Bernstein: AI will become the most disruptive long-term theme in U.S. healthcare services

Overview

This report is part of Bernstein's global AI series and discusses the potential industry structure, profit pool, and valuation impacts on U.S. healthcare services once AI is fully integrated and commercialized in the early 2030s. The author argues that healthcare is one of the industries most likely to be significantly transformed by AI, because AI can simultaneously affect cost, accessibility, patient experience, clinical efficiency, and payer operations.

Core views

The core view of the report is that AI's impact on healthcare services will be broad and long-lasting, but the benefit pathways will differ across subsectors. Care automation and consumer agents will have the greatest long-term impact, while provider and payer efficiencies are more likely to be realized earlier. Hospitals and care providers can improve EPS and margins by reducing labor cost intensity, increasing clinical capacity, and improving operating efficiency; overall MCO margins may remain stable because part of the efficiency gains may be passed on to customers, but large MCOs may gain share through AI capabilities and consumer agents. UNH is identified as the biggest AI beneficiary, and HCA is also seen benefiting from provider efficiencies and care expansion.

Analysis framework

The report uses a combination of thematic decomposition and profit pool estimation, dividing healthcare AI applications into five categories: care expansion and automation, AI-driven consumer agents, precision medicine, provider efficiencies, and payer automation and efficiencies. It then estimates for each application the spending scope affected, potential profit pool, speed of adoption, beneficiaries, and constraints, and maps these to the healthcare services companies covered by Bernstein.

Methodology notes

• AI industry impact frameworkFive-category healthcare AI application framework

  Care automation, consumer agents, provider efficiencies, payer efficiencies, precision medicine

  The report uses five application categories to break down the sources of value, adoption timeline, and beneficiaries of AI in healthcare services.

• Profit pool estimationScope and Opportunity

  Addressable spending scale and potential profit pool

  Using baselines such as U.S. healthcare spending, provider costs, payer operating expenses, and net specialty drug spending, the report estimates the revenue, savings, or profit pools AI may generate.

• Investment impact rankingCoverage Rank Order

  Ranking of AI beneficiaries among covered companies

  The report ranks UNH as the biggest beneficiary and highlights companies such as HCA and ELV for their benefit pathways in care automation, consumer agents, or provider efficiencies.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• UNH

  One of the biggest AI beneficiaries

  Strengths

  Optum Insights has the opportunity to deliver AI capabilities to providers, and UNH can also benefit in consumer agents, payer efficiencies, and total medical cost management.

  Weaknesses

  Some MCO efficiency gains may be passed on to customers, and competition in consumer agents is intense.

  Comparison

  The report identifies UNH as the most important AI beneficiary in its coverage universe.

  Risks

  Regulation, privacy, algorithmic liability, execution of AI deployment, and the pace of consumer agent adoption.

• HCA

  Key beneficiary on the hospital side

  Strengths

  Provider efficiencies and care expansion can reduce the share of compensation costs, improve margins, and potentially expand care capacity.

  Weaknesses

  Deep clinical workflow automation is slower to implement and depends on EHR integration, data quality, and clinical acceptance.

  Comparison

  Compared with MCOs, hospitals' economic gains from care automation and provider efficiencies are more direct.

  Risks

  Regulatory constraints, clinical liability, resistance from medical staff, and adjustments to compensation structures.

• ELV

  Beneficiary of consumer agents and payer efficiencies

  Strengths

  The report shows ELV's virtual assistant interface, indicating it is already deploying capabilities related to consumer agents.

  Weaknesses

  It needs to prove that AI can provide accurate, trustworthy, and compliant recommendations in highly sensitive healthcare scenarios.

  Comparison

  Similar to UNH, ELV can benefit from MCO consumer entry points and operating automation.

  Risks

  Privacy, liability boundaries, consumer trust, and regulatory scrutiny.

• Hospitals and other care providers

  Major winners at the industry level

  Strengths

  AI can reduce administrative burden, increase clinical capacity, shorten wait times, and improve cost structures.

  Weaknesses

  The investment threshold and data infrastructure requirements are high, and smaller institutions may lag behind.

  Comparison

  The report believes large hospitals and physician organizations are better positioned than smaller competitors to invest early and benefit first.

  Risks

  Workforce reshaping, clinical validation, regulatory approval, and system integration.

• MCO

  Overall more neutral but internally differentiated

  Strengths

  AI can streamline claims, authorization, customer service, care management, and cost estimation, improving NPS and member experience.

