AI Was Supposed to Replace Workers. So Why Are Companies Spending Billions on Them?

The article argues that AI leaders are investing heavily in the workers many expected to be replaced, suggesting the AI boom may enter a new phase. It cites a free presentation by former Pentagon advisor Jim Rickards but provides no specific financial metrics, spending figures, or market-moving developments. Overall, the piece is more interpretive than actionable for near-term trading impact.

Analysis

The signal here is less about substitution and more about an arms race in scarce human capital. When the leaders are still paying up for talent, it implies the AI cycle is earlier than many discount, which supports a longer runway for compute demand and cloud spend; semis and hyperscaler infrastructure remain the cleaner beneficiaries over the next 6-18 months.

The second-order risk is that the market keeps treating AI as an operating-margin miracle, while the near-term reality may be higher opex, higher SBC, and slower operating leverage. That creates a vulnerability for high-multiple software and any name whose bull case depends on fast AI monetization rather than on actual usage growth; the first place this shows up is usually next-quarter guidance, not the headline release.

Contrarian take: consensus is probably overfocused on AI displacing workers and underappreciating that the buildout still requires a lot of expensive labor, integration, and oversight. That makes services/integration and infrastructure exposure more durable than pure application-layer exposure, but the article itself is promotional and not a verified fundamental catalyst, so this is better treated as an alert than a standalone trade.

Key Decisions for Investors

• No immediate standalone trade on this item; wait for verified evidence in earnings/10-Qs that AI-related compensation and opex are materially changing before leaning short software margins.
• Prefer long SMH on pullbacks versus short IGV over the next 1-3 months; the setup is that compute demand stays elevated while software monetization remains slower than the market expects.
• Hold or add to MSFT/GOOGL/META only on weakness, not on the headline; falsify the bullish thesis if the next print shows R&D/SBC rising faster than revenue by more than ~2 percentage points.
• Set an earnings alert for MSFT, META, and GOOGL: if management raises opex guidance without a matching lift in revenue expectations, consider a short in a rich software basket.
• Avoid chasing high-beta AI-app names until there is clearer evidence of revenue conversion; if the market keeps rewarding infrastructure over apps, relative performance should favor XLK/SMH over IGV for the next 3-6 months.