US health insurers question whether AI saves them real money: they spend more on administration even as they earn more

By ACCESS Newswire (Black Book Research report) · November 4, 2026.

Claims adjustment and administrative expenses across U.S. health insurers reached about $69.2 billion in the first half of 2026, up 2.7%, according to Black Book Research's "Payer, Health Plan & Managed Care Outsourcing Outlook 2027" report. Meanwhile, the sector's margin rose from around 1.8% to around 4.1%, and net income was roughly $27.7 billion.

In other words, the sector is earning more, but its administrative bill keeps growing. That contrast is the report's starting point: executives are asking whether outsourcing and AI produce recurring savings or merely shift costs, eliminate open positions or generate productivity that never turns into cash.

The 82-page report covers 18 service areas, 54 vendor profiles and 36 qualitative dimensions. It draws on a 2025-2026 program with 8,194 verified respondents. Among them, 86% cite prior authorization and utilization management as a priority (the procedures by which the insurer decides whether to cover a treatment before it is performed). 82% cite interoperability, and 77% cybersecurity and third-party oversight. It's worth remembering that these are Black Book's own data, and that it distributes the report free with registration.

According to Black Book, AI moves the equation in both directions. Some insurers outsource to free up budget to invest in AI. Others buy services that are already AI-enabled. And automation can make it cost-effective to bring back in-house tasks that were previously subcontracted, because they require less labor. The report identifies provider data, credentialing and claims administration as candidates for insourcing or hybrid models.

The regulatory calendar adds pressure. From 2027, insurers must meet expanded interoperability and prior authorization API (application programming interface) requirements, and in 2028 the requirements for electronic claims attachments arrive. As Douglas Brown, founder of Black Book, puts it, an insurer can outsource a process, but not regulatory responsibility.

**Our take.** The report's most valuable warning is not technological but accounting-related: an unfilled position or a "productivity improvement" is not money in the bank. It's the same pattern we see in other sectors, where AI is sometimes used as a justification for cost-cutting rather than as a proven cause. Here, at least, there is a call to separate the service case, the AI investment case and the financing case.

There is also a reading on employment. In the short term, administrative and back-office work at insurers is among the most exposed, both in internal workforces and in outsourced service centers, inside and outside the U.S. That automation can repatriate tasks doesn't mean the jobs come back: the same work will require fewer people. That transition will be hard and deserves planning, not slogans.

And yet there is an opportunity. If administrative spending truly falls, and prior authorization becomes faster and more transparent by regulatory mandate, the ultimate beneficiary should be the patient: less friction between a diagnosis and its treatment. That fits with our underlying thesis: AI will contribute most when it frees resources for care and not just for margins. That's the metric the sector should be held to in 2027: not how much AI is bought, but whether costs fall and access to care improves.

A final caveat: the material comes from a press release by a consultancy that sells sector analysis, and we cannot verify its methodology. The aggregate figures should be taken as indicative, and the full report is what would allow an assessment of whether its conclusions hold up.