  Weaknesses

  Efficiency gains may be passed on to customers, and industry profit margins may not improve significantly.

  Comparison

  Large MCOs with strong technology capabilities may gain share from laggards.

  Risks

  Algorithm transparency, bias control, CMS rules, privacy, and controversies around automated claims processing.

• PBMs

  Moderate beneficiary

  Strengths

  AI can improve claims automation, benefit coordination, pricing, formulary design, and rebate optimization.

  Weaknesses

  Some gains may be offset by margin pressure.

  Comparison

  Compared with hospitals and UNH, PBM benefits are more focused on operating efficiency than structural profit pool expansion.

  Risks

  Regulation, pricing transparency pressure, and payer bargaining.

Key data

• Scope of care automation impactApproximately $2.7 trillion in U.S. physician and hospital care spending; potential opportunity above $100 billionAssumes AI enables 10%-20% more care delivery and lowers the share of clinical compensation costs.
• Scope of consumer agent impactApproximately $4.5 trillion in healthcare spending; potential profit pool of $50 billion to above $100 billionThe report assumes consumer agents can capture 1-2 percentage points of the value from savings, better outcomes, and improved satisfaction.
• Provider efficiency opportunityMore than $500 billion in related administrative, operational, billing, and insurance-related costs; savings rate of about 15%-20%Near-term opportunities include ambient listening, documentation, billing, revenue cycle management, staffing, and inventory management.
• Payer efficiency opportunityPayer operating expenses or SG&A of about $200 billion to $300 billion, of which roughly half falls within AI's application scopeMainly applied to claims, call centers, enrollment, eligibility verification, care management, and self-service.
• Precision medicine opportunityNet specialty drug spending exceeds $200 billion; assumes about a 20% reduction in ineffective medication useAI helps personalize treatment through genomics and testing data, reducing trial-and-error and inefficient drug spending.
• Key ratings and target pricesUNH Outperform, target price $444; HCA Market-Perform, target price $541; ELV Outperform, target price $384The report believes UNH and HCA are the most likely near-term beneficiaries of AI disruption.

Impact & implications

From an investment perspective, AI is more likely to create value first in operating efficiency, document automation, claims processing, customer service, and consumer navigation, before gradually expanding into care extension, semi-automated clinical functions, and long-term automated care. Providers may achieve margin expansion through lower labor costs, higher throughput, and stronger scale advantages; MCOs may use automation and consumer agents to reduce administrative costs, improve experience, and gain share, though industry-wide margin improvement may be offset by customer givebacks.

Risks

• The regulatory framework remains centered on human accountability, licensing, and standards of care, making it difficult for fully autonomous clinical AI to quickly obtain clear liability attribution.
• Medical malpractice liability, explainability of clinical judgment, and informed patient consent may slow the deployment of care automation.
• Data privacy requirements such as HIPAA and AI governance rules will raise compliance costs and lengthen adoption timelines.
• Clinicians and healthcare staff may worry that AI threatens job security and the value of professional training, leading to organizational resistance.
• Patients may distrust AI in high-risk medical decisions in terms of reliability, accuracy, bias, and empathy.
• Insufficient data pipelines, legacy IT systems, and EHR integration among providers and payers may limit end-to-end automation.
• The closer AI gets to clinical decision-making or final authorization decisions, the higher the regulatory and audit risk.
• Precision medicine requires strong clinical validation; otherwise it is difficult to prove that AI can diagnose accurately and improve outcomes.

What to watch

• Whether UNH's Optum Insights can become an AI enablement platform for providers and export capabilities to smaller MCOs, PBMs, and providers.
• Whether hospitals such as HCA can generate measurable margin improvement in documentation, scheduling, imaging, care expansion, and revenue cycle management.
• Whether consumer agents at MCOs such as ELV and UNH can improve member experience, NPS, cost transparency, and care navigation efficiency.
• Whether care automation can gradually expand from narrow-function areas such as radiology to semi-autonomous or autonomous clinical workflows.
• How regulators define the boundaries of AI education, medical advice, clinical decision-making, and liability.
• Whether AI deployment in claims, prior authorization, customer service, and care management by payers can reduce administrative cost ratios.
• Whether differences in AI investment capacity between large and small institutions lead to share shifts.
• Patient and clinician acceptance of AI tools, and whether there is backlash due to bias, errors, or privacy issues.