Insights

AI's looming path to failure in healthcare: as we automate, we must renovate

Erik Abel, PharmD, MBA · January 22, 2026 · 14 min read

Change my mind. The healthcare industry's obsession with AI ignores the real bottleneck, legacy infrastructure. In 2025, why are we still faxing records or transferring images on discs? Maybe we should fix that before proposing AI as the cure-all. To take it a step further, what might AI have to say about the adoptability of AI in the U.S. healthcare system, its deficiencies, and the quicksand that surrounds it?

Introduction

Robert Pearl, M.D. recently sparked an important discussion on the role of generative AI in healthcare, highlighting its potential to reduce medical errors and improve patient outcomes. His post posed a critical ethical question.

What are the moral implications of NOT using AI in medicine?

This triggered a more visceral response based on my own experience, for which I posed an equally thought-provoking perspective. Before we expect AI to transform healthcare, we need to address the fundamental barriers holding back innovation, legacy infrastructure, interoperability challenges, and fragmented data systems. Ultimately I arrived at the question of what systematic barriers exist that impose limitations on advancing care through this fascinating technology. I also wondered aloud whether anyone has applied a tool like a PESTLE analysis to the U.S. healthcare ecosystem in recent times before jumping to the conclusion that AI is our solution.

To explore this further and satisfy my own curiosity, I conducted an experiment. I asked ChatGPT Atlas 4.0 and Claude AI Sonnet 3.7 (both paid versions) to conduct a PESTLE analysis (Political, Economic, Social, Technological, Legal, Environmental) of the U.S. healthcare system with a focus on outdated infrastructure. The results reveal a consistent theme, a diagnosis and treatment plan of sorts, for those with curious eyes.

What is a PESTLE analysis and why is it used?

A PESTLE analysis is a strategic planning framework used by companies, policymakers, and industries to assess external factors that influence their ability to operate, innovate, or enter a market, and it is often a tool used in international business. It evaluates the Political, Economic, Social, Technological, Legal, and Environmental forces shaping an industry's ecosystem. In healthcare, this type of analysis is critical for identifying structural barriers, such as outdated technology, regulatory constraints, and misaligned incentives, that can either hinder or enable progress.

ChatGPT vs Claude: a U.S. healthcare PESTLE analysis

Executive summary

ChatGPT output. The U.S. healthcare system remains constrained by outdated legacy infrastructure that inhibits digital transformation and data-driven advancements. The industry is burdened by fragmented regulations, technological stagnation, misaligned incentives, and entrenched vendor interests, all of which hinder innovation and efficiency. Before AI can fulfill its potential, healthcare must address foundational issues, particularly around interoperability, data liquidity, and vendor-neutral cloud adoption. This PESTLE analysis evaluates these systemic barriers and outlines strategic recommendations to modernize healthcare's digital infrastructure.

Claude AI output. Having analyzed the U.S. healthcare system through a PESTLE framework, I believe we are approaching a critical breaking point where legacy infrastructure is not merely slowing progress, it is actively endangering care quality and system sustainability. The conventional narrative often frames this as a technological challenge, but my assessment reveals a more fundamental truth. Healthcare's legacy problem is primarily one of misaligned incentives, institutional inertia, and regulatory capture.

The observation that we must start more foundationally with data exchange and interoperability, before or at least in parallel to pursuing AI advancements, is precisely correct. The healthcare industry's stubborn adherence to outdated technologies, COBOL-based claims processing, fax machines, and CD/DVD transfers for medical imaging, creates a fragile foundation incapable of supporting transformative innovation.

Political factors (P)

ChatGPT output. Healthcare infrastructure reform is hindered by bureaucratic stagnation, powerful vendor influence, and a disconnect between policy discourse and operational necessities. Rather than addressing fundamental issues like interoperability and outdated systems, political attention is disproportionately directed toward AI and high-profile innovations, diverting resources from essential modernization efforts.

