Two Years In, AI Spending Still Waiting for the Productivity Receipt

The S&P 500 closed at -0.03% as of 26 June 2026, a number so microscopically negative it practically rounds to an existential shrug. Meanwhile, technology sector capital expenditure on artificial intelligence infrastructure has accelerated dramatically. The arithmetic is straightforward: companies are spending money they don't have on productivity gains they can't measure. Welcome to the corporate productivity paradox.

Two years into what executives insist on calling a gold rush, the earnings call transcripts tell a story of unwavering conviction married to increasingly creative accounting. CFOs have become philosophers, defending billions in AI capex deployments while labour productivity metrics remain stubbornly, almost insultingly flat. The gap between what's being spent and what's being shown—in actual, measurable productivity improvements—has widened into something approaching a chasm.

Look at the pattern from recent earnings season. Tech companies articulate AI integration plans with the confidence of men who have never met a pivot table. They speak of transformer models and inference optimization and enterprise automation as though repetition somehow generates return. Downstream, the actual productivity numbers tell a different story. Where are the unit economics improvements? Where is the headcount reduction justified by autonomous workflows? Where, frankly, is the money?

The tech sector isn't alone in this peculiar theatre. Across the S&P 500, the narrative has become remarkably consistent: AI spending is essential, non-negotiable, and cannot yet be meaningfully measured against operational outcomes. This is not new. This is not innovation. This is capital deployment divorced from accountability, wrapped in enough technical vocabulary to make a Stanford computer scientist blush.

CFOs have developed a playbook for these conversations. When pressed on ROI during earnings calls, they pivot to runway arguments: this is a multi-year investment, the infrastructure must precede the applications, competitive positioning demands participation regardless of current returns. None of this is inherently false. But framed together, across hundreds of earnings calls, from companies with hundreds of billions in aggregate market value, it reads less like strategy and more like collective institutional hedging.

The real tell isn't what's being said. It's what's not being said. You will not find a single earnings call transcript from the past two years where a CFO candidly acknowledges that AI capex spending has not yet produced measurable productivity gains commensurate with the deployment. You will not find executives conceding that they're spending partly out of fear, partly out of genuine uncertainty about where value actually accrues, and partly because everyone else is doing it. Instead, you get the verbal equivalent of a consulting deck: strategic imperative, transformational potential, long-term competitive advantage.

The equity markets have noticed. That -0.03% S&P performance tells you that investors have priced in the uncertainty. They're not fleeing. But they're not celebrating either. The market, in its infinite patience for executive optimism, is essentially saying: prove it when you can measure it.

Here's what makes this different from previous technology adoption cycles. In the past, when companies deployed substantial capital into new infrastructure—cloud migration, e-commerce platforms, digital transformation—there was a relatively clear operational metric to watch. Did costs drop? Did throughput increase? Did customer acquisition improve? With AI, companies are spending on infrastructure for applications that don't yet exist at scale. They're buying picks and shovels for a gold rush that might not generate gold. Or might. Nobody actually knows.

The tragedy is that artificial intelligence will almost certainly unlock genuine productivity gains. History suggests that transformative technologies eventually do. But the interim period—the one we're living through—is a graveyard of optimistic capex and flattened productivity metrics. CFOs are defending spending that cannot yet be defended through actual results. And the market, with characteristic patience, is waiting to see if the emperor's new AI actually generates any earnings to show for the wardrobe.