The Great AI Payoff Illusion

Three years ago, CEOs across every industry repeated the same refrain: artificial intelligence is the technology that will transform every sector of the economy. Billions of dollars were committed to investments, thousands of pilot projects were launched, and consulting firms were hired to guide the digital transformation. Now, in October 2026, the most rigorous analysis available shows a different outcome than promised: AI has not delivered any measurable financial gain to companies outside the technology sector.

Torsten Slok, chief economist at Apollo Global Management, published this weekend a detailed analysis based on three fundamental metrics — revenue per worker, profit margins, and total factor productivity growth — and the conclusion is unambiguous. For the 493 companies that make up the S&P 500 outside the Magnificent Seven, all indicators remain flat since before the AI investment boom.

The numbers that matter

Slok's central argument rests on three charts that cut through any ambiguity. The first measures real revenue per worker adjusted for inflation (base Q4 2019). For the Mag7 group, the indicator jumped from 100 to 125 between the last quarter of 2019 and the second quarter of 2026. For the rest of the S&P 500 — the so-called S&P 493 — the index remained at 100.7 in mid-2026. Essentially zero movement over six years.

The second chart measures profit margins. The Mag7 saw their margins rise from around 20% to over 30% in the same period. The S&P 493 kept margins essentially flat at around 12%. The third chart measures total factor productivity growth — the metric economists consider the true measure of innovation and efficiency. For the Mag7, productivity grew 1.8% annually since 2022. For the rest of the American corporate economy, growth is zero.

The capital cost problem

The Hudson Hill Capital analysis, published in September, deepens the problem by introducing capital cost into the equation. The report notes that the five largest technology companies (Microsoft, Alphabet, Amazon, Meta, and Oracle) directed between $810 and $840 billion in capital expenditure alone in 2026 — a jump from $408 billion in 2025 and $87 billion in 2019. If you measure productivity by labor alone, the indicator improves. But when you add capital cost and energy to the denominator — which any financial analyst would do — the gain disappears.

The Federal Reserve Bank of San Francisco adjusted for capital utilization and found that total factor productivity rose only 0.3% in 2025 and fell 0.3% over the four quarters to Q2 2026, while capital grew 2.8% annually. In other words: AI is consuming far more capital than it is generating in additional productivity. The result is a net decline in productivity when you count everything.

Token spending as cost center

Traditional companies' reaction to the data has been swift. "Mainstream corporate America is pulling back and starting to view token spending not as a source of growth but as a cost center that weighs on margins," Slok said in the report. Companies like SAP have already announced AI spending freezes. The message is clear: after spending millions on pilots and PoCs that never made it to production, CIOs are realizing the ROI is negative.

Conor Twomey, CEO of AI One, who has spoken with roughly 500 large companies in 2026, points out that most are still stuck in the "chatbot era" — the notion that ChatGPT represents the ceiling of what AI can do, when in reality it's the floor. A study he cites shows a successful chatbot implementation saves roughly 26 minutes per employee per week — and that benefit accrues to the employee, not the business. Real value begins when AI does meaningful work, such as AI-assisted software development, and even so the first signs of returns are concentrated among software developers, not the broader economy.

The telecommunications bubble parallel

Hudson Hill Capital draws a troubling comparison with the telecommunications bubble of 1996-2002: over $500 billion invested, only 2.6% of fiber activated, and bondholders recovered about 20 cents on the dollar. The analogy is that AI suppliers (Nvidia, Dell, AMD, TSMC) are living in supplier nirvana while AI buyers (Mag7) are financing with capital that hasn't yet hit the income statement — but that is changing.

As depreciation rises (from the five largest companies, it went from $88 billion in 2023 to $171 billion over the latest twelve months), costs will begin to pressure results. Goldman estimates depreciation will cut 5% from 2027 S&P 500 earnings growth. Meanwhile, technology companies are raising capital in the markets — roughly $290 billion between January and October 2026, with $221 billion in bonds and $70 billion in equity — to fund the next round of data centers.

The question is not whether AI will be transformative, but when — and what price the broader economy will pay in the transition.

Corporate America is beginning to ask the same question, and the answer it's finding is making it nervous.

Sources: Bloomberg, Apollo Global Management, Hudson Hill Capital, Business Insider

✓ Independent sources cross-checked and verified before publishing