Amazon's Healthcare Pivot: What Haven's Failure and One Medical Reveal About Self-Insured Strategy

Amazon's Healthcare Experiment: The Real Lessons

Amazon spent three years trying to build its own healthcare delivery system through Haven, the joint venture with Berkshire Hathaway and JPMorgan Chase launched in 2018. By March 2021, Haven shut down. The company had burned through millions and produced zero cost savings for its 1.3 million combined employees.

Two years later, Amazon took a different approach: it acquired One Medical for $3.49 billion. This pivot reveals something crucial for self-insured employers managing rising healthcare costs.

Haven failed because vertical integration—owning clinics, hiring doctors, building infrastructure—doesn't solve the core problem: employers can't negotiate primary care prices low enough to offset the structural overhead. One Medical works differently. It's a membership-based primary care platform that negotiates with existing health systems rather than replacing them.

The distinction matters for CFOs and benefits leaders. Here's why.

Why Haven Failed: The Vertical Integration Trap

Haven's model assumed that owning the delivery system would create leverage. The joint venture would open corporate clinics, direct employees there, control costs through scale. This approach had intuitive appeal—eliminate middlemen, own the value chain.

The reality was different:

• Primary care penetration was low. Most employees never visited Haven clinics. Even heavily marketed programs see 15-25% utilization among eligible populations.
• Infrastructure costs were fixed. Once clinics opened, costs didn't decrease if utilization stayed flat. Haven couldn't scale efficiently.
• Negotiating leverage didn't materialize. Haven had no ability to steer specialist referrals or control downstream costs. Owning primary care alone doesn't reduce emergency department use or hospitalization rates at meaningful scale.
• Market consolidation resisted integration. Health systems and hospital networks—especially in major metro areas where Amazon has concentrated employee populations—didn't cede leverage to a corporate employer player.

By 2021, Haven had invested heavily with minimal return. The pandemic accelerated the shutdown, but the business model was already broken.

One Medical's Different Bet

One Medical operates on a membership model: employers pay per employee per month (typically $150-200/member/month on top of medical insurance premiums), and members get access to same-day or next-day primary care, virtual visits, and care coordination.

This model doesn't try to own the entire delivery system. It layers on top of existing insurance. One Medical:

• Negotiates with specialists and hospitals for referrals (Amazon's bargaining power helps here)
• Focuses on primary care as a gateway to reduce unnecessary specialty referrals and ER visits
• Charges a transparent membership fee separate from medical claims
• Operates in roughly 200 U.S. markets with existing networks

For Amazon, the acquisition makes sense. The company has bargaining power to:

1. Negotiate better referral rates for One Medical members into specialist networks
2. Steer 1.3 million employees onto a platform with known utilization rates
3. Use claims data from One Medical members to identify cost reduction opportunities

But this is not vertical integration. It's horizontal consolidation: Amazon is buying distribution and primary care access, not replacing the healthcare system.

What Self-Insured Employers Actually Learn

The shift from Haven to One Medical has three takeaways for your organization:

1. Direct Contracting Beats Ownership

Owning clinics or health systems creates fixed costs and coordination problems. Direct contracting—negotiating rates with existing providers—reduces costs without the infrastructure burden.

Self-insured employers should prioritize direct contracts with:

• Regional primary care networks (often 20-30% discounts off fee-for-service)
• Specialist centers of excellence (40-60% savings on knee, hip, spine, and cardiac procedures)
• Urgent care and telehealth partners (15-25% reduction in ER utilization)

You don't need to own these providers. You need contractual leverage and member incentives to use them.

2. Primary Care as Cost Control, Not Revenue

Haven treated primary care as something to own. One Medical treats it as a lever for utilization management.

The data supports this:

• Patients with established primary care relationships use the ER 30-40% less frequently
• Care coordination for high-cost chronic conditions (diabetes, heart disease, COPD) reduces annual medical costs by 15-20% per managed patient
• Same-day/next-day primary access reduces urgent care utilization by 25-35%

But these savings only materialize if primary care has authority to coordinate downstream care and manage utilization. One Medical's model gives them that—a dedicated care team with time and incentives to coordinate.

Your direct contracting strategy should include embedded care coordination in primary care networks. This isn't just access. It's accountability for cost reduction.

3. Membership Models Beat Insurance-Only Models

One Medical's separation of membership fees from medical claims creates alignment. Members pay upfront for access; the provider has incentive to keep them healthy and out of high-cost settings.

Self-insured employers should consider:

• Concierge or membership-based primary care networks ($100-200/member/month)
• Integrated occupational health and primary care (especially for large employer populations)
• Voluntary health coaching and preventive programs with financial incentives

These models work because they're transparent and separate accountability for primary care from claims processing.

Bottom Line

Amazon didn't learn that employers should own healthcare systems. It learned that employers with scale should directly contract with high-performing primary care networks and use that primary care layer to manage downstream utilization.

If you're self-insured, you don't need to build clinics or hire doctors. You need direct contracts with primary care networks, claims data visibility to identify cost drivers, and care coordination focused on your highest-cost populations.

That's the lesson from Haven's failure and One Medical's acquisition. Apply it.