A major issue is the political discourse's fixation on AI and emerging technologies while neglecting the deteriorating digital backbone of healthcare. This misalignment between aspirational policy goals and the industry's operational realities ensures that systemic inefficiencies persist, stalling progress where it is most needed. Regulatory inertia and entrenched vendor interests obstruct true interoperability, preventing healthcare from modernizing its digital infrastructure effectively.

Federal vs. state regulatory disparities. Conflicting federal and state regulations create inconsistencies in digital health adoption and data exchange.
21st Century Cures Act and TEFCA. While promoting interoperability, these mandates are hindered by vendor resistance and misaligned business models.
HIPAA's double-edged sword. Originally designed to protect patient data, HIPAA now serves as both a safeguard and an obstacle to seamless data sharing.
Value-based care (VBC) vs. fee-for-service (FFS). VBC models encourage data-sharing efficiencies, but FFS remains dominant, limiting IT investments in infrastructure.
Health IT vendor concentration. A few dominant EHR vendors (Epic, Cerner/Oracle, Meditech) prioritize proprietary systems over open APIs and interoperability.

Claude AI output. The political landscape surrounding healthcare infrastructure reveals deeply entrenched barriers to modernization.

Fragmented authority and regulatory hesitance. Multiple federal agencies (ONC, CMS, FDA) share overlapping jurisdiction over healthcare technology without clear coordination. This fragmentation creates policy gaps where no single authority can drive comprehensive modernization.
Lobbying influence on infrastructure policy. The healthcare technology market is shaped more by lobbying power than innovation merit. Established vendors spent over $160 million on lobbying between 2019 and 2023, effectively maintaining barriers to transformative change through regulatory capture.
Short-term political horizons vs. long-term infrastructure needs. The political cycle's focus on short-term wins disincentivizes the multi-year investments needed for fundamental infrastructure modernization. Politicians can announce initiatives, but rarely remain in office long enough to ensure implementation and accountability.
Cloud strategy political implications. Government policies regarding data sovereignty, privacy, and security increasingly impact cloud provider selection, requiring healthcare organizations to maintain flexibility in their infrastructure strategies as political landscapes shift.

What is particularly troubling is how political rhetoric about healthcare innovation focuses overwhelmingly on AI and cutting-edge technologies while ignoring the crumbling infrastructure underneath. This disconnect between political narratives and technical realities perpetuates a cycle where foundational problems remain unaddressed.

Economic factors (E)

ChatGPT output. Healthcare's financial structures disincentivize IT modernization, keeping legacy systems entrenched. Payers and providers operate under financial constraints that prioritize immediate cost containment over long-term transformation, perpetuating inefficiencies and limiting digital adoption. The cost of maintaining outdated infrastructure is high, yet stakeholders resist investment due to short-term budget pressures and unclear ROI from digital transformation.

Legacy system maintenance costs. Many organizations still operate COBOL-based claims processing and outdated EHRs, making upgrades prohibitively expensive.
Payer reluctance to modernize IT. Commercial payers prioritize short-term cost control over long-term digital transformation.
Data lock-in costs. EHR vendors charge high fees for data extraction, hindering interoperability efforts.
Underinvestment in IT. Healthcare invests only about 4% of revenue in IT, compared to 10% or more in industries like banking and retail.
Slow AI adoption due to poor data liquidity. AI promises efficiency gains but struggles with fragmented and low-quality data.

Claude AI output. The economics of healthcare's legacy infrastructure reveal profound structural barriers to modernization.

The interoperability business dilemma. Major health IT vendors have built business models that financially benefit from limited interoperability. When data exchange becomes difficult or expensive, organizations are incentivized to stay within a single vendor ecosystem, reinforcing market concentration.
Maintenance economics vs. modernization investment. Healthcare organizations face an increasingly impossible equation. Maintenance costs for legacy systems consume 70% to 80% of IT budgets, leaving insufficient resources for transformative investments. This creates a downward spiral where organizations cannot invest their way out of legacy constraints.
False economy of technological conservatism. The perceived cost savings of maintaining legacy systems versus modernizing often fail to account for hidden costs such as security vulnerabilities, reduced operational efficiency, clinician burnout, patient dissatisfaction, and the opportunity cost of delayed innovation.
Cloud provider financial leverage. Single-provider cloud strategies expose healthcare organizations to significant financial risk through potential price increases, specialty service premiums, and data egress fees. Cloud-neutral approaches provide economic flexibility and negotiating leverage.

