The betrayal behind the data-center backlash: AI promised to break the rules of class but is just rewarding them so far

Research cited in the article estimates AI-exposed workers are losing roughly $28 billion annually in wage growth, affecting 5.8 million workers (3.7% of the labor force), while the lowest wage quartile saw a 10.7-percentage-point relative wage-growth decline after 2023. At the same time, nonfinancial corporate profits rose $400.9 billion in Q2 2026 and margins reached 15.2% of gross value added, near postwar highs, amid pricing power rather than demonstrated AI-led productivity gains. AI-related equities are projected to contribute nearly 40% of S&P 500 earnings growth this year and next, concentrating gains among higher-income households with equity exposure. Public resistance is escalating as 75 data-center projects worth $130 billion were blocked or delayed in Q1 2026, driven by concerns over local costs and electricity bills.

Analysis

The investable transmission mechanism is not broad AI labor displacement but a widening split between asset owners and wage-dependent consumers. If payroll growth remains weak while equity gains stay concentrated, discretionary demand should soften first in lower-income cohorts, pressuring dollar retail and subprime credit before aggregate consumption data deteriorates. This favors defensively positioned consumer staples over Dollar General (DG), Dollar Tree (DLTR), and lenders with lower-FICO exposure such as Capital One (COF) over the next 1-3 quarters.

For AI beneficiaries, the more immediate risk is political and infrastructure friction rather than labor regulation. Permitting delays, interconnection constraints, and local ratepayer backlash can lengthen data-center build schedules; that shifts value from GPU vendors with expectations tied to rapid deployment toward incumbent power generation, transmission, and grid-equipment suppliers able to monetize scarcity. Vistra (VST), Constellation Energy (CEG), GE Vernova (GEV), Eaton (ETN), and Quanta Services (PWR) have more direct pricing exposure to constrained power delivery than hyperscalers, though each is increasingly consensus-owned.

APO and MS have limited direct fundamental exposure to the wage thesis. MS benefits indirectly if AI-linked equity appreciation sustains fee-based wealth assets, but that is vulnerable to a concentrated mega-cap correction; APO's relevant exposure is private-credit and infrastructure financing, where delayed data-center projects could slow deployment but extend demand for structured capital. The article's causal claim that AI is suppressing wages is not yet sufficient to underwrite a macro short: stagnant wages may reflect low labor mobility, demographics, and cyclically weaker hiring rather than technology substitution.

Contrarian view: data-center opposition is likely underpriced as a timing problem, not necessarily a demand-destruction event. A 6-18 month permitting slowdown would reduce near-term capex conversion and cloud revenue expectations, but also raises the value of already-permitted sites, contracted generation, and regulated transmission assets. The thesis fails if hyperscaler capex guidance remains intact while power-project completion dates do not slip, or if state/federal policy overrides local restrictions and accelerates interconnection reform.

Key Decisions for Investors

• Initiate a 3-6 month pair: long CEG or VST / short a basket of AI-data-center capex proxies (equal-weight ORCL and DELL). Structure at modest size because both legs retain AI beta; the thesis is that power scarcity captures economics if deployment timelines elongate. Exit if hyperscaler capex guidance rises while contracted power pricing or nuclear/merchant power forwards fail to improve.
• Overweight ETN and PWR versus NVDA for the next 6-12 months as a second-order AI infrastructure allocation. Grid equipment and transmission backlog conversion is less dependent on immediate model-demand monetization; key risk is a broad capex recession or a meaningful drop in power-load forecasts.
• Reduce exposure to lower-income discretionary and subprime-credit sensitivity: underweight DG, DLTR, and COF versus XLP or Walmart (WMT) over 1-3 quarters. Confirm with sequential deterioration in wage growth, revolving-credit delinquencies, or retailer traffic; reverse if real wage growth reaccelerates and hiring broadens.
• Keep MS as a relative financial-sector overweight versus more credit-sensitive banks, but do not add solely on this narrative. Add only if wealth-management net new assets and fee-based balances remain resilient through the next earnings print; a sharp correction in AI mega-caps and weaker advisory activity would invalidate the relative thesis.
• Set an event-driven alert around state data-center moratoria, utility commission rate cases, and quarterly hyperscaler capex guidance. A cluster of additional restrictions or explicit project-date slippage supports the CEG/VST and ETN/PWR legs; federal permitting reform or accelerated interconnection approvals would compress the scarcity premium.