Big Tech's AI Spending Hasn't Crushed Return on Capital — Yet

The tech industry's unprecedented rush to build AI infrastructure has raised questions about whether companies can generate meaningful returns from their multibillion-dollar investments. New data suggests that for most major players, the financial impact on core efficiency metrics has been surprisingly modest.

Return on invested capital (ROIC) — a measure of how effectively companies convert spending on equipment, facilities, and acquisitions into profit — has held relatively steady for most tech giants through mid-2026, according to analysis first reported by Axios.

The scorecard

Alphabet and Amazon have actually seen ROIC increase in recent quarters despite aggressive data center buildouts. Microsoft has maintained roughly flat returns. Meta represents the clearest exception, with ROIC dropping from approximately 32% to 22% over the past year. Oracle has also experienced a modest decline.

The standout performer is Apple, which has watched its ROIC climb to 78.8% — the highest among major tech companies — after largely declining to participate in the capital expenditure arms race that has consumed its peers.

Why it matters

ROIC provides a more complete picture of AI investment efficiency than profit margins alone. While margins show the basic economics of making and selling products, they don't account for the upfront capital required to build the infrastructure. For AI data centers, those costs are enormous. This metric helps answer whether companies are actually creating value or simply spending money in a collective bet that may not pay off.

The calculation

ROIC divides after-tax operating profit by average total invested capital from balance sheet reports. It reveals how much earnings power a company generates per dollar of investment — crucial context when evaluating whether AI spending makes financial sense.

Analysts note that companies with diversified revenue streams, particularly established cloud computing businesses, have absorbed AI infrastructure costs more effectively than those without such cushions.

The caveat

These figures may understate the true capital commitment. Big Tech has accumulated more than $1.5 trillion in off-balance-sheet obligations that aren't fully reflected in standard ROIC calculations. If those commitments were factored in completely, the efficiency picture could look considerably different.

The data also raises a counterfactual question: Would these companies be performing even better financially if they weren't engaged in this collective spending spree? Apple's trajectory suggests the answer might be yes.

The analysis was originally reported by Axios, drawing on the most recent quarterly earnings cycle through Q2 2026.