Perhaps most concerning is how economic pressures have created an infrastructure inequality gap in healthcare. Large, resource-rich health systems can invest in modernization while smaller community hospitals and rural providers fall further behind, creating a two-tiered system with profound implications for care equity.

Social factors (S)

ChatGPT output. Cultural resistance and workforce limitations create friction in healthcare's digital evolution. Clinicians remain overburdened by administrative tasks, patients expect more seamless digital experiences, and a shortage of healthcare IT professionals hampers the industry's ability to implement data-driven advancements. Meanwhile, patients demand a digital experience comparable to other industries but are stuck navigating antiquated systems.

Risk-averse industry culture. Healthcare remains hesitant to adopt new technology due to regulatory complexity and fear of disrupting workflows.
Clinician burnout and IT frustration. EHR systems are often viewed as administrative burdens, leading to user dissatisfaction and inefficiency.
Healthcare IT workforce shortage. The lack of skilled data analysts and informaticists slows the adoption of AI and data-driven solutions.
Patient expectations vs. reality. Consumers expect digital convenience, but the system still relies on CDs, DVDs, and flash drives for imaging transfer along with paper-based processes.

Claude AI output. The social dynamics surrounding healthcare technology reveal critical barriers to progress.

The clinical-technical divide. Healthcare organizations typically separate clinical and technical leadership, creating persistent communication gaps where neither group fully understands the other's constraints and requirements. This division complicates modernization efforts that require deep integration of clinical and technical considerations.
Patient expectations vs. healthcare reality. Today's patients, accustomed to seamless digital experiences in other industries, increasingly expect the same from healthcare. The growing gap between these expectations and the reality of paper forms, faxed records, and CD-based imaging creates mounting pressure for change.
Workforce adaptation challenges. Healthcare's aging workforce faces significant challenges adapting to new technologies. The average nurse is 52 years old, and the average physician is 51, demographics that complicate digital transformation efforts and create resistance to change.
Cloud skills development gap. The transition to cloud-based infrastructure requires specialized expertise that differs significantly from traditional healthcare IT skills. Organizations pursuing cloud-neutral strategies face additional challenges in developing staff proficient across multiple platforms.

What is often overlooked is how the persistence of legacy infrastructure directly contributes to healthcare's well-documented burnout crisis. Clinicians spend 1 to 2 hours on EHR and administrative work, such as prior authorization, for every hour of direct patient care, a ratio that reflects not just poor software design but the fundamental limitations of outdated infrastructure.

Technological factors (T)

ChatGPT output. The continued reliance on outdated IT infrastructure severely restricts healthcare's ability to integrate AI, real-time data exchange, and cloud-based solutions. Without modern data architectures and real-time interoperability, the industry remains trapped in inefficiencies that limit advancements in patient care, diagnostics, and operational efficiency, for example fax machines, eFax, and flat-file exchanges via SFTP.

COBOL-based payer systems. Claims processing remains reliant on outdated mainframe technology, limiting real-time processing and efficiency.
Fax and e-fax persistence. Over 70% of medical communication still occurs via fax, stifling real-time data exchange.
DICOM image accessibility gaps. Imaging data is primarily stored as reports rather than full imaging files, restricting AI-driven diagnostics.
Health information exchanges (HIEs) remain incomplete. Many HIEs suffer from partial participation, outdated architecture, and limited interoperability.
Interoperability via APIs remains theoretical. While FHIR APIs have promise, real-world implementation is inconsistent and lacks enforcement mechanisms.
Lack of AI-ready data infrastructure. Without structured, standardized, and accessible datasets, AI adoption remains limited.

Claude AI output. The technological landscape of healthcare infrastructure reveals critical vulnerabilities that align precisely with the concerns raised.

