95% of Companies Are Seeing Little to No ROI From AI - And the Gap Between Spending and Returns Is Widening
The AI investment boom has a return problem. While tech giants are committing hundreds of billions to AI infrastructure, the companies deploying that infrastructure are increasingly struggling to justify the cost. Multiple research reports converging on the same conclusion: the gap between AI spending and AI returns has become one of the most consequential issues in enterprise technology.
Despite tens of billions of dollars poured into AI, as many as 95% of companies are seeing little to no return on investment, highlighting a growing disconnect between expectations and real-world outcomes. While executives tout productivity gains, many employees report minimal time savings, and a large majority of organizations still lack meaningful, value-driving AI use cases. Yahoo Finance
Microsoft canceled most of its Claude Code licenses, in part over costs. Uber's COO said AI costs are getting "harder to justify." One AI consultant told Axios that a client recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees. Yahoo Finance
The Four Problems Blocking Enterprise AI ROI
The ROI gap is not random - it has identifiable causes. Corporate AI adoption is running into four specific problems. First, use cases: most people default to automating tasks they dislike rather than tasks most valuable to the company. Second, costs: enterprise AI plans are not truly "all you can eat," and even simple chatbot queries carry heavy token costs - one CTO told Axios that employees were using AI models to check the weather. Third, humans remain the bottleneck to more efficient adoption. Fourth, data: when enterprises are hesitant to give AI agents unfettered access to proprietary data, those agents become less effective. Yahoo Finance
The data problem is particularly structural. AI agents that cannot access the data they need to perform their tasks produce inferior outputs - but the data that would make them most useful is often the data organizations are most reluctant to expose to AI systems. That tension has no easy resolution.
Who Is Winning and Why
The ROI picture is not uniformly bleak - it is deeply uneven. One of the clearest trends is the uneven distribution of profits across the AI ecosystem. Semiconductor companies have captured the bulk of financial gains, while enterprises, model developers, and application companies continue to struggle with monetization. Yahoo Finance
Within the enterprise, a McKinsey survey of 1,993 respondents from 105 countries found that a majority reported either cost benefits or revenue gains from using AI. The most commonly reported savings were in software engineering, manufacturing, and IT, while revenue gains were commonly reported in sales and marketing, strategy and corporate finance, and product and service development. Yet despite the benefits companies have generated from AI, only a minority reports that the benefits outweigh the investment. Fortune
The companies breaking through share a common characteristic. Nicolai von Bismarck, a McKinsey partner leading its service operations practice, said: "Many of the companies that are breaking through share a common characteristic: They treat AI transformation the way they would treat any comprehensive operating-model transformation, with strategic discipline, executive accountability, and a clear theory of how value gets captured."
The Spending Side Is Getting Worse
While enterprise ROI struggles, hyperscaler spending is accelerating. Goldman Sachs strategist Ben Snider estimates megacap US hyperscalers will allocate $755 billion to capital expenditures in 2026, an 83% year-over-year jump. This spending is estimated to reach 100% of cash flows from operations this year, leaving little room to return cash to shareholders. Hyperscalers cut buybacks by 64% year over year in Q1 and now allocate 20% of total spending to buybacks and dividends, compared to an average of 34% from 2017 to 2022. Yahoo Finance
The economic structure of this moment is unusual. The companies building the infrastructure are spending beyond their operating cash flows, while the companies deploying that infrastructure are mostly not getting returns. The logical resolution is either that returns materialize in the next 12-24 months, or that spending gets cut. Market volatility in June 2026 reflects investors pricing in the risk of the second scenario.
What Business Leaders Need to Do Differently
From four years advising C-level executives on AI for business strategy, I have watched this pattern repeat across every new technology cycle. The companies that eventually generate real returns share three practices that distinguish them from the 95% that don't.
First, they measure before they deploy. Baseline data on the process AI is supposed to improve - time spent, error rates, cost per transaction - exists before the AI tool goes live. Without a baseline, you cannot demonstrate improvement.
Second, they start with revenue-generating use cases, not cost-cutting ones. AI automation applied to sales pipeline management, pricing optimization, or customer retention generates returns that show up in revenue. Applied to checking the weather or summarizing internal meetings, it generates token bills that show up in IT costs.
Third, they treat implementation like a change management project, not a software deployment. The bottleneck is almost never the AI capability. It is getting humans to change how they work.
The 95% figure is alarming - but it is also an opportunity signal. The gap between what AI can do and what most organizations are extracting from it is enormous. The companies that close that gap in the next 18 months will have a durable competitive advantage over those still searching for their first meaningful use case.
Cut Through the Noise
What percentage of companies are seeing ROI from AI in 2026?
Up to 95% of companies are seeing little to no return on their AI investments, according to research cited in multiple 2026 reports. A McKinsey survey found that while a majority of companies reported some cost or revenue benefit from AI, only a minority said the benefits outweigh the investment. The gap is attributed to poor use case selection, uncontrolled token costs, data access limitations, and failure to treat AI adoption as an operational transformation.
What are the biggest reasons enterprise AI investments fail to deliver ROI?
Four problems dominate: employees using AI for low-value tasks rather than high-value ones; token-based billing that generates unexpected costs (one enterprise spent $500 million on Claude in a month without spending caps); humans remaining the bottleneck to changed workflows; and organizations refusing to give AI agents access to the proprietary data that would make them effective.
How much are tech companies spending on AI infrastructure in 2026?
US hyperscalers are projected to spend $755 billion on capital expenditures in 2026, an 83% year-over-year increase, according to Goldman Sachs. This equals approximately 100% of their operating cash flows for the year, causing a 64% year-over-year reduction in buybacks and dividends. Microsoft alone has guided to $190 billion in 2026 capex.
Which AI use cases are actually generating returns in 2026?
The use cases showing the strongest returns are software engineering (productivity gains from AI coding tools), manufacturing (process optimization), IT operations (automated incident response), sales and marketing (AI-assisted pipeline management and content), and fraud detection in financial services. AI applications in strategy, corporate finance, and product development also report strong revenue gains.
