Big Tech’s $690B AI Buildout Has One Test: Enterprise ROI

That is why the current buildout can be rational even before enterprise ROI is fully proven. AInvest describes the 2026 data-center expansion as happening amid supply constraints and says AI infrastructure investment is outpacing software value capture ^[7]. In other words, Big Tech is racing to control a scarce input before the end market has completely matured.

But this is not automatic validation. Building early reduces the risk of missing demand, while increasing the risk that capacity arrives before enough customers are ready to pay for it at scale.

Enterprise ROI is still the weak link

Enterprise AI adoption and enterprise AI payoff are not the same thing. McKinsey’s 2025 Global Survey found that nearly two-thirds of respondents say their organizations have not yet begun scaling AI across the enterprise; 64% say AI is enabling innovation, but only 39% report enterprise-level EBIT impact ^[27]. McKinsey also notes that organizations are beginning to redesign workflows and put senior leaders into AI governance roles as they try to capture bottom-line value ^[22].

Coverage of MIT’s GenAI Divide is more bearish. Digital Commerce 360 reported that, despite an estimated $30 billion to $40 billion in enterprise generative AI spending, 95% of organizations had not seen measurable financial return, while only 5% of integrated pilots were extracting millions in value ^[24]. That should be read as a warning signal, not proof that enterprise AI cannot work. The evidence points to a divide between scaled, integrated deployments and pilots that never reach the profit-and-loss statement.

The signals that decide whether the buildout pays off

Utilization

The central question is whether AI data centers and specialized chips remain heavily used. High utilization turns a fixed-cost buildout into sellable capacity. Weak utilization exposes overbuild and makes it harder for providers to absorb the cost of new infrastructure.

Pricing power

AI compute has to command prices that support returns. If cloud providers compete away pricing before enterprises scale usage, revenue growth may still disappoint relative to the capex burden.

Enterprise-level financial impact

Use-case wins and demos are not enough. The stronger proof point is enterprise-level financial impact, where McKinsey’s survey still shows a gap between innovation benefits and EBIT impact ^[27]. The more enterprises redesign workflows rather than bolt AI onto old processes, the stronger the case for durable AI cloud demand ^[22].

Investor tolerance

Markets are already distinguishing between AI spending stories. Fortune reported that after Alphabet, Meta, and Microsoft discussed higher AI spending, Meta fell more than 6% after hours, Microsoft was essentially flat, and Alphabet rose almost 7% ^[1]. That uneven reaction suggests investors want a credible path from capex to returns, not just a larger AI budget.

Who carries the most risk?

The most resilient capacity is capacity that can serve many paid workloads. A broad cloud platform has more possible ways to monetize AI infrastructure than a buildout tied to a narrow or still-unproven demand base.

Futurum points to the core imbalance: pure-play AI vendors led by OpenAI and Anthropic are growing rapidly, but their combined revenues remain only a fraction of the infrastructure investment being deployed on their behalf ^[2]. That does not mean the capex is doomed. It means the margin of safety depends on whether enterprise customers turn AI into sustained demand rather than isolated experimentation.

Bottom line

Big Tech’s AI infrastructure spending is sustainable for now, but only conditionally. While compute is scarce, the largest cloud providers have strategic reasons to build ^[7]. But capex estimates in the $650 billion-plus range will be judged by utilization, pricing, and enterprise ROI—not by model hype alone ^{[2][5][24][27]}.

If enterprises turn AI into recurring production workloads with measurable financial impact, the buildout looks like a long-term cloud platform shift. If most organizations remain stuck before enterprise-wide scaling, the same spending starts to look like overbuild.