HIE implementation gaps. While Health Information Exchanges exist in most regions, their practical utility remains limited by incomplete participation, inconsistent data quality, and poor integration with clinical workflows. Most HIEs function more as data repositories than truly active exchanges influencing point-of-care decisions.
Imaging data limitations. As noted, most healthcare systems only exchange reports about imaging studies rather than the images themselves. This limitation profoundly restricts clinical utility, creates redundant testing, and blocks AI implementation that would require access to the original imaging data.
Pharmacy data isolation. Community pharmacy systems represent information islands, disconnected from broader clinical data ecosystems despite capturing vital medication adherence and potentially clinical monitoring data. This isolation creates dangerous blind spots in medication management.
Cloud architecture flexibility. Cloud-neutral approaches using containerization, microservices, and platform-agnostic development create significantly more flexible technology foundations than either legacy systems or single-provider cloud strategies.
Technological anachronism. The persistence of COBOL-based claims systems, fax communications, and CD/DVD-based image transfer is not merely an embarrassing anachronism. It represents fundamental infrastructure limitations that directly impact patient safety, clinical efficiency, and innovation capacity.

Legal factors (L)

ChatGPT output. Regulatory complexity both protects patient data and impedes necessary innovation. Disjointed state and federal policies, combined with ongoing data blocking by vendors, create persistent barriers to achieving true interoperability and scalable healthcare analytics. Fragmented privacy laws and ongoing EHR data blocking continue to prevent the seamless interoperability that is crucial for value-based care and AI-driven analytics.

State-specific data laws. Varying regulations, such as CCPA in California and the NY SHIELD Act, complicate multi-state data sharing and compliance.
Prescription Drug Monitoring Program (PDMP) limitations. PDMP integration with EHRs is inconsistent, limiting its usefulness for clinicians.
EHR and data blocking remains prevalent. Despite ONC anti-blocking rules, vendors still make data access costly and cumbersome.
Telehealth licensing barriers. Despite progress post-COVID, state-based licensing restrictions hinder telehealth's full potential.

Claude AI output. The legal framework surrounding healthcare technology creates a complex web of constraints.

Regulatory framework misalignment. Healthcare regulations were largely designed for paper-based processes and later adapted to digital systems. This retrofitting creates persistent misalignments between regulatory requirements and technological capabilities.
Data ownership and control disputes. Unclear legal frameworks regarding healthcare data ownership create persistent barriers to information exchange. Vendors, providers, and patients all claim different levels of ownership and control rights over the same information.
Security and privacy implementation extremes. Fear of HIPAA violations drives extreme risk aversion in technology decisions. Organizations frequently choose restrictive implementations that exceed actual compliance requirements, limiting information flow beyond what regulations actually require.
Multi-cloud compliance advantages. Cloud-neutral strategies offer significant advantages in navigating the complex and evolving compliance landscape. Different cloud providers maintain varying certifications and regional capabilities, allowing organizations to match workloads with appropriate compliance environments.

Most concerning is how current regulatory frameworks fail to address the security vulnerabilities inherent in legacy systems. HIPAA's security rule requirements were largely written before ransomware became healthcare's most significant threat, creating a compliance framework that does not adequately address today's most pressing security challenges.

Environmental factors (E)

ChatGPT output. Legacy healthcare IT infrastructure contributes to inefficiency, waste, and sustainability challenges. Data centers running on outdated hardware, reliance on physical storage media, and supply chain vulnerabilities hinder the industry's ability to transition to scalable, cloud-native solutions. Wasteful practices, like the continued use of CDs and DVDs for imaging, underscore the need for modern, cloud-based solutions that are both environmentally and operationally superior.

Reliance on on-premises data centers. Many hospitals still use inefficient, legacy IT infrastructure rather than adopting cloud-based solutions.
Medical waste from legacy storage devices. CDs and DVDs remain widely used for imaging transfers, contributing to unnecessary waste.
Supply chain vulnerabilities. The dependency on international components for medical IT systems exposes the industry to disruptions.

