56% of CEOs got nothing from AI, but told Wall Street otherwise

Of the 374 S&P 500 companies that mentioned AI in their most recent earnings calls, the vast majority described their implementations as entirely positive. Meanwhile, PwC surveyed 4,454 CEOs across 95 countries and found that 56% reported zero measurable improvement in either revenue or costs from their AI investments. Only 12% achieved gains in both.

That gap between the boardroom narrative and the balance sheet is not a rounding error. It is the defining corporate contradiction of 2026.

The AI ROI paradox nobody is discussing publicly

Apollo Chief Economist Torsten Slok put it bluntly: "AI is everywhere except in the incoming macroeconomic data." His observation echoes a pattern economists have seen before. In 1987, Nobel laureate Robert Solow noticed that despite massive investments in computing, productivity growth had actually declined, from 2.9% annually between 1948 and 1973 to just 1.1% afterward. His famous line: "You can see the computer age everywhere but in the productivity statistics."

A National Bureau of Economic Research study covering 6,000 executives across the U.S., U.K., Germany, and Australia found that 90% of firms reported zero impact on employment or productivity over three years of AI usage. Average AI usage among those firms? Just 1.5 hours per week. A quarter of respondents did not use AI at the workplace at all.

The Solow paradox is back, and this time the numbers are larger.

Three structural reasons your AI spending produces nothing

PwC Global Chairman Mohamed Kande identified the core problem during Davos 2026: "Somehow AI moves so fast that people forgot that the adoption of technology, you have to go to the basics." His diagnosis pointed to three foundational gaps.

1. Dirty data, expensive models

Most organizations lack the clean, structured data AI systems require. Deloitte's global AI ROI analysis found that only 6% of companies achieved payback within one year, compared to the 7-to-12 month timeline typical for conventional technology investments. The bottleneck is rarely the model. It is the data infrastructure underneath it.

2. The productivity J-curve nobody budgeted for

MIT Sloan researchers analyzing tens of thousands of manufacturing firms found that AI adoption caused an initial 1.33 percentage point productivity drop. Established firms suffered most: older organizations saw declines in structured management practices after implementing AI, accounting for nearly one-third of their productivity losses. The technology demands parallel investments in training, workflow redesign, and process adaptation that most budgets never included.

At the individual level, research shows AI often increases workload rather than reducing it. Workers spend time prompting, verifying outputs, and managing tools instead of doing the work AI was supposed to handle.

3. Competitive panic disguised as strategy

Despite 56% seeing no returns, 91% of organizations plan to increase AI spending this year, according to Deloitte. BCG's AI Radar report found that companies plan to double their AI budgets in 2026, allocating roughly 1.7% of revenues. One executive captured the dynamic perfectly: "If we do not do it, someone else will, and we will be behind."

This is not investment driven by evidence. It is spending driven by fear. When only 6% of companies generate real AI profits, the 94% pouring money in are essentially subsidizing a competitive arms race with no proven payoff.

Why this matters more than the dot-com bubble

The dot-com crash destroyed speculative startups. The AI productivity paradox is different: it is draining operating budgets of established companies while their public narratives suggest the opposite. Gartner predicts 30% of generative AI projects will be abandoned after proof-of-concept by end of 2025, with 60% of AI projects unsupported by AI-ready data abandoned through 2026.

The pattern suggests more than half of companies regret their AI workforce decisions for a reason: the technology was deployed to replace processes that were never properly understood in the first place.

What the 12% who succeeded actually did

The companies seeing real AI ROI share three characteristics Kande identified: clean data infrastructure built before AI deployment, business processes redesigned around AI capabilities (not bolted on top of existing workflows), and governance frameworks that measure actual outcomes rather than adoption metrics.

None of these are glamorous. None make good earnings call soundbites. But they separate the economics of AI tooling that delivers from the kind that bleeds budgets quarterly.

The question is not whether your company should invest in AI. It is whether your company has earned the right to. Without the foundations, you are not investing. You are donating.

---

Related Reading:

• AI Layoffs Secret: They're Firing on Potential, Not Proof
• Deep Work Is Failing 68% of Knowledge Workers