Aon Sees 9.5% Employer Health Cost Hike Looming for 2027
Aon projects employer health costs up 9.5% in 2027. Overalls CEO Jon Cooper argues AI-driven job shifts make non-medical benefits the real differentiator for employers.
By Amara Osei
4 min read

What's News
- Aon projects U.S. employer health care costs will rise 9.5% in 2027 before employer offsets.
- KFF finds workers' contributions toward family coverage rose 9% this year.
- 73% of employers call non-medical benefits the most cost-effective well-being support, per MetLife's survey.
- OpenAI reports an internal model resolved more than 100 long-standing math problems.
- CMS projects health spending to grow faster than GDP through 2034.
Employer health care costs will rise 9.5% in 2027 before companies take offsetting steps, Aon projects, and that single figure frames the benefits problem Jon Cooper, CEO of Overalls, lays out in a new essay: medical costs will keep eating the benefits budget while AI rewrites who companies need and what those people do.
The workforce shift is already moving. Cooper writes that AI agents now write code, run tests, and carry out research assignments. OpenAI reports that an internal model has resolved more than 100 long-standing math problems.
Still, most companies haven't closed the gap between handing individuals better tools and changing how the business runs. Agents need access to company information, permission to act, and a way to get help when something goes wrong, Cooper argues. Research co-led by Chicago Fed economist Ezra Karger points to uneven adoption as a reason technical progress takes time to reach the wider economy.
What's the timeline?
Cooper's forecast runs in three steps:
- By 2028: more tasks handed to AI, with people checking the results.
- By 2030: companies reorganize around that work.
- By 2032: agents carry assignments across departments, bringing people in only when a decision or approval is needed.
Which jobs change first?
Software, law, and consulting will move fastest, Cooper writes. As coding, research, and drafting take fewer hours, people will focus on deciding what matters and checking the work. That creates a training problem: firms will have to figure out how people become experts when the work that used to teach them has been automated.
Stanford research already finds weaker hiring for young workers in AI-exposed jobs, though it does not establish causation.
Physical work will change more slowly. Better planning helps a manufacturer, but someone still has to maintain the equipment. Patients still need nurses, and homes still need electricians. Anthropic's scenarios show how cheaper knowledge work can support demand for hands-on work.
How bad is the medical cost problem?
The numbers stack up fast. Beyond Aon's 9.5% projection for 2027, CMS projects health spending growing faster than GDP through 2034. KFF finds workers' contributions toward family coverage rose 9% this year.
Employers can spend considerably more and still deliver bad news at open enrollment, Cooper writes. Employees see a bigger deduction from their paycheck. Explaining that the increase would otherwise have been worse doesn't make it feel like a better benefit.
At the same time, routine assignments will take fewer hours, and companies will depend more on experienced people. Losing someone with hard-earned judgment or trusted customer relationships will be harder for a lean team to absorb.
Cooper's verdict: good medical coverage remains essential, but competing on medical alone is becoming a losing proposition.
Do the cost-control tactics work?
Employers are trying three main levers, Cooper notes:
- Allowances for individual insurance, which Mercer says big companies are eyeing for select employee groups.
- Higher-deductible plans.
- Medical captives that pool large-claim risk.
These approaches change how costs and risks are shared. They don't stop medical inflation.
What replaces medical as the differentiator?
Non-medical benefits offer more room to make each dollar count. In MetLife's survey published earlier this year, 73% of employers called them the most cost-effective way to support employee well-being.
Cooper's company, Overalls, sells one answer: life navigation. The idea is a person who helps find care for a parent, sort out a disputed bill, or handle a family emergency — one place to turn, with someone who does the research, makes the calls, and stays with the problem until it's sorted.
AI can handle the routine legwork, Cooper writes, but human support still matters when life gets complicated, citing Aon analysis on AI, data, and personalization in benefits.
Leave and PTO, housing assistance, income protection, and financial well-being support also help employers stand out. People remember who helped when life got hard, Cooper writes — and that memory shapes where people choose to work, and whether they stay.
For benefits leaders, the 2027 projection is the deadline that matters: budgets built around medical inflation alone will keep losing ground, while the 73% MetLife figure suggests where employers are already moving their money.
Original: openai.com
More from Amara Osei
Show full bio
Senior reporter covering companies, deals and strategy in entrepreneurship, startups and venture funding at Business Bearings.
2,136 articles