Day 2 Notes From The 44th Annual J.P. Morgan Healthcare Conference

J.P. Morgan continues to surprise us at lunch, with the Mayor of San Francisco sharing on Monday an update on recent improvements in the city. Then on Tuesday at the Grand Ballroom lunch, it was the first episode of my favorite new sit-com "Everyone but Abe..."

You know, the one where for the first time ever, Dr. Mehmet Oz, the 17^th Administrator of the Centers for Medicare & Medicaid Services (CMS) flies out to the J.P. Morgan Healthcare Conference with almost his entire senior staff (the folks who actually run Medicare and Medicaid), except for... Abe Sutton, the Director of the Center for Medicare and Medicaid Innovation (CMMI). The audience was treated to a whirlwind tour of CMS's accomplishments over the past year, a hearty dose of MAHA perspective and compliments for President Trump, the estimated number of the people impacted by ACA and Medicaid changes is projected by CMS to be only 800,000 people), and repeated asks by Dr. Oz as to whether Abe Sutton was actually hiding somewhere in the ballroom (he was still in Washington, coming back from parental leave).

Dr. Oz also said that President Trump saved the Medicaid program, which otherwise would have expired as a result of its ever-increasing cost. Let's contrast that with a new Commonwealth Fund study estimating that, if Medicaid mandatory work requirements are imposed as legislated, U.S. hospitals' annual operating margins will drop in 2027 by 12.5% to 14.2%. Keep in mind that we heard at the conference separately today that many hospitals among the roughly half of U.S. hospitals that actually are profitable this year operate only at a 2% to 3% margin post-Covid. Doing the math, that would put these hospitals, if the estimates are correct, underwater with margins roughly negative 10% to 12%. And that is before any further impact from ACA subsidy expiration or possible site-neutrality or 340b program changes.

Dr. Oz made it very clear that the Trump Administration will be going after what he described as high levels of fraud at the state level, such as in Minnesota and California, and suggested these moneys could be made available to fund hospitals and healthcare (though no structure was shared for such funding). He also said that states would no longer be able to artificially inflate their Medicaid dollars owed by the federal government through sophisticated approaches like provider taxes. According to CMS, funding responsibilities should remain within states, meaning Oklahoma should not have to fund California.

Here are some other key takeaways of substance:

• CMS is going to go slow now to later go fast on Medicare Advantage. Rather than act precipitously, CMS is taking time to study Medicare Advantage and consider new policies that will address the "loss of faith" in Medicare Advantage and the way in which some companies have taken advantage of the system. So, based on the comments shared by CMS, it is unlikely that we will see material MA changes that would affect 2026 or 2027, but there may be significant changes that could affect future years. Dr. Oz said CMS wants to get the new policies right the first time, as it's important in the view of Chris Klomp, the new head of Medicare, to create stability and predictability for Medicare Advantage. The November 2025 CMS Request for Information on how to improve and modernize Medicare Advantage kicked off that re-evaluation process (response deadline for the RFI is January 26, 2026).
• CMS leadership reaffirmed their commitment to risk and value-based reimbursement models, effusively praising the new CMMI Access Model for technology-enabled chronic disease management and the LEAD model ACO program starting in 2027. WISeR will test new prior authorization requirements for original Medicare. (And no, Abe Sutton still wasn't there to receive the praise heaped on by Dr. Oz.)

Revenue Cycle Management and AI

Today was revenue cycle management day at the conference, with Ensemble and Smarter Technologies both presenting. Some common takeaways were evident:

• AI is transforming revenue cycle management (RCM), with Smarter Technologies reporting not only reductions in costs through AI but increased revenue for client hospitals in the range of 40 to 80 basis points. If the smaller hospital system post-COVID operating margin is now in the 2-3% range (assuming they are making money, which half are not), then added revenue in that range could add almost 20-30% to operating margin – which is transformative and a lifeline for hospitals. We have seen multiple smaller health systems considering sale transactions currently because, while they are profitable at these margins, they have weak balance sheets and don't have the necessary cash flow to undertake transformational initiatives. If a platform-based RCM vendor can increase revenue and margin, that could give a new lease on life to hospitals (at least until the Medicaid and ACA changes weigh in...).
• The combination of AI plus human experts appears to be the favored approach, resulting in better performance, more rapid and continuous training of LLMs and acceleration of improvement implementation. Smarter Technologies discussed how they are deploying new models into their RCM products on a weekly basis, a rate of substantive functional improvement not previously seen with hosted or SAAS-based RCM software systems.
• Collaboration between AI and human experts also allows for human insertion into the development process of what Anthropic calls "taste, judgment and agency." Taste and judgment refer to the human ability to discern quality, context and relevance. As RCM company Ensemble shared today, contextual reasoning still is needed for many healthcare tasks and is best done by humans. For example, taste in data curation is not just about quantity but also quality and relevance, so that the first information received by the user feels right and is responsive. We also want an LLM model that is perceived as specialized but not rigid, striking a balance that nonverbally accords with human "insider/outsider" instinctual warning patterns and doesn't impede the building of trust and acceptance by the user. Used correctly, human "taste" can guide the LLM post-training process to a result where the models provide warm and helpful responses, rather than robotic templates or responses that feel creepy or smarmy or that can be off-putting and reduce usage or adoption by customers. Per Mike Kriger, Chief Products Officer at Anthropic, taste can factor in value beyond metrics, that intrinsically human nonverbal sense of knowing when something feels right or wrong. There's a reason that when we value someone's perspective, we say that the person has "good taste." It's because we are in accord and willing to accept that person's definition or perspective, rather than having to edit it ourselves or simply reject it. Applying that systematically, taste-driven development might begin with taste distillation by looking at samples and identifying what feels right and what doesn't, essentially creating a taste blueprint. That distilled knowledge can then be injected into the agent's post-training process, influencing later iterations (whether of code writing, data presentation, color schemes, textual voice or sound timbre). Through an iterative loop, with collaboration between the AI model and human experts, the end product can be improved to balance expressive design with functional rigor.
• Making sure your RCM AI has an orchestration layer where data can be taken in from multiple sources is critical to a scalable, adaptable RCM system. That layer then allows for longitudinal analysis and process improvements by making it easier to identify where something went wrong. That analysis itself can be fed back into the LLM model for further post-training and refinement.
• It's an AI arms race in RCM now between payors and providers, and it will continue. As AI gets better both offensively and defensively, who will bear the cost of ongoing technology upgrades and the possible loss or recoupment of prior savings? Who will address the costs and disruptions of increased payor integrity audits of providers? Does this suggest that health systems may accelerate the risk shifting of this RCM exposure to a full-platform outsource arrangement, continuing the trend we have seen? We have not seen in capital expenditure budget percentages shared by leading health systems at the conference over the past two days enough dollars to fund at scale a payor/provider technology arms race. As one RCM company said today, scale is needed to negotiate and deal effectively with today's payors, and a single health system in one or limited markets cannot generally wield sufficient scale to achieve a balance with a national payor.

