Apollo’s Torsten Slok says AI still hasn’t paid off for most U.S. companies

Artificial intelligence has not produced a measurable financial payoff for most U.S. companies outside Big Tech, according to Apollo Global Management economist Torsten Slok. His argument, published Oct. 24 and reported by TechCrunch and Startup Fortune, is that the AI boom has so far lifted the firms selling the tools and infrastructure more than the businesses buying them.

That is a pointed challenge to the market’s dominant assumption that AI spending will quickly turn into higher productivity, stronger revenue and fatter margins across corporate America. Slok’s analysis focuses on the S&P 493 — the S&P 500 minus Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla — and concludes that there is no evidence AI is improving productivity for the broad group of non-tech companies.

His conclusion matters because the current AI investment cycle has been built around the idea that businesses will eventually use the technology to do more with less. If that does not show up in the numbers, the investment case for broad AI adoption becomes more complicated, especially for companies already absorbing the costs of software, chips, data centers and implementation.

What Slok is arguing

Slok’s core point is simple: outside the largest technology names, AI has not yet shown up in the financial statements of corporate America.

According to TechCrunch, he says AI has delivered “zero financial payoff” to companies beyond the tech sector. Startup Fortune reported the note on Oct. 24 and said the economist’s charts show flat profit margins and flat revenue per worker for the S&P 493. In his view, the absence of movement in those measures suggests that AI has not yet translated into measurable productivity gains for the rest of the market.

Slok’s framework also draws a line between the companies buying AI and the companies building the ecosystem around it. The Magnificent Seven — the largest U.S. tech stocks often viewed as the main beneficiaries and funders of the AI buildout — have seen revenue per worker and profit margins rise since the spending wave began, according to Startup Fortune. By contrast, the other 493 companies in the S&P 500 have remained essentially flat on those same measures.

Why he thinks the results are lagging

Slok’s critique is not just about software adoption. It is about what business value actually looks like.

According to TechCrunch, he argues that AI’s usefulness in business depends on judgment, not only on cheap intelligence. That is a narrower and more skeptical definition of utility than the one often used in AI marketing, where speed and scale are frequently presented as proof of impact.

He says the technology misses the parts of management that are hardest to automate: accountability, trust, culture, unwritten rules, ethical trade-offs and fairness. In his view, those are not peripheral considerations. They are the things that determine whether a tool can improve a company’s bottom line in the real world.

“Despite massive investments and widespread hype around artificial intelligence, the technology has delivered zero financial payoff to corporate America outside of tech.”

“AI may be a great calculator, but in business, intelligence is mostly judgment”

Why it matters

The significance of Slok’s note goes beyond one economist’s opinion. It cuts to the central question investors, executives and workers are asking about AI: who is actually capturing the value?

If the answer is still mostly the infrastructure builders and AI sellers, then the economics of the boom look very different from the public narrative around productivity. Firms outside Big Tech may be spending more without seeing much immediate return. That could pressure margins, slow broader adoption or force companies to justify AI projects with softer benefits that are harder to measure than revenue growth or profit expansion.

The note also arrives as capital spending on AI continues to surge. Startup Fortune highlighted Microsoft’s 2026 AI capex guidance of $190 billion and Meta’s 2026 capex plan of $135 billion to $145 billion. Microsoft reported $41 billion in latest-quarter capital spending, a 70% year-over-year increase, and Startup Fortune said the company’s implied calendar-year 2026 capex guidance reaches $175 billion after an accounting change. Goldman Sachs, meanwhile, has estimated total 2026 capital spending by U.S. hyperscalers at $800 billion.

That spending is often described as the engine behind AI’s next stage. But if the buyers of AI are not seeing a measurable boost in output or margins, the market may eventually ask whether the buildout is earning its keep.

Why the market is watching margins

Margins and revenue per worker are not abstract accounting exercises. They are among the simplest ways to test whether technology is making a company more efficient.

Startup Fortune said those measures rose for the Magnificent Seven during the AI spending wave while staying essentially flat for the other 493 S&P 500 companies. That comparison does not prove AI caused the improvement, but it does suggest that the broad productivity story remains uneven at best.

It also explains why Slok’s note has resonated with investors. If AI is becoming a cost center before it becomes a profit engine, the next phase of the story may be less about adoption headlines and more about whether those expenditures can eventually produce a measurable return.

How did we get here?

The current debate is a byproduct of the wider AI buildout that has dominated markets since the technology’s latest surge. Big Tech has acted as both the principal spender and the principal supplier of AI infrastructure, while other companies have been encouraged to adopt tools that promise productivity gains across sales, customer service, coding, finance and operations.

In public markets, that has created a split. The largest technology firms have been rewarded for their role in financing and enabling the AI economy. The rest of the market has mostly been asked to absorb the costs and wait for the payoff.

Slok’s note lands in the middle of that gap. He is not saying AI has no value. He is saying the value has not yet shown up where many executives and investors expected it most: in the income statements of companies outside the tech elite.

What the S&P 493 tells us

The S&P 493 is a useful proxy because it strips out the seven biggest tech names that have dominated the AI narrative. TechCrunch reported that Slok’s chart tracks the group after excluding Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla.

By removing those companies, Slok isolates the rest of corporate America — the manufacturers, retailers, industrial firms, healthcare companies and other businesses that have been told AI should improve efficiency. If the numbers stay flat there, the case that AI is transforming the wider economy becomes harder to prove.

