The health systems and health plans that have pushed AI furthest into their operations are among the least equipped to prove what it is returning, according to a survey of 64 U.S. healthcare CFOs and finance leaders published Aug. 12 by Deloitte.
Only 18% of organizations Deloitte classified as “AI scalers” said they have mature financial attribution capabilities, meaning they consistently measure AI’s effect on revenue growth or cost savings with defined baselines and clear ownership of key performance indicators. Among “AI starters,” 31% reported that level of maturity.
Editor’s note: The survey was conducted in spring 2026 among 32 finance leaders at health systems with more than $1 billion in revenue and 32 at health plans with more than 500,000 members. Forty-four percent qualified as AI scalers, defined as organizations where more than 33% of generative AI initiatives launched over the past two years reached scaled deployment across multiple functions. Deloitte said the labels were built for analytical comparison and do not represent a formal maturity model.
Eight things to know:
1. Investment intent is highest among the organizations with the thinnest measurement discipline. Seventy-five percent of AI scalers said they plan to increase spending on generative and agentic AI over the next 12 months, compared with 67% of AI starters.
2. Scalers are also more confident in payback. Eighty-five percent expect AI investments to break even in less than five years from the time of investment, compared with 72% of starters. Among scalers, 64% said AI payback is comparable to or faster than that of other enterprise technologies such as EHRs or revenue cycle tools.
3. Return expectations follow the same pattern. Sixty-four percent of scalers estimate annualized cost savings of 5% or more from AI within one to two years, compared with 45% of starters. On the revenue side, 75% of scalers expect annualized growth of 2% or more, compared with 63% of starters.
4. Most organizations sit in a middle tier of measurement. Sixty-eight percent of scalers said they rely primarily on before-and-after comparisons for some initiatives, compared with 56% of starters. Deloitte attributed the gap to the complexity of wider deployment: multiple teams shape the same outcome, baselines shift and benefits show up indirectly or over longer time horizons.
5. AI’s cost structure is a second complication. Consumption is increasingly metered in tokens rather than licenses or seats, making a meaningful share of AI spending a variable input cost rather than fixed overhead. Deloitte said a use case can improve cycle time or productivity and still fail financially if prompts, retrieval, outputs, orchestration steps or agent-to-agent activity push costs up faster than benefits, a risk concentrated in high-volume workflows such as revenue cycle, contact centers, prior authorization, clinical documentation and member services. Many organizations still lack the cost visibility and allocation mechanisms to track token consumption by use case, workflow, team or outcome.
6. Deloitte recommends each AI use case carry a “value ledger” documenting the value driver, the baseline, the owner of the relevant KPIs, a consumption metric and a review cadence. It also suggests finance separate AI’s impact into what is directly attributable, what is reasonably contributory and what is not yet monetized, so leaders can recognize capacity gains without overclaiming savings. A prior authorization agent, for example, could be tracked on nurse review minutes avoided, approval turnaround time, avoidable-denial reduction and cost per authorization.
7. Public messaging mirrors the internal gap. In a separate analysis of 17,622 newsroom and press release articles published by 62 health systems and health plans between January 2023 and May 2026, Deloitte reviewed 408 technology-focused pieces and found that narratives citing demonstrated value rose from 9% in 2023 to 19% in 2026. Expected value remained the dominant frame throughout, accounting for roughly 35% to 45% of coverage. When organizations did cite proven results, they more often pointed to operational or clinical improvement than to cost savings, revenue growth or margin gains.
8. The findings come as CFOs weigh AI against competing capital demands and prepare for Medicaid funding cuts that begin phasing in after 2026, a stretch in which every dollar of technology spending is likely to draw sharper board scrutiny. As Deloitte put it: “Momentum may justify initial investment. Sustained investment will involve evidence.”
Click here to access the Deloitte report.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