Claude AI output. The environmental implications of healthcare's legacy infrastructure deserve greater attention.

Paper-dependent workflows. Despite decades of digitization efforts, the average hospital still uses 29.5 tons of paper annually. This persistent paper dependence reflects the limitations of current digital infrastructure to fully support clinical and administrative workflows.
Duplicate testing environmental impact. The inability to efficiently share diagnostic test results across organizations leads to significant test duplication. Beyond the financial waste, this redundancy has substantial environmental impacts through increased resource consumption and medical waste generation.
Disaster recovery vulnerabilities. Legacy systems often lack robust disaster recovery capabilities, creating significant vulnerabilities to climate-related disruptions, an increasingly concerning gap as extreme weather events become more common.
Cloud resource optimization. Cloud-neutral strategies enable organizations to optimize workload placement for energy efficiency, potentially reducing the carbon footprint of healthcare operations. Different providers maintain varying commitments to renewable energy and efficiency, allowing environmentally conscious deployment decisions.

The healthcare industry's technological conservatism creates a substantial environmental burden that receives insufficient attention in sustainability discussions. Modern, cloud-based infrastructure would significantly reduce healthcare's carbon footprint while simultaneously improving system resilience.

Strategic recommendations

ChatGPT output.

Modernize payer IT systems. Transition from COBOL-based systems to cloud-native, API-driven solutions. Establish federal incentives for payers to upgrade infrastructure.

Mandate true interoperability beyond EHR vendors. Strengthen FHIR enforcement and require open API access. Penalize continued data blocking by major vendors.

Accelerate AI readiness via structured data standardization. Require standardized DICOM imaging sharing to improve AI diagnostics longitudinally. Mandate structured clinical documentation beyond free-text notes.

Phase out fax communication in healthcare. Establish a five-year federal phase-out plan for fax-based medical communication. Require adoption of secure digital alternatives with financial support for small providers.

Leverage cloud-based imaging and AI solutions. Encourage centralized, vendor-neutral cloud access for imaging, such as PACS 2.0. Prohibit the continued use of CDs and DVDs for imaging transfer.

Establish federal digital infrastructure grants. Create funding programs to assist small and mid-sized hospitals with IT modernization. Incentivize the adoption of vendor-neutral cloud-based solutions.

Strengthen data exchange across the healthcare ecosystem. Improve pharmacy data integration beyond claims data to include clinical interventions. Ensure PDMPs are interoperable with EHRs to enhance prescribing decisions. Expand HIE participation by enforcing real-time data-sharing standards.

Claude AI output. Addressing healthcare's foundational infrastructure challenges requires bold interventions across multiple domains.

Foundational infrastructure modernization

National image exchange network. Create a federally supported national platform for DICOM image exchange that eliminates CD and DVD-based transfers while ensuring appropriate privacy protections.
Pharmacy data integration initiative. Develop technical standards and financial incentives for integrating community pharmacy data with broader clinical information systems, leveraging pharmacists as clinical information sources.
Legacy system retirement program. Establish a Medicare Advantage bonus payment structure for payers that fully transition from COBOL-based systems to modern, API-driven claims processing platforms.
Cloud-neutral reference architectures. Develop healthcare-specific reference architectures that enable organizations to implement cloud solutions while avoiding single-provider lock-in through containerization and microservices approaches.

Regulatory and policy reform

Meaningful information blocking enforcement. Implement substantial financial penalties for EHR vendors and healthcare organizations that impede data exchange through excessive fees or technical barriers.
Data exchange certification enhancement. Replace the current self-attestation model for interoperability certification with rigorous testing requirements and real-world performance measurements.
Federal HIE architecture standards. Develop national technical standards for HIE implementation that ensure consistent functionality and interoperability across state and regional boundaries.

Economic incentive realignment

Infrastructure modernization tax credits. Create targeted tax incentives for healthcare organizations investing in foundational infrastructure modernization, with enhanced benefits for small and rural providers.
Value-based payment tie-ins. Explicitly incorporate infrastructure modernization requirements into value-based payment models, creating direct financial incentives for eliminating legacy systems.
Interoperability performance bonuses. Implement Medicare payment bonuses for organizations demonstrating exceptional performance on objective interoperability metrics.

