American hospitals have validated algorithms. Their large-scale deployment encounters institutional and governance obstacles. According to the NAM discussion paper published on March 9, 2026, the fragmentation of health architecture and data limits digital transformation. The digital health market in North America is experiencing expected growth, associated with telehealth, reimbursement conditions, and demographic and health factors.
The Essentials
- According to the National Academy of Medicine (March 9, 2026), clinical AI encounters obstacles in empirical validation, monitoring, digital architecture and interoperability, among other infrastructure reforms.
- Boston, San Francisco, and Nashville are home to major hospital systems and constitute zones of critical stakes for the digital transformation of health.
- American energy federalism fragments the response: each state regulates its grid, complicating coordination without preventing the existence of federal and federal-state coordination mechanisms.
- The North American digital health market is the subject of growth projections, conditioned on the resolution of validation and governance issues.
- Solutions exist: hospital microgrids, dedicated renewable energy contracts, revision of federal grid allocation priorities.
The NAM Diagnosis: The Algorithm Awaits the Current
The National Academy of Medicine published on March 9, 2026 the discussion paper “Toward a National Health Digital and Data Architecture” on the conditions for digital transformation of American health. Its title is sober, Toward a National Health Digital and Data Architecture, but its content is less so. The text recommends establishing protocols for validation, monitoring, and governance of AI systems.
Clinical AI, whether radiological diagnostic assistance, early sepsis detection, predictive bed management, or automated pharmacovigilance, requires considerable computing power, available continuously and without interruption. The data centers hosting these systems consume energy in large quantities, creating tensions with network capacity. Demand has accelerated faster than networks have expanded.
The result is a queue. Data center operators report capacity extension delays of several years in the most constrained areas. Some hospital systems that have signed contracts with AI providers report that capacity extension delays are affecting their deployment schedules.
Boston, San Francisco, Nashville: Three Cities, One Problem
These three cities concentrate a disproportionate share of medical research, tertiary care, and clinical training institutions on the continent. Boston is home to Massachusetts General Hospital, Brigham and Women’s, the Dana-Farber Cancer Institute, and systems affiliated with Harvard and MIT. San Francisco hosts the UCSF Medical Center and is part of California’s digital health ecosystem. Nashville is home to HCA Healthcare, the world’s largest private hospital operator, as well as the Vanderbilt University Medical Center.
These three metropolises are also, each in their own way, nodes of digital infrastructure. The concentration of major health institutions in these three cities creates critical issues of alignment between AI investments and available clinical evidence. In Boston, Massachusetts’ electrical grid manages growing demand with production and transmission capacity inherited from a pre-digital era. In San Francisco, California has undertaken ambitious energy transitions, but grid reliability remains a sensitive subject since the preventive outages of the 2010s. In Nashville, Tennessee’s rapid economic growth has outpaced infrastructure investments.
In these three cities, major hospital systems face electrical capacity challenges for their deployment.
Energy Federalism as a Friction Multiplier
The electrical problem would already be difficult to solve if it depended on a single decision-maker. American federalism makes it structurally more complex. American electrical regulation is shared between federal authority, notably for interstate transmission and wholesale markets, and state and local authority for distribution and retail. The Federal Energy Regulatory Commission (FERC) oversees wholesale markets and interstate interconnections, while states and municipalities generally regulate distribution and retail rates.
For a hospital system operating in multiple states, which is the case for most large groups like HCA or CommonSpirit Health, electrical planning falls under distinct federal, state, and regional authorities, which can complicate certain multi-state projects. Each expansion of digital infrastructure in a new jurisdiction involves navigating a different regulatory environment, with different approval delays and different interlocutors.
The NAM calls for inter-agency coordination and a national digital and data architecture. The digital transformation of health presupposes a certain homogeneity of infrastructure conditions, which exists only partially on the data side, where standards like HL7 FHIR are advancing, unlike on the energy side, where jurisdictional silos remain intact.
Measures Taken to Unlock the Situation
The blockage is real, but it is not static. Several actors have begun to circumvent network constraints while waiting for national capacity to strengthen.
The first path is that of hospital microgrids. Some large hospital systems are investing in on-site energy production and storage capacity, solar panels, batteries, cogeneration, to reduce their dependence on the general grid and stabilize the power supply to their data centers. Brigham and Women’s Hospital in Boston has announced significant investments in this direction. These solutions remain costly and are accessible only to institutions with a solid financial position.
