METAL LAB

EY Chief Economist Warns AI Could Create a Winner-Take-All Economy

Gregory Daco argues that AI-driven productivity gains protect corporate margins, not workers' incomes.

EY Chief Economist Warns AI Could Create a Winner-Take-All Economy

Image: Fortune (screenshot)

Summary

  • Gregory Daco, chief economist at EY-Parthenon, warned that AI-driven productivity gains protect corporate margins but not workers' incomes.
  • In the second quarter of 2026, US corporate margins hit a record 14.9% of GDP, while labor's share of income fell to 52.8%, the lowest level since 1947.
  • At the same time, OpenAI said its latest model has overtaken Anthropic, while Anthropic is finalizing a $15 billion pre-IPO credit facility.

A Warning Amid the OpenAI-Anthropic Rivalry

OpenAI has said its latest AI model has overtaken Anthropic, and Anthropic is finalizing a $15 billion pre-IPO credit facility, according to reports from the Financial Times and Bloomberg. As the two companies grow in tandem, Gregory Daco, chief economist at EY-Parthenon, warned that AI-driven productivity gains will create a winner-take-all economy that channels profits to a handful of large companies.

Productivity gains flow directly along a solid line into corporate margins, pushing them to a record high, but the path from margins to labor income is broken by a dotted line, with a gate in between where a handful of companies capture the profits. As a result, labor's share of income has fallen to its lowest level since 1947.Productivity gains flow directly along a solid line into corporate margins, pushing them to a record high, but the path from margins to labor income is broken by a dotted line, with a gate in between where a handful of companies capture the profits. As a result, labor's share of income has fallen to its lowest level since 1947.
Image: Generated by METAL AI

Daco told Fortune, "Productivity growth protects margins, not income." He noted that in past technological shifts, from the late-19th-century railroad boom to the dot-com revolution of the 1990s, large, vertically integrated companies captured the gains first, while smaller firms struggled under persistent cost pressure, policy uncertainty, and high interest rates.

In plain terms, the gains from AI-driven productivity flow into corporate margins before they reach workers' wages. Companies that are already large, like OpenAI and Anthropic, are better positioned to capture capital faster in this dynamic.

Margins at a Record High, Labor's Share at a Record Low

The numbers behind Daco's warning are stark. In the second quarter of 2026, US economic output grew 1.7%, helped by a 0.3% increase in hours worked, and nominal compensation rose 2.6%. But once oil-driven inflation from spring through summer is factored in, Daco added, real compensation was essentially flat to slightly down.

Corporate margins, by contrast, hit a record 14.9% of GDP, and labor's share of income fell to 52.8%, the lowest level since the government began tracking the data in 1947. Daco said 50% is not a floor, and that labor's share could fall further.

IndicatorValue
Corporate margins as % of GDP14.9% (record high)
Labor's share of income52.8% (lowest since 1947)
Q2 2026 economic growth1.7%
Nominal compensation growth2.6%

The Same Signal Shows Up in Bond Markets and Jobs Data

JP Morgan's Kriti Gupta and Nick Roberts, in a note shared with Fortune, said the recent rise in bond yields is no longer primarily about Federal Reserve policy uncertainty. They wrote that it reflects a combination of concerns over global fiscal deficits, rising bond issuance by hyperscalers, and higher refined fuel prices. They added that with US refineries running near maximum capacity, refined fuel prices aren't coming down easily even though crude oil supply itself remains ample, and that this could feed through to consumer-facing prices like airfares.

US jobs data due out the same day drew unusually close attention. July's report had come in much weaker than expected, with nonfarm payrolls falling unexpectedly and the prior two months' figures revised down, leading Deutsche Bank's Jim Reid to say doubts had grown over whether the Federal Reserve would raise rates at its September FOMC meeting. Deutsche Bank economists expected nonfarm payrolls to add 65,000 jobs this time, with unemployment holding at 4.1%.

PIMCO's Tiffany Wilding and Lotfi Karoui, in an insights piece for the firm, pointed out that 90-day delinquency rates on subprime auto loans have risen sharply in recent years while delinquency on prime loans has stayed stable. They added that it remains to be seen whether this subprime weakness is a leading indicator of broader economic stress, or whether it would take an external shock — such as a labor market disruption or a sudden end to the AI capex cycle — to turn it into something more serious.

Editor's Take

What makes Daco's warning notable is the timing. It landed at the exact moment OpenAI announced its latest model had overtaken a rival and Anthropic was finalizing a $15 billion pre-IPO credit facility — alongside news that labor's share of income had dropped to its lowest level on record. That's not a coincidental overlap; it's two sides of the same trend. Companies with the capital to pour into AI infrastructure are the ones capturing the gains from productivity growth first, and that money cycles back into bigger models and bigger credit facilities.

A survey we covered last month found that one in five US workers is already delegating tasks to AI instead of colleagues, and Daco's analysis shows the next chapter of that trend: the pace at which work shifts to AI is running far ahead of the pace at which those gains flow back into wages. In practical terms, Korean companies can fall into the same trap. If a company announces labor-cost savings from AI adoption without examining how much of the resulting margin improvement actually gets reinvested or passed on as wages, it ends up reproducing exactly the winner-take-all pattern Daco describes, in Korea too.

With both bond markets and jobs data now shaky, the market is likely to react sharply to every single jobs number between now and the September FOMC meeting. If today's jobs report comes in close to Deutsche Bank's estimate of 65,000 added jobs, the Fed would have grounds to focus on inflation instead — but the decline in labor's share of income that Daco flagged isn't something one or two rate moves can reverse. That gap is likely to widen further over the coming quarters.

Comments