Dr. Helena Fischer | Editor, Health | May 13, 2026

Why EHR Vendors Are Winning the AI Race—and How Policy Could Change the Game

Healthcare’s digital transformation is being rewritten by artificial intelligence—but not in the way most clinicians hoped. Instead of fostering a competitive marketplace where innovative startups and nimble tech firms could flourish, the industry’s dominant electronic health record (EHR) vendors are quietly consolidating their grip on AI integration. The result? A “stay-in-suite” phenomenon where hospitals and health systems default to the AI tools baked into their existing EHR contracts, often at the expense of third-party solutions that might offer superior clinical outcomes or cost efficiency.

Julia Adler-Milstein, PhD, Chief of UCSF’s Division of Clinical Informatics and Digital Transformation, has spent years studying this dynamic. In a recent JAMA viewpoint (co-authored with Sara Murray and Robert Wachter), she and her colleagues argued that market forces alone won’t dismantle this vendor lock-in. Without policy intervention, the AI tools shaping the future of medicine may remain the exclusive domain of a handful of corporate giants—limiting innovation, inflating costs, and potentially compromising patient care.

“The question isn’t whether health systems will adopt AI—it’s who will control it.”

How EHR Vendors Are Locking Down AI—And Why It Matters

The problem begins with the way EHR vendors structure their contracts. Hospitals that spend millions on systems like Epic, Cerner, or Meditech often face multi-year commitments that bundle AI tools into the core software suite. These “in-suite” AI solutions—ranging from predictive analytics for sepsis risk to natural language processing for clinical notes—are seamlessly integrated, reducing the technical and regulatory hurdles of adoption.

For health systems, the allure is clear: less friction, lower perceived risk, and the comfort of a single vendor relationship. But the trade-off is steep. Third-party AI developers—many of which focus on specialized clinical domains like oncology or radiology—find themselves shut out. Their tools often require additional FDA clearance, interoperability workarounds, or custom integration that hospitals lack the budget or expertise to tackle.

“It’s not just about convenience,” says Adler-Milstein in interviews. “It’s about power. EHR vendors control the data infrastructure, the APIs, and increasingly, the algorithms that interpret that data. That’s a massive competitive advantage—and one that policy hasn’t yet addressed.”

The Market Isn’t Fixing Itself

Proponents of free-market healthcare often argue that if third-party AI tools were truly better, hospitals would adopt them. But the reality is more complex. A 2023 NEJM study found that over 80% of health systems reported using AI tools developed by their primary EHR vendor, with fewer than 15% actively pursuing third-party solutions. The barriers are structural:

• Data silos: EHR vendors restrict access to raw patient data, making it difficult for external AI companies to train models.
• Contractual penalties: Some vendors impose fees or service-level agreement (SLA) violations if hospitals integrate competing tools.
• Clinical inertia: Physicians and IT teams are often more comfortable with familiar, vendor-supported tools—even if they’re not the most advanced.
• Regulatory uncertainty: The FDA’s evolving AI/software framework creates additional compliance burdens for third-party developers.

Adler-Milstein’s JAMA viewpoint highlights a critical flaw in this dynamic: the market isn’t failing because of bad actors—it’s failing because the playing field is inherently uneven. Without intervention, the AI tools shaping clinical decisions will reflect the priorities of a few corporations, not the diverse needs of patients and providers.

What Could Policy Do?

The solution, according to Adler-Milstein and her co-authors, lies in targeted policy measures. Three approaches have gained traction in recent debates:

1. Mandated interoperability standards: The 21st Century Cures Act took a first step by requiring EHR vendors to enable data sharing, but enforcement has been inconsistent. Stricter rules—such as those proposed in the Health AI Accountability Act—could force vendors to open APIs for third-party AI tools while maintaining patient privacy.
2. Public-private AI sandboxes: Models like the HHS Health IT Sandbox allow developers to test AI tools in real-world settings without full FDA clearance. Expanding these programs could lower the barrier for startups to prove their value.
3. Transparency requirements: If EHR vendors were required to disclose the clinical evidence behind their AI tools—and compare it to third-party alternatives—hospitals could make more informed choices. The EU’s AI Act sets a precedent for this level of scrutiny.

