The biggest risk in 2026 isn’t that AI fails. It’s that companies fail AI.