What TechCrunch and Startup Fortune add

The two outlets largely reinforce the same basic conclusion, but they emphasize slightly different parts of the story.

TechCrunch focused on Slok’s thesis that AI has delivered no measurable financial payoff to most U.S. companies outside tech. It also highlighted his broader view that business success depends on judgment and organizational realities that AI cannot capture cleanly.

Startup Fortune, meanwhile, pushed further into the market mechanics. It said the note was published Oct. 24 and placed the findings alongside the widening capital-spending boom. It also underlined the difference between the Magnificent Seven and the rest of the S&P 500, while noting that companies outside the top seven are mentioning AI more often on earnings calls without seeing those mentions turn into stronger financial results.

That distinction matters. A company can sound bullish about AI on an earnings call without proving that the technology is helping profits. Startup Fortune’s framing suggests that narrative enthusiasm and financial results are not yet the same thing.

What does this mean for companies?

For executives, Slok’s analysis is a warning against assuming that AI adoption automatically improves performance.

Companies may still need to spend on AI to remain competitive, but the note implies that the return on those investments should not be taken for granted. If the gains are delayed, uneven or confined to certain functions, then AI becomes harder to justify as a broad-based efficiency lever.

Where the pressure could show up

• Margins: If AI spending continues before measurable gains arrive, operating margins may face pressure.
• Capital budgets: Large outlays on data centers, chips and related infrastructure may be harder to defend.
• Investor expectations: Boards and shareholders may demand clearer evidence of payoff.
• Adoption strategy: Companies may shift from broad AI experimentation to narrower, higher-return use cases.

Capital expenditures are the best public clue to how serious companies are about AI infrastructure. But capex is not the same thing as productivity. Spending can rise long before results appear, and in a weak-return environment, that gap can become a problem.

The bond-market stress around AI names

Startup Fortune added another layer to the story by pointing to stress in AI infrastructure financing, especially around CoreWeave. The company’s five-year credit default swap spread hit a record 1,194.5 basis points on Oct. 15, which Startup Fortune said implied a 63% default probability.

The same report said CoreWeave shares fell more than 15% on Oct. 14 amid a bond-market selloff. It also said the company ended June 30, 2026 with about $35 billion of debt and $16 billion of debt-like lease liabilities, and it briefly sold $1.25 billion of bonds due in 2032 at a 15% yield.

Those figures do not prove that the AI industry is in trouble. But they do show that investors are paying much more attention to the financial structure behind the boom. When debt costs rise and credit protection becomes more expensive, the market is saying that risk is no longer theoretical.

Startup Fortune described the broader situation as an AI debt binge of $466 billion. Taken together with Slok’s note, that backdrop raises a simple question: if the spending keeps climbing while the financial payoff remains hard to detect, who eventually absorbs the bill?

Background: the AI spending wave

The recent AI buildout has been powered by a small group of enormous spenders. Microsoft and Meta are two of the clearest examples in the fact sheet, with Microsoft guiding to $190 billion in 2026 AI capex and Meta outlining a $135 billion to $145 billion capex plan for the same year.

Microsoft’s latest-quarter capex of $41 billion, up 70% year over year, shows how fast those outlays have escalated. Startup Fortune said Microsoft’s implied calendar-year 2026 capex guidance would be $175 billion after an accounting change, reinforcing the scale of the company’s commitment.

Goldman Sachs’ $800 billion estimate for 2026 capital spending by U.S. hyperscalers gives the spending wave a macro backdrop. That kind of capital intensity has obvious winners: chipmakers, cloud providers, data-center builders and other infrastructure suppliers.

What Slok is questioning is the second half of the equation — whether the companies buying access to AI are getting enough in return to justify the expense.

Why the timing matters now

The timing of the note is important because it lands after a long stretch of optimism about AI’s economic effects. Markets have already priced in a great deal of future benefit. If the measurable payoff does not arrive soon, valuation assumptions may need to be revisited, especially outside the largest tech names.

At the same time, the fact sheet suggests the market itself is beginning to test the durability of the AI story through credit spreads, bond yields and equity volatility. That is often what happens when a technology narrative moves from excitement to accountability: investors start asking not just what can be built, but what can be monetized.

What happens next?

The next phase will likely hinge on whether company-by-company data starts to show AI-driven gains outside Big Tech. That is the central open question in the fact sheet, and it is the one Slok’s note puts under the brightest light.

There are a few specific things to watch:

1. Earnings calls: Whether more companies keep mentioning AI without a corresponding improvement in results.
2. Margin trends: Whether the S&P 493 eventually shows better profit margins or revenue per worker.
3. Capex discipline: Whether firms continue to fund AI spending at current levels if returns remain elusive.
4. Credit markets: Whether stress around AI infrastructure names broadens beyond CoreWeave.

There is also a broader strategic question. Even if AI eventually creates value for non-tech companies, the lag could be long enough to matter. In that case, the early gains may continue to accrue primarily to the firms building the tools and the infrastructure, while the buyers wait for the payoff that justifies the expense.

For now, that is the market’s central tension: a technology with enormous hype, enormous spending and, at least in the data Slok cited, no measurable bottom-line benefit for most of corporate America outside Big Tech.

By the numbers

Slok’s conclusion may not be the final word on AI’s economic impact. But it does sharpen the debate at the exact moment when the costs are visible, the hype remains high and the benefits outside Big Tech are still difficult to prove.