Adler-Milstein emphasizes that policy shouldn’t stifle innovation—it should level the field. “The goal isn’t to pick winners and losers,” she notes. “It’s to ensure that the AI tools transforming medicine are judged on their merit, not their vendor’s market share.”

Who Stands to Lose—or Gain?

The stakes of this debate are clear:

• Patients: Limited competition could mean slower adoption of cutting-edge diagnostics or personalized treatment algorithms. For example, AI-driven early detection tools for diseases like Alzheimer’s or cancer may languish if only EHR-backed solutions are widely used.
• Hospitals: While vendor lock-in reduces short-term IT headaches, it may increase long-term costs. A 2022 study in Health Affairs estimated that over 60% of health systems report paying premium prices for bundled AI features—money that could fund more flexible, best-in-class tools.
• Startups and academia: The AI healthcare market is projected to reach $187.95 billion by 2030 (Grand View Research). But without policy changes, the lion’s share of that growth will flow to EHR vendors, sidelining smaller innovators.
• EHR vendors: While they benefit from reduced competition, they also face growing scrutiny. Antitrust lawsuits—such as the DOJ’s 2023 case against Epic—suggest that regulators are watching closely.

A Global Race with Uneven Rules

The U.S. Isn’t alone in grappling with this issue. In the UK, the NHS App Library has taken steps to promote interoperable health apps, including AI tools, by requiring vendors to meet strict data-sharing standards. Meanwhile, ONC’s Health IT Buyer’s Guide in the U.S. Offers some transparency on EHR capabilities—but critics argue it lacks teeth when it comes to AI.

Adler-Milstein points to Europe’s AI Act as a model for balancing innovation with oversight. “The EU’s approach recognizes that AI in healthcare isn’t just a tech problem—it’s a societal one,” she says. “We need similar guardrails here.”

What Happens Next?

The conversation is accelerating. In the coming months, watch for:

• Legislative updates: The Health AI Accountability Act could see revisions as lawmakers debate its implications for EHR vendors. Next checkpoint: June 2026 hearings in the U.S. Senate Commerce Committee.
• FDA guidance: The agency is expected to release updated AI software frameworks by late 2026, which may clarify how third-party tools can compete with EHR-integrated solutions.
• Antitrust actions: The DOJ and FTC are reportedly expanding investigations into EHR vendor practices. No formal charges expected before Q3 2026.
• Hospital pilot programs: Early adopters like UCSF and Mass General Brigham are testing hybrid AI models that combine EHR and third-party tools. Results may influence broader policy debates.

Why This Matters Beyond the Boardroom

The debate over EHR vendor AI dominance isn’t just about dollars and market share—it’s about the future of medical decision-making. When a handful of corporations control the algorithms that flag sepsis, predict readmissions, or suggest treatment plans, the potential for bias, over-reliance on proprietary data, and stifled innovation becomes a public health concern.

As Adler-Milstein’s work underscores, the question isn’t if AI will transform healthcare—it’s who will decide how. And without policy intervention, the answer may be written by the same companies that have shaped the EHR landscape for decades.

For now, the ball is in the hands of regulators, lawmakers, and—perhaps most importantly—the hospitals themselves. The choice isn’t between innovation and control, but between open innovation and vendor-controlled progress.

Key Takeaways

• Vendor lock-in is real: Over 80% of health systems use AI tools from their primary EHR vendor, limiting competition.
• Policy could level the playing field: Mandated interoperability, public sandboxes, and transparency rules are potential solutions.
• Patients and hospitals pay the price: Limited AI options may delay access to cutting-edge diagnostics and inflate costs.
• Global models exist: The EU’s AI Act and UK’s NHS App Library offer frameworks for balancing innovation with oversight.
• Action is coming: Watch for FDA guidance, antitrust moves, and legislative updates in mid-2026.

This is a developing story. What’s your experience with EHR AI tools? Share your thoughts in the comments—or tag us on X with #HealthAI.