The Free Lunch Is Over, Except Now It’s Not: What Near-Zero Software Costs Mean for Every Player in Healthcare
Abstract
This essay argues that the collapse of software development costs, driven by AI coding tools, will be one of the most disruptive forces in healthcare over the next five to ten years, arguably more disruptive than any single regulatory change or clinical breakthrough. The implications cut differently across hospitals, payers, pharma, and vendors, but the common thread is that software as a moat is largely dead, and the winners will be those who figured that out early.
Key claims:
- Software development costs are falling 80-90% for many use cases, with agentic coding tools like Cursor, Devin, and GitHub Copilot dramatically compressing build timelines
- For hospitals and health systems, this means internal IT teams become credible builders again, threatening incumbent EHR and middleware vendors
- For payers, it means utilization management, prior auth, and claims adjudication logic can be rebuilt internally at a fraction of historical cost, destabilizing a generation of point solutions
- For pharma, clinical trial software, regulatory submission tooling, and commercial analytics platforms become commoditized, shifting value to data and relationships
- For health tech vendors, any company whose core defensibility was “we built the thing and you can’t” is in serious trouble
- The real winners are those sitting on proprietary data, clinical workflows, and regulatory relationships that software alone cannot replicate
This is not a five-year story, it’s a two-year story
Table of Contents
The Actual Premise: Software is Becoming a Commodity Input
What This Means for Hospitals
What This Means for Payers
What This Means for Pharma
What This Means for Health Tech Vendors
Where the Real Moats Are
What Investors Should Actually Be Doing About This
The Actual Premise: Software is Becoming a Commodity Input
There’s a useful analogy buried somewhere in the history of electricity. Before widespread electrical grids, manufacturers built their own power generation on-site. It was expensive, it required specialized expertise, and it was a legitimate competitive differentiator to have reliable power when your competitor didn’t. Then the grid happened, and power became a utility, and overnight the differentiator evaporated. Nobody today builds a factory and considers their access to electricity a competitive moat.
Software in healthcare has operated for the last thirty years roughly like private power generation. Building it was expensive and slow. A mid-sized health system trying to custom-build a care management platform was looking at multi-year timelines, eight-figure budgets, and a constant risk of the whole thing collapsing when three key engineers left for Google. So instead, everyone bought. They bought Epic. They bought Salesforce. They bought a hundred point solutions for a hundred specific workflows. And the vendors who built those things had real moats, because the switching costs were brutal and the alternative was attempting to rebuild internally, which was essentially impossible at reasonable cost.
That equation is breaking down fast. Tools like GitHub Copilot, Cursor, and the newer agentic coding platforms are compressing development timelines by anywhere from 50 to 90 percent depending on the use case. Some enterprise teams are reporting that work that used to take a senior engineer six weeks is getting done in three days. The models are not perfect, the output requires review, and complex distributed systems still require serious human architecture decisions. But for the enormous category of healthcare software that is essentially business logic wrapped in a UI with some integrations, the cost structure is collapsing.
This matters more in healthcare than almost anywhere else because healthcare is uniquely full of software that is essentially business logic wrapped in a UI with some integrations. Prior auth platforms. Care gap identification tools. Claims repricing engines. Quality reporting dashboards. Population health analytics. Contract modeling tools for value-based care. Virtually every category of health tech point solution that raised a Series A in the last decade is, if you strip away the branding, a set of rules encoded in software running against healthcare data. And encoding rules in software is exactly what these new tools are extraordinarily good at.
The people who built those companies are smart and they know this. The honest ones will tell you privately that they are terrified. The less honest ones are writing blog posts about how AI will create more demand for their platform, which is technically true in some narrow sense and deeply misleading about the structural shift happening underneath them.
What This Means for Hospitals