LLM conclusions

ChatGPT output. The U.S. healthcare system must first address its fundamental data exchange challenges before AI can meaningfully improve efficiency and care outcomes. A vendor-neutral cloud strategy is crucial to breaking down data silos and enhancing interoperability. By modernizing legacy IT infrastructure, enforcing real interoperability, and adopting standardized, structured data, the healthcare ecosystem can transition from a fragmented, outdated system to a streamlined, data-driven future. AI will be transformative, but only if it is built on a foundation of accessible, high-quality data.

Claude AI output. The triggering observation that we must address fundamental infrastructure issues before pursuing AI advancements reflects an essential truth. No amount of algorithmic sophistication can overcome the limitations of fragmented, inaccessible, and low-quality data. Healthcare's fascination with cutting-edge AI while maintaining fax machines and COBOL systems represents a profound disconnection between aspirations and foundations.

The path forward requires acknowledging an uncomfortable reality. Incremental approaches have failed to drive meaningful change in healthcare infrastructure. After two decades of digitization efforts, the industry still struggles with basic data exchange while maintenance costs for legacy systems consume an ever-larger portion of limited IT budgets.

Transforming healthcare's infrastructure will require coordinated action across political, economic, social, technological, legal, and environmental domains. No single policy, technology, or market force alone can overcome the deeply entrenched barriers to modernization. Cloud-neutral strategies provide a crucial pathway for this transformation, enabling healthcare organizations to leverage modern capabilities while maintaining the flexibility needed in an uncertain regulatory and technological landscape.

The promise of technology to transform healthcare remains powerful, but fulfilling this promise requires first addressing a fundamental question. Is there any other industry more tightly holding onto old technology than healthcare? Until we confront this reality, healthcare will continue to invest in advanced technologies built atop a foundation of sand, impressive in concept but ultimately unsustainable in practice.

Some thoughts and takeaways of my own

We can all likely agree that healthcare is complex. However, if we look beneath the surface of the current AI buzz, I think many would concur that AI alone cannot and will not solve healthcare's inefficiencies, especially if our foundational infrastructure remains outdated. Interestingly, both LLMs, ChatGPT 4.0 and Claude Sonnet 3.7, landed on the same insight. Before AI can deliver scalable, transformative impact, we must start with, or at least not ignore, the basics.

As healthcare leaders, it is inspiring to push boundaries and envision futures in medicine, wellness, health, and care. Equally important to that vision is truly seeing the world around us with open eyes, inclusive of the less glamorous work of modernizing foundational IT systems, improving HIE connectivity, and enabling structured data capture to aim for modernized clinical and business process interoperability. In this case, maybe AI itself is offering us a tough pill to swallow called reality, where before or while we automate, we must invest in the infrastructure that can actually support innovation. We need to promote rising tides so as to lift all ships.

I want to recognize Dr. Robert Pearl, whose early Thursday morning post woke me before the cup of coffee, inspiring me to return to my business toolbox for clinicians, learnings honed during long nights at The Ohio State University within the Fisher College of Business, after long days in the Cardiothoracic ICU intermingled with operational and administrative meetings focused on transforming care delivery. We all have landmark texts and teachings that serve as beacons moving forward. Some of mine in business include frameworks like Porter's Five Forces, Aguilar's PESTLE, Kotter's Leading Change, and Cialdini's principles of influence, all of which remind us of one thing.

Strategic thinking should not be abstract. It should be a practical lens through which we solve real-world problems.

Often, the subtleties so easily overlooked are where the real work begins.

Technology is amazing. But at the end of the day, humanism is what we are aiming for on this path to improvement, because both the problem and the solution are on us.

All views, analyses, and services presented here reflect my independent professional perspective, informed by more than two decades of real-world experience, and do not represent the views or positions of any current or former employer or affiliated organization.