‘I don’t think there’s a floor’: Workers’ share of America’s income is at a record low before the AI boom even begins

The article argues the AI “productivity boom” is likely deflationary, but notes labor’s share has fallen to 52.8% (lowest since 1947) while corporate margins hit a record 14.9% of GDP—suggesting gains are concentrated in capital rather than workers. Q2 output rose 1.7% with only 0.3% from more hours, while compensation (+2.6%) is described as roughly flat to slightly contracting in real terms amid prior oil-driven inflation. It also highlights how capital-intensive AI buildouts face headwinds from higher long-term rates (weak hiring, struggling housing) and underscores potential leakage via imports: annualized net imports of “large computers” (GPU servers) reached ~$450B vs ~$50B in 2023, implying zero net GDP contribution despite massive investment. Overall, the message is that AI investment may not translate into broadly distributed income, raising policy and market concerns.

Analysis

The first-order trade is not “AI equals GDP,” it’s “AI equals a larger share of output accruing to capital.” That mechanically favors the narrow set of firms that own the chips, racks, power, networking, and cloud distribution layers, while leaving labor-heavy, price-taker businesses with no offsetting wage tailwind. In market terms, this argues for continued earnings dispersion: mega-cap tech and semiconductor suppliers can still compound margins, but broad index earnings breadth may disappoint because the investment cycle is concentrated, imported, and not yet translating into economy-wide unit labor cost relief.

The bigger near-term loser is anything rate-sensitive that relies on broad-based income growth: housing, small-cap cyclicals, and domestically oriented consumer names. If long rates stay sticky, the AI capex boom crowds out mortgage activity and private fixed investment outside the data-center ecosystem, which is a second-order drag that can show up over the next 1-3 quarters before it is visible in headline GDP. If wage growth rolls over while inflation stays sticky, the policy response risk increases: more populist pressure, antitrust scrutiny, and tax/depreciation changes aimed at the AI winners, not the laggards.

The contrarian miss is that consensus is treating AI spending as an unambiguous growth engine, when the more probable near-term effect is margin concentration and labor-share compression. That’s bullish for index-level EPS but bearish for social and political durability; once the narrative shifts from “productivity” to “who captures the surplus,” valuation multiples can re-rate lower for the same winners. The falsifier is clear: if AI capex starts producing visible wage acceleration and broader credit demand within 2-4 quarters, the current concentration trade loses force; absent that, the dispersion regime likely persists.

Key Decisions for Investors

• Long SMH vs short ITB/XHB for 1-3 months: semis and AI infrastructure should keep capturing incremental spend while housing remains rate-choked; target 8-12% relative outperformance if long yields stay elevated.
• Pair long QQQ or XLK vs short IWM for the next quarter: large-cap tech owns the capital gains from AI while small caps remain most exposed to weak labor income and tighter financing; stop if Russell breadth materially improves or 10Y yields break down sharply.
• Accumulate pullbacks in NVDA/AVGO on 2-4 week horizons, but size as a momentum trade, not a macro thesis: upside persists as long as capex guides higher, but watch for any commentary on ROI scrutiny or deferred orders as the key falsifier.
• Watch/list only: DLR and EQIX into any rate rally. Data-center demand is real, but these names are rate-sensitive; they work best if Treasury yields stabilize rather than spike. If 10Y backs up another 25-50 bps, prefer semis over REITs.
• Set an alert on homebuilder earnings revisions (LEN, PHM): if management cuts backlog or incentives rise again, it confirms the second-order income squeeze and supports staying short housing proxies.