The second path is through long-term renewable energy contracts, directly negotiated with producers. Several health data center operators have signed Power Purchase Agreements that guarantee them dedicated capacity, without going through the queues of regional networks. This approach displaces the problem rather than solving it: it secures energy for actors with the critical size to negotiate, and leaves smaller facilities without an answer.
The third path is regulatory. The Reuters Digital Health 2026 conference saw an unusual consensus emerge between hospital operators and technology managers: an explicit call for prioritization of health data centers in network allocation plans, on the same level as traditional critical infrastructure. The U.S. Department of Energy is conducting capacity analyses that could inform this debate. This is a policy lever that has not yet been activated, but whose discussion has entered the doors of federal agencies.
Clinical AI Waiting: What the Delay Actually Costs
The issue is not limited to a matter of investment or economic competitiveness. Delays in deploying clinical AI can have clinical implications. Early sepsis detection tools, a generalized infection associated with at least 350,000 annual deaths in the United States according to Centers for Disease Control data, have shown in some observational or prospective studies an association with variable mortality reductions. Deployment delays raise questions of access and equity that digital health economic models do not systematically address.
The same reasoning applies to radiological diagnostic assistance. The FDA authorizes the commercialization of certain medical devices incorporating AI; deployment delays can result from multiple factors, including clinical validation, monitoring, and integration. Some tools are FDA-approved, but clinical validation, monitoring, integration, and governance remain deployment issues. Clinical trials have been conducted. Regulatory validations are progressing.
Digital medicine is running up against a constraint that belongs to another industrial century.
This gap also raises questions about the geography of benefits. AI is transforming labor markets at very different rates across sectors; in health, it risks widening the gap between major metropolises capable of investing in alternative solutions and regional systems that depend entirely on an under-capacity national infrastructure.
Expected Changes Between 2030 and 2035
Market projections for North American digital health are to be read as indicators of direction, not as certainties. They describe a potential conditional on the resolution of infrastructure bottlenecks. If electrical capacity in health metropolises does not keep pace, some of these investments will migrate to better-equipped jurisdictions, or to cloud architectures hosted in less constrained states, which raises other questions about data sovereignty and clinical latency.
The 2030-2035 horizon is also the one at which the Department of Energy models network expansion needs for 2030, 2035, and 2040. The Grid Deployment Office, created in 2021, is handling projects to strengthen national interconnections, several of which directly concern the Boston and San Francisco regions. These programs are advancing, but network constraints can delay some data center projects.
The institutional question raised by the NAM remains open: building a national health digital and data infrastructure would likely require more inter-agency coordination, resources, and dedicated authority. The analogy with the internet in the 1990s is useful: the scale-up of the digital network had then benefited from federal coordination that had made it possible to harmonize investments in telecommunications infrastructure. Coordination mechanisms between energy and health digital sectors show gaps. Debates on training large-scale AI models show that questions of energy infrastructure do not arise only in the United States, but nowhere else does federalism add such a dense layer of fragmentation.
The undertaking is not insurmountable. States like Virginia, which already hosts the world’s largest concentration of data centers, have developed accelerated authorization procedures and predictable pricing mechanisms that could serve as a model. The next National Governors Association conference on digital infrastructure, scheduled for fall 2026, could be an opportunity to formalize this type of exchange between states. The pressure from hospital operators, who now have a direct economic and clinical interest in unlocking the situation, is a lever that policymakers have not yet fully mobilized.
Sources
- National Academy of Medicine, Toward a National Health Digital and Data Architecture: Laying the Foundation for Digital Transformation (June 2026), https://nam.edu/perspectives/toward-a-national-health-digital-and-data-architecture-laying-the-foundation-for-digital-transformation/
- Reuters 2026 Digital Health Conference, journalistic coverage (reuters.com/business/healthcare)
- US Department of Energy, Grid Deployment Office, electrical grid capacity analysis reports (energy.gov/gdo)
- Centers for Disease Control and Prevention, data on sepsis mortality in the United States (cdc.gov)
- North American digital health market reports, market projections 2025-2035 (aggregated sector sources, no single verifiable URL)
