AI Spending Hit $2.5 Trillion in 2026 — So Why Can't Most Companies Prove It's Working?

Two numbers from 2026 tell very different stories about the same industry. The first: worldwide spending on AI is forecast to total $2.52 trillion this year, a 44% increase over 2025, according to Gartner — making it the fastest-growing technology expenditure category in enterprise history. The second: fewer than one in three CFOs can point to a specific financial return from that spending. If you've sat in a budget meeting trying to justify an AI line item and come up short on hard numbers, you're in surprisingly good company — most of the industry is in the same spot. Both numbers are real, and understanding why they coexist says more about where AI actually stands right now than either one alone.

Adoption is no longer the bottleneck

By any measure, AI has moved past the early-adopter phase. According to McKinsey's State of AI survey, 88% of organizations now use AI in at least one business function, and 72% use generative AI specifically — up from just 33% in 2024, a genuinely fast shift for enterprise software.

The money backs this up. Crunchbase reported Q1 2026 global venture funding reached $300 billion, with AI companies capturing $242 billion of it — 80% of all venture dollars that quarter. OpenAI alone raised $122 billion, followed by Anthropic at $30 billion and xAI at $20 billion. The broader AI market, roughly $390.9 billion in 2025, is on track for around $539.5 billion in 2026, per Grand View Research — estimates vary by firm, but not the direction.

None of this reads like a slowdown.

The part that hasn't caught up

Despite that adoption curve, McKinsey's data shows nearly two-thirds of organizations haven't actually begun scaling AI across the enterprise — most deployments are still pilots. Only 23% have adopted AI agents at scale, despite far more having experimented. And only 39% report any EBIT impact they can attribute to AI; roughly 6% qualify as "AI high performers" attributing more than 5% of EBIT to it.

Gartner's own research points the same way: in a survey of 353 data, analytics, and AI leaders from late 2025, only 39% were confident their current AI investments would positively affect financial performance — record spending on something less than half are confident will pay off.

Forrester has been most direct about naming this a correction. In its 2026 predictions, it forecasts enterprises will defer 25% of planned AI spending into 2027, driven by a widening gap between vendor promises and delivered value. Fewer than one in three CFOs can identify a specific financial return from AI, and only 15% of AI decision-makers have seen an EBITDA increase from it over the past year.

Why both things are true at once

The simplest explanation is that "adopting AI" and "getting value from AI" turned out to be very different projects on very different timelines. Turning on a generative AI tool for a team is fast and cheap. Rewiring a workflow or decision process around it — the part that shows up in EBIT — is slower, more expensive, and easier to get wrong. Gartner found organizations with successful AI initiatives invest up to four times more, as a share of revenue, in the unglamorous foundational work: data quality, governance, change management. The spending headlines are about the fast, cheap part; the ROI numbers are about the slow, expensive part most organizations haven't reached yet.

This also explains why the money hasn't slowed even as skepticism grows. Deferring 25% of planned spend to 2027 isn't cutting it — it's CFOs asking for a business case before the next round, rather than approving by default like many did in 2024-2025. The shift is from technology-led AI investment to finance-led: the money still comes, but it now has to clear a bar.

What this means if you're evaluating AI spending yourself

For anyone deciding how much to invest in AI tools — a large enterprise budget or a small team's subscriptions — the practical lesson isn't "AI doesn't work." It's that the tool itself rarely determines whether you see a return. The "high performer" minority treated adoption as the starting line, not the finish line, and invested in the unglamorous work of integrating AI into how decisions actually get made. Buying the tool is the easy 2026 story; making it pay off is still the harder one.