Only 6% of Companies Get Real Value from AI. Here Is How to Fix That

Quick Facts

• Only 37% of AI initiatives deliver expected business value, and just 6% of companies qualify as AI high performers, according to McKinsey's 2026 State of AI survey.
• Cost overruns are the top brake on AI projects, stalling three times as many initiatives as compute shortages, per VentureBeat's October 2026 Pulse survey.
• Organizations with full visibility into AI costs are five times as likely to have established ROI, KPMG found in its Q2 2026 AI Pulse of 2,145 leaders.

Enterprise AI spending is rising fast, but returns are not keeping pace. A new framework published by VentureBeat defines AI economics as the relationship between total cost of ownership and the business value those investments produce, giving technology and finance leaders a consistent method to evaluate what to start, stop, optimize, or scale.

The core test is straightforward: Is the business value real and measured against a baseline? Is the cost per outcome lower than the previous method? If an initiative cannot answer both questions, it should not be scaled.

The ROI Gap Is Wide

The numbers show how far most companies are from capturing returns. McKinsey's 2026 State of AI survey found that 80% of respondents say AI improved their individual productivity. Only 37% reported a positive EBIT impact. Just 6% are classified as high performers.

Adoption is spreading, though. Nearly nine in ten respondents say they use AI regularly in at least one business function. The share of companies scaling AI enterprise-wide rose from 38% to 44% in a year. For agentic AI specifically, large enterprises scaling agents jumped from 27% to 40% over the same period.

A PwC January 2026 survey found that 56% of CEOs reported neither revenue growth nor cost reduction from AI over the previous 12 months. Only 12% achieved both outcomes, a group PwC calls the vanguard, distinguished by stronger AI foundations, defined roadmaps, and enterprise-scale deployment.

Cost Visibility Is the Missing Piece

The measurement problem starts before the ROI calculation. Only 7% of the 2,145 leaders in KPMG's mid-2026 AI Pulse said they had established ROI on AI. Only 26% judge AI spending by a measured financial return they could defend to their finance team, yet 79% expect their AI budgets to increase.

KPMG also found that 86% of organizations with established ROI operated a formal cross-functional or enterprise-wide AI management layer. The correlation between governance and results is direct.

Cost Overruns Are Stopping Projects Cold

VentureBeat's October 2026 Pulse report found that 38% of respondents said cost overruns had stalled an AI project. That is three times the number affected by compute shortages. Nearly half of organizations in Futurum's second-half 2026 survey reported running over budget on AI, and one-quarter delayed or cancelled initiatives due to unforeseen costs.

Memory, GPU, and software prices are climbing alongside the broader AI buildout. Infrastructure costs are rising across the board, compressing margins for companies that have not built unit economics into their AI planning.

The Inference Paradox

Gartner analyst Will Sommer identified a structural cost problem building inside AI deployments. "Product leaders cannot rely on more efficient token economics to rationalize AI costs," Sommer told VentureBeat. "Each successive generation of AI capability will necessitate more, and often more expensive, tokens."

Gartner calls this the Inference Paradox: cheaper models make more capable applications affordable, but those applications demand more inference to operate. AI agents require 5 to 30 times more tokens per task than basic AI chatbots. Worldwide spending on AI inference is forecast to reach $23.3 billion in 2026, already surpassing the $19 billion expected for AI training. Gartner projects that inference costs per agentic workflow will increase more than fivefold through 2028.

What Separates Scalers from Starters

IBM research found that only 25% of AI initiatives deliver expected ROI and just 16% have scaled enterprise-wide. The companies that do scale share common traits: they tie AI spending to measurable business outcomes, they have cross-functional oversight, and they evaluate cost per outcome before committing to expansion.

BearingPoint's October 2026 data showed that nearly three-quarters of organizations implementing AI reported measurable impact, but only 13% scaled AI as planned. The gap between impact and scale is a resource allocation problem as much as a technology problem.

The conclusion the framework draws is direct: the leaders who win the next phase of enterprise AI will not be those who spend the most. They will be those who know enough about the economics of each investment to decide where to spend next.