Health Systems and K-Curves

The K-Curve has been used recently to describe the ongoing bifurcation of the U.S. economy, with widely divergent outcomes for lower-income and higher-income people. The higher earners are on the upward slope of the K, benefiting from the stock market, surviving the effects of tariffs on consumer goods and not cutting back significantly on their spending, while the lower earners are on the downward slope of the K, facing ongoing financial strain, increased debt, feeling the pressures of higher costs from food, energy, healthcare and education, while wages have not grown enough to offset those pressures. In effect, the K curve is a symbol of rising inequality.

At the J.P. Morgan conference, we saw the K-Curve in action for the hospital and health system sector. Many of the largest and most highly respected, branded hospitals and health systems presented, with a common message of:

• Significant year on year growth often in the double-digit range of 10 – 15%, with revenue increasing due to higher fee for service utilization and lagging or absent deployment of value-based or risk-based reimbursement models.
• Margins in the high single digits, often 8% to 12%.
• Days Cash on Hand well above 200 days, often in the 250 – 300+ days range, with strong balance sheets that often show a 200 – 400% cash to total debt ratio.
• Low weighted average cost of capital, usually at or around 3%, as most of these systems were using bonds or cash on hand from current earnings for major capital expenditures, with little variable rate debt exposure.
• Expansion being planned and occurring, with multiple renderings and photos of new towers, innovation districts, community ambulatory centers being built or opened, as well as a merger and acquisition pipeline that is being filled.
• Operational restructurings to unify service lines under common governance across facilities, to create high quality, accountable local management and to divest, partner or close non-core services and facilities, especially those with higher churn and lower margins. Management is being done by service lines and not by hospitals.
• A differentiation is occurring between conglomerate status where multiple unrelated businesses do not create operating synergies (bad) and effective revenue diversification strategies that support the core and buffer headwinds (good).
• A continued move into ambulatory care in the communities served – "meeting the patient where they are" – and remedying the lack of ambulatory surgery centers, urgent care centers and imaging centers through acquisitions, partnerships and development.
• Moving case mix index (CMI) upward in complexity as patients with lower acuity are seen more in ambulatory settings, in hospital at home programs or in alternative care sites. This intentionally also allows hospitals serving aging populations to be more intensive and focus more on complex coordinated care, with the highest, best use of resources and capital, with opportunities for increased Medicare reimbursement.

What does this mean for hospitals? If these patterns persist – and if no externality occurs (like a pandemic, a recession or wholesale regulatory changes) – the largest health systems with healthy balance sheets should continue to see their revenue grow and be able to invest in growth and broader services. Maintenance of the fee for service paradigm means that, as Americans get older and sicker, the larger hospitals and health systems will be able to meet those needs, while surviving and, in some cases, thriving. From a systemic perspective, though, this does not likely bend the total cost of care curve downward.

The bottom half of the K-Curve will not do as well. Median health system operating margins held near 1% throughout 2025 (per Strata), with smaller health systems reporting declines in operating revenue and margins from the prior year. 37% of hospitals in the U.S. were reported by Kaufman Hall to be operating at a loss in 2025, down slightly from 40% in 2024. The looming changes to Medicaid and the expiration of ACA subsidies also may drive down smaller health system revenue, with negative impact.

The continuing move to ambulatory care voiced by the larger presenting health systems at the conference also will hit hard for smaller health systems. Those smaller health systems may not be able to recoup the revenue lost when an ortho or cardiac procedure moves out of a hospital operating room into an ambulatory surgery center, or when oncology cases are lost to the nearest academic medical center or national cancer center as a result of employers putting in place a center of excellence program. And, with weak balance sheets, they can't develop or acquire the necessary facilities to retain the transferring revenue leading to a loss of physicians and a resulting downward spiral.

We expect to see health system mergers and acquisitions and joint operating agreements continue to accelerate in 2026 and 2027, with accompanying closures or bankruptcies for hospitals that cannot find partners. Many hospitals and health systems will wait too long, rather than seeking a partner earlier, prior to financial degradation that can result in a worse deal or potentially no willing partner at all. "Don't wait, act boldly" should be the mantra for leadership in these difficult times.

On to Day 3 of the conference, with further thoughts on oncology, neuroscience, ambient listening and the "once in a lifetime" moment.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.