Top economist says AI just hasn’t delivered on the productivity hype—and it means a ‘painful repricing’ of markets is very possible
Apollo Chief Economist Torsten Slok warns there’s reason to be concerned about AI being unable to yet generate returns on investment.
Overview
The clock is ticking on AI to deliver on its promises of transformed workplace and economic productivity, and if lags in returns on investment continue, the markets are in for a rude awakening, according to one top economist.
Torsten Slok, the influential chief economist for Apollo Global Management, argued in a recent blog post that there’s a growing gap around AI-enhanced productivity. Basically, you can only see it at tech companies, not most of the Fortune 500.
While some sectors like software and tech can easily integrate AI into their operations, Slok argued that deploying this technology is slow-going for the vast majority of the economy. It takes time and effort due to regulatory hurdles, data protection, and workflow integration, meaning structural productivity gains are slow, and returns of investment have yet to be seen. Slok said he thinks it may happen—eventually. And by that point, the stock bubble may have burst, because the market has priced in returns sooner rather than later.
“The key issue is the length of the ROI runway outside the tech sector,” Slok said. “The bottom line is that a mismatch between current earnings expectations and the actual time firms need to generate ROI on AI investments could have significant implications for many AI company valuations today.”
Slok cites Bloomberg and Macrobond data indicating that despite profit margins for the Magnificent Seven increasing from around 15% to 25% between the first quarters of 2023 and 2026, profit margins for the rest of the S&P 493 have hovered around 10%. The Bloomberg 500 Index follows the same pattern as the S&P, remaining at a steady 12% profit margin over the same period of time.
Most concerning to Slok is what happens if this gap grows as AI deployment and productivity gains continue to sputter. A seminal and controversial MIT study published last year found only 5% of companies saw a meaningful return in investment from generative AI pilot projects. The Apollo economist warned that as expected earnings, or current market pricing, continue to outpace actual earnings, markets will face a “painful repricing” that threatens to decelerate the AI boom.
“Put differently, companies will slow their AI spending if they don’t see ROI quickly,” he said.
Where is the economy seeing AI’s faltering returns on investment?
U.S. industry giants are already reckoning with hiccups to their mass automation efforts. In a visible sign of the human expertise needed to really leverage AI productivity gains, Ford hired 350 “gray beard” engineers—veterans in the industry, including former employees—to train junior staff and reprogram ineffective AI tools. The automaker has continued to deploy AI vision systems across 33 global plants, with more than 1,000 cameras performing millions of assembly line inspections, but recognized the technology was not as effective without human oversight.
“Artificial intelligence is a fantastic tool, but it’s only as good as the information you use to train it,” Charles Poon, Ford’s vice president of vehicle hardware engineering, told reporters last month. “Over prior years, we didn’t pay as much attention as we should have to the experience of our most knowledgeable engineers that have been with us through many product cycles.”
Ford follows the lead of companies like IBM, which slashed thousands of jobs last year amid increased spending on cloud services. In March, the company announced it would triple entry-level hiring in the U.S. across all business units, arguing more jobs are necessary in AI-first workplaces.
As it stands today, this human labor is far cheaper than the automation tools companies are pouring investments into, further calling into question AI’s productivity benefits in the workplace. Nvidia’s vice president of applied deep learning, Bryan Catanzaro, said earlier this year that the cost of AI still far exceeds that of human labor, an admission that coincided with an era of tokenmaxxing, where tech companies like Meta incentivized AI use through internal employee leaderboards, which led to workers using the tech just for the sake of it, all the while driving up costs.
Details
According to Slok, the race to most effectively use tokens is more of an indication of companies struggling to get their money’s worth from AI and failing to produce real workplace gains from it.
“Companies will slow their AI spending if they don’t see ROI quickly,” Slok said. “And the current focus on token optimization is an early warning that AI implementation could be a bumpier, slower road than expected.” (Slok has separately argued that AI will create more jobs, not less, as he’s become Wall Street’s main exponent of the relevance of Jevons paradox; he also thinks it will lead to a boom in small business entrepreneurship.)
Why has AI yet to deliver on its promises?
Peter Cappelli, a professor of management at the University of Pennsylvania’s Wharton School, was early to spot the issue Slok highlighted, leading a case study on Ricoh, a digital services company, that was published in the Harvard Business Review. In short, he found that people are greatly underestimating “just how much work is involved in” realizing productivity and ROI gains, as he told Fortune earlier this year.
“If you’re listening to the people who make the technology,” Cappelli said about the AI class, “they’re telling you what’s possible, and they’re not thinking about what is practical.”
The gap between the possible and practical uses of AI is driven by “AI shame,” or the pressure for these emerging technologies to be effective, particularly amid rising pressure from investors. It’s a phenomenon observed in tokenmaxxing tech companies, where leadership has mandated increased AI use, but has failed to articulate tangible use cases or goals associated with AI use.
Boston Consulting Group found in a recent study that deploying AI just for the sake of it may actually stymie the technology’s productivity gains. The consultancy’s 2026 Global AI at Work report, which surveyed about 12,000 frontline employees, found that when 42% of respondents reported eight hours of saved time per week as a result of regular AI use, most said they received little to no guidance on how to use the time they saved, and half said they weren’t using that freed up time to complete more strategic work.
“This whole tokenmaxxing thing has probably run its course, and now it’s hitting their cost base in a pretty big way,” David Martin, Global leader of BCG’s People & Organization practice, told Fortune. “A lot of companies just gave AI to everyone, regardless of position, and I think now they’ll say, ‘Well, let’s be more thoughtful about who has access, and what is the business case? And are we delivering on it, ultimately?’”
In the case of Ricoh, Cappelli said, when the company outsourced low-level administration work to process insurance claims to AI, the process required about $500,000 in outside consultant fees, as well as $200,000 per month on AI fees, ultimately resulting in costs that were three times higher than if an employee were to complete the administrative work manually. The company reduced headcount only modestly, Capelli said, from 44 to 39 employees.
Ultimately, Ricoh increased the productivity of the division three-fold, but it took time and its example underscores Slok’s concern around what AI has to offer: Productivity gains are possible, but they are not without immense initial costs in both time and money.
“So that’s the payoff,” Capelli said. “But it’s not cheap [and] it took a hell of a long time to do.”
This story was originally featured on Fortune.com
Source
Originally published at fortune.com.
