Amazon Tried to Build Healthcare Twice. Then It Started Writing Checks.

In January 2018, three of the most powerful companies in America stood on a stage and promised to fix healthcare. Amazon, Berkshire Hathaway, and JPMorgan Chase announced a not-for-profit joint venture to lower costs for their combined U.S. workforce, and the mere announcement was treated as an earthquake.⁶ Three years later, the venture - named Haven - was disbanded with 57 employees and almost nothing shipped.⁷ Amazon's second attempt, a build-it-yourself clinic service called Amazon Care, lasted barely longer before being shut down too.⁸ Then something changed. Amazon stopped trying to invent healthcare and started buying it - roughly $4.65 billion of it.

The official story is that Amazon kept failing in healthcare. The truer story is that Amazon kept failing at one thing - building healthcare from scratch - and then quietly switched to a different thing it is very good at: acquiring the parts that already work and wiring them into Prime.

Two ventures Amazon built itself, and both died

Haven looked unstoppable on paper. Combine the world's best logistics company, the world's most patient capital, and one of the world's largest banks, point them at a market everyone agrees is broken, and surely something breaks loose. Nothing did. The structural flaw was hiding in the founders themselves: each of the three companies ended up running its own healthcare projects separately, with its own employees, which left the shared venture with nothing distinctive to do.⁷ You cannot disrupt an industry as a side project that your own parents keep undercutting. Haven was a coalition that obviated itself.

Amazon Care was the second swing - a clinic-and-telehealth service Amazon built and sold to enterprises. It met the same wall from a different angle. When Amazon pulled the plug, the explanation came straight from the executive in charge: it was 'not a complete enough offering for the large enterprise customers we have been targeting, and wasn't going to work long-term.'⁸ Note the tell. The shutdown landed only about a month after Amazon agreed to buy One Medical.³ The product wasn't simply weak; it was redundant. Amazon had already decided to buy the thing it had been trying to build.

The pivot hiding inside the failures

Here is the thesis a smart friend can repeat at dinner: Amazon's healthcare strategy isn't a string of flops - it's a disciplined retreat from build to buy. The pattern is unmistakable once you line up the dates. The first acquisition, PillPack, came in 2018 - the same year Haven launched - and turned into Amazon Pharmacy.² When the home-grown bets sputtered, Amazon doubled down not on its own engineering but on its checkbook, agreeing in 2022 to acquire One Medical, a membership-based primary care provider, in an all-cash deal valued at $3.9 billion.³ One Medical was not a science project. It was hundreds of working clinics with paying members, ready to be bundled into Prime on day one.

Why buy what you could build? Because the asset Amazon needs in healthcare is the one thing it cannot manufacture by hiring engineers: a functioning, regulated, physician-staffed care delivery network with patient trust already baked in. Logistics genius moves pills. It does not conjure a doctor a patient already likes. Haven proved that a brilliant coalition with no operating clinic is just a press release. One Medical proved the opposite - that a real, running clinic is exactly the kind of market-ready infrastructure Amazon can absorb. The lesson Amazon learned, expensively, was where its edge stops.

Why this still isn't the disruption everyone fears

Even a disciplined buy strategy can stop short of a coherent one, and Amazon's does. Stack what it now owns - an online pharmacy that ships pre-sorted doses to your door, and a chain of primary care clinics you can fold into a Prime membership.²³ That is a genuinely better front door to care. But a front door is not a house. The pieces that actually move the U.S. healthcare cost curve - the insurance plans, the reimbursement contracts with payers, the deep clinical integration with hospitals and specialists - are precisely the rails Amazon has not bought and cannot easily build. What it has assembled is a premium consumer add-on for people who already have good coverage, not a systemic replacement for the people who don't.

The regulatory record underlines the limits. The FTC issued a Second Request on the One Medical deal, extending the mandatory waiting period and signaling real scrutiny of Amazon swallowing another slice of the care economy.⁴ The agency ultimately declined to sue before the deal closed in February 2023, but the friction is the point.⁵ Every additional rail Amazon would need to become a true disruptor - insurance, payer contracts, hospital integration - is a rail that draws antitrust attention precisely because Amazon is the one trying to own it.

The fair objection is that this reads the failures too kindly - that Haven and Amazon Care were simply botched, and the acquisitions are a separate, lucky second act. But the timeline argues otherwise. PillPack was bought the same year Haven launched, and Amazon Care was killed a month after the One Medical deal was signed.³⁸ These weren't sequential mistakes; they were a company watching its own homemade ventures stall and reallocating capital toward what it could acquire. The honest counter cuts the other way too: even read generously, Amazon has bought a better waiting room, not a new health system. The most expensive lesson it learned was the shape of its own moat - and the most important question it hasn't answered is whether logistics ends at the clinic door.