Analysis
Workers' share of total US income fell to 52.8% of GDP in the latest data, the lowest level since the government began tracking the measure in 1947, while corporate profit margins hit a record 14.9% of GDP, Fortune reported, citing EY-Parthenon chief economist Gregory Daco.
The underlying growth numbers explain why the split is widening rather than narrowing. Second-quarter GDP growth came in at just 1.7%, with only 0.3 percentage points of that attributable to additional hours worked -- meaning nearly all of the quarter's growth came from higher output per worker rather than more workers or more hours, and that higher output isn't showing up in paychecks. Wage compensation rose 2.6% nominally, which after inflation amounted to flat-to-slightly-contracting real wages. Meanwhile, AI server imports have exploded to a $450 billion annualized pace, up from roughly $50 billion annually through 2023 -- a ninefold increase in capital spending on AI infrastructure happening at the same time labor's share of the economic pie shrinks to its lowest recorded level.
“Wage compensation rose 2.6% nominally, which after inflation amounted to flat-to-slightly-contracting real wages.”
Daco's framing is blunt: "Productivity growth protects margins, not income," and he specifically warned there may be "no floor" to labor's decline, adding that "as long as you continue to see concentrated gains on the capital side...labor's share could keep plummeting." Notably, Daco's own data shows this concentration trend predates the AI boom itself -- driven by automation, cost discipline and capital spending broadly -- and AI has so far delivered "further concentration" on top of an already-declining trend rather than starting a new one, creating what Fortune's reporting calls a "winner-takes-all" environment where large firms capture the gains while smaller companies face persistent margin pressure.
Historically, prior technological revolutions -- electrification, computing, the internet -- eventually spread their productivity benefits economy-wide, even if the early years concentrated gains narrowly. Daco's caveat is explicit: there's "no guarantee that AI follows the same timetable," and the AI server import curve, still accelerating nine years into its steepest slope, offers no evidence yet that the concentration phase is ending.