Will Manidis: legacy healthcare software companies — not startups — will win the AI adoption race
Mar 7, 2025 with Will Manidis
Key Points
- Legacy healthcare software companies with entrenched install bases and decades of switching costs will dominate AI adoption in medicine, not startups, because hospital administrators—not physicians—control procurement.
- Manidis predicts a wave of AI healthcare startups will fail over the next six months as investors chase thinly differentiated tools while model costs commoditize toward open-source alternatives like Llama.
- For engineers, joining legacy EMR companies to drive AI adoption from within offers better returns than founding new healthcare AI startups as the value shifts toward incumbents with existing relationships and data.
Summary
Will Manidis — investor, former startup founder, and current employee at a large publicly traded EMR company — makes a contrarian case for where AI in healthcare will actually compound: not in startups, but in legacy software businesses that already own the install base.
The argument is straightforward. Ambient scribes like those from Abridge are a front door to the electronic medical record, not a full software suite. And unlike most enterprise software markets, the person who uses healthcare software — the physician — is not the person who buys it. Hospital administrators make procurement decisions, which means novel distribution plays that work in consumer or B2B tech simply don't transfer. Manidis expects a wave of AI-enabled healthcare startups over the next six months and predicts most will fail, with some of the more aggressive pill-mill-style operators ending up in legal jeopardy.
Legacy distribution is the most underpriced asset in healthcare right now, in Manidis's view. A company with $140 million in recurring revenue built over decades carries switching costs, trust, and a piloting surface that no new entrant can replicate. The winning play is to take that installed base and layer modern AI on top of it — and he says that's already happening across his company's product suite.
For technically strong engineers, Manidis sees real alpha in joining one of these legacy companies rather than founding yet another AI health startup — working up to a leadership role and driving AI adoption from the inside. As model-layer costs commoditize (he points to locally-run Llama as a pressure on OpenAI's roughly $140 billion valuation), the value shifts toward whoever already has the relationships and the data.
On markets broadly, Manidis is neither bull nor bear. He expects several more 10x moves up, but also several 10x drawdowns along the way, with fragility increasing as venture capital chases thinly differentiated AI tools. He runs net cash, no leverage, no credit cards — and recommends the same to anyone asking.