Employees using AI are working faster, but the economy isn’t more efficient. A look at what happened in the pre-Internet era might explain why

The article argues AI may be creating a lag between rising labor productivity and weak measured total factor productivity, echoing the 1990s internet/productivity paradox. Evidence cited includes time savings from AI users, but also delayed output realization, burnout risk, and limited economy-wide impact in current data. Overall, the piece is cautiously constructive on AI’s long-term productivity potential but neutral in the near term.

Analysis

The market is likely underpricing the lag between AI capex and measurable economy-wide output. That gap matters because listed beneficiaries will likely monetize the investment cycle first, while broad productivity gains may not show up in macro data for several quarters or even years; in the meantime, multiples can expand on narrative alone. The key second-order effect is labor substitution at the margin without immediate headline layoffs: firms can freeze headcount, stretch existing staff, and defend margins, which should be most visible in software, consulting, back-office automation, and IT services before it reaches the rest of the market.

The biggest loser is not labor broadly, but low-differentiation service providers whose pricing power depends on billable hours or manual throughput. If AI is truly acting as capital deepening, firms that sell workflow automation, cloud inference, data infrastructure, and enterprise integration should capture the spend, while labor-intensive middlemen face margin compression and slower revenue-per-employee growth. Watch for a divergence between revenue growth and headcount growth as the market starts rewarding firms that can show operating leverage from AI adoption, even if aggregate productivity statistics stay muddled.

The contrarian risk is that current AI spend proves closer to the 1990s infrastructure buildout than to an immediate earnings step-up: capex can stay elevated while realized productivity remains deferred, creating a temporary profit squeeze for adopters. That argues for caution on the broad software basket if valuations already discount near-term efficiency gains. The catalyst to reverse the current complacency is a sequence of 1-2 quarters where companies report higher margins and flatter hiring without a corresponding jump in revenue, which would validate AI as a margin tool before it becomes a GDP story.

The cleanest expression is to own the infrastructure beneficiaries and hedge the adoption losers. The trade should work best over 6-12 months as enterprise budgets roll into fiscal planning cycles and as management teams guide to labor-light growth; the main risk is AI ROI disappointment or a macro slowdown that delays conversion of capex into profits.

Key Decisions for Investors

• Long SMH vs short XLY or IYF over 6-12 months: AI infrastructure beneficiaries should capture the first wave of spend while service-heavy sectors face margin compression; target 8-12% relative outperformance if capex conversion holds.
• Long MSFT / NVDA / AVGO on 3-6 month pullbacks: these names monetize the adoption cycle before macro productivity data turns; favor entries after earnings or broader tech de-ratings, with a 2:1 risk/reward into the next budget cycle.
• Short selected IT services / BPO names (e.g., ACN, CTSH on strength) for 3-9 months: if AI reduces billable labor demand, revenue growth may remain okay while wage leverage and utilization pressure margins; cover if management shows net-new AI-driven bookings acceleration.
• Pair long QQQ / short equal-weight software ETF (if available) for 6 months: use it as a hedge against the risk that software pricing power lags infrastructure monetization; thesis breaks if software vendors start posting clear AI-driven ARPU expansion.
• Avoid extrapolating broad productivity gains into cyclicals yet; wait for two consecutive quarters of margin improvement without headcount growth before adding industrial or consumer discretionary exposure tied to a stronger labor-income backdrop.