Most of the 30 companies in the Dow Jones Industrial Average now say they use artificial intelligence. Far fewer are willing to say what it costs them or what they get back, according to a new analysis from Flexera.
The enterprise software company reviewed annual filings from the 30 Dow companies as of August 2026. The findings were published on Wednesday, Nov. 4.
Risks Get More Space Than Returns
More than three-quarters of the companies reported actively using AI and named specific ways they use it. Only two disclosed clearly identifiable amounts they had invested in AI.
Those same two companies were the only ones to report a measured business result tied to AI. No other Dow company put a number on what AI delivered in its annual filing.
Companies were far more willing to talk about AI risk. About 9 in 10 of the Dow companies listed company-specific AI risks in their filings, compared with the two that outlined AI outcomes.
“The challenge is that there is no single, widely accepted formula for AI economics yet,” said Conal Gallagher, CIO and CISO at Flexera. He said developers, analysts and salespeople may all use AI, but the costs and expected results are different for each.
Gallagher defined measurable value as a company’s ability to link spending to a specific business outcome. That could mean higher productivity, lower operating costs, faster delivery or revenue growth.
He said companies need a clear view of their total AI spending before they can calculate a return. That is hard because AI costs can show up across software licenses, cloud services, data platforms, tokens and AI agents, often under different budgets.
Other Surveys Point to the Same Gap
Other research shows a similar pattern. A survey of more than 1,500 organizations by software company Aptean found that 98% had adopted AI, but less than half said it was essential to their core work.
An SAP survey of 2,600 director and C-suite executives found that most companies are struggling to find and define AI returns. The survey covered midmarket and enterprise organizations.
An August McKinsey survey of more than 1,700 executives found that 37% linked any measurable earnings to AI. At the same time, 80% said AI had improved their productivity.
“A striking finding this year is the gap between individual gains and enterprise impact,” McKinsey Senior Partner Dan Tinkoff wrote in the report.
Gallagher said companies may be cautious about public reporting because measurement methods are still developing. He noted that every AI prompt, response and API call can create cost, while benefits such as time saved are hard to tie directly to AI.
He recommended measuring value one use case at a time instead of seeking a single company-wide figure. That means naming a business owner, setting the full cost, defining the outcome and tracking results over time.
Gallagher also said AI spending can become concentrated among a few users without finance teams noticing until the bill arrives. One company working with Flexera found that just four users were responsible for 35% of its AI costs.
Gallagher said he expects companies to face pressure to disclose AI spending in the near term. He said CFOs, CIOs and investors are asking for clearer evidence of AI’s financial impact.
