The Centers for Medicare and Medicaid Services has spent the past year building something it calls the Health Tech Ecosystem, and this week it took stock of the progress, laying out the categories of commitment companies can make and pointing to the more than 700 organizations that have signed on. The commitments are called pledges, and they are voluntary. Companies promise, in a few sentences each, to make patient health data more accessible and interoperable.
The obvious question is why a voluntary pledge would succeed where two decades of mandates and tens of billions of federal dollars largely did not. Interoperability, the ability of one health system's records to talk to another's, is one of the oldest unsolved problems in American healthcare, and the government has been trying to force it since 2009. That history is the essential context for judging whether this new, gentler approach is a smart pivot or a hopeful repackaging of a problem that has defeated every prior attempt.
Why the records still do not talk to each other
Start with the problem, because its persistence is the whole point. When you see a new doctor, your records often do not follow you. A specialist may not see what your primary-care physician ordered; a hospital may not have the scan another hospital took last month. Patients fill out the same clipboard forms over and over because the systems holding their information cannot, or will not, share it.
This is not for lack of trying. The 2009 HITECH Act spent roughly $35 billion pushing hospitals and doctors to adopt electronic health records, and it succeeded at digitizing records while largely failing to connect them. The country ended up with electronic records that sit in incompatible silos, which in some ways is worse than paper, because it created an illusion of modernization without the interoperability that was the actual goal. Subsequent rules under the 21st Century Cures Act took aim at "information blocking," the practice of deliberately impeding data sharing, with regulation and penalties.
So the federal government has already tried the mandate approach, twice, with real money and real enforcement behind it, and the records still do not flow the way patients assume they do. That is the track record against which a voluntary pledge has to be evaluated.
Why the data does not want to move
To judge whether pledges can work, you have to understand why interoperability failed under mandates, and the reasons are structural rather than technical. The technology to share data has existed for years. The barriers are about incentives.
For much of the industry, keeping data siloed is good business. If a hospital system's records do not easily transfer, patients are stickier, because leaving means starting over, and competitors cannot easily poach them. Electronic-health-record vendors have historically built systems that work smoothly within their own ecosystem and awkwardly with rivals, because friction at the boundary protects their market. Data has been a competitive moat, and asking companies to share it is asking them to lower a wall they built on purpose.
This is why the problem resisted mandates. When the underlying incentive rewards hoarding, regulation produces minimal compliance, meeting the letter of the rule while preserving the practical friction. Companies did the bare minimum the law required and no more, because doing more meant surrendering an advantage. Any approach that does not change that incentive, pledge or mandate, runs into the same wall.
The case that a pledge could work anyway
Given that history, there is a serious argument for why the voluntary approach might succeed where force did not, and it is worth taking seriously rather than dismissing.
The logic is that a mandate produces resistance and minimal compliance, while a voluntary framework that companies help design can produce buy-in. When the industry shapes the standards, the standards are more likely to be practical, and companies that co-authored them are more invested in making them work than in evading them. The pledge approach also uses reputational competition as leverage: publishing which companies have joined, and eventually which have delivered, creates pressure to participate and to be seen delivering, so that non-participation becomes a competitive liability rather than a safe default. If enough major players join, the holdouts look like the ones blocking patients' access to their own data, which is a bad position to defend publicly.
There is also a genuine technological shift that makes this moment different from 2009. AI assistants and modern data-exchange standards like FHIR have matured to where bridging incompatible systems is more feasible than it was, and the current initiative leans on both. The tools available now are better suited to the problem than the ones the HITECH era had to work with, which is a real reason, independent of the policy design, that progress could come faster this time.
Why the skepticism is also warranted
The doubts are equally grounded, and they center on the nature of a pledge. Each commitment is two to five sentences and largely aspirational and flexible, by the description of lawyers tracking the program. A representative one has patient-facing app companies pledging to connect to aligned networks and, with patient consent, securely access relevant data to deliver personalized support. That is a statement of intent, not a binding, measurable, enforceable obligation.
The concern follows directly. A voluntary pledge with soft language and no penalty for non-delivery is easy to sign and easy to let slide. Seven hundred organizations signing on is impressive as a headline and tells you little about how many will actually change their systems, because signing costs nothing and delivers immediate reputational benefit, while delivering requires real engineering, real money, and, in some cases, surrendering the competitive advantage that data hoarding provides. The gap between pledging and delivering is exactly where the initiative will be judged, and pledges are structurally weakest at closing it, because they lack the one thing a mandate has, which is consequences for doing nothing.
The deeper worry is that the pledge does not resolve the incentive problem at the root of the failure. If sharing data still costs a company its competitive moat, a voluntary promise gives it even less reason to follow through than a mandate did, since there is no penalty at all. The initiative is betting that reputational pressure and industry buy-in substitute for enforcement. Whether they do is genuinely unknown, and the honest position is that it has not been demonstrated yet.
What would actually tell you it is working
The way to judge this is to ignore the pledge count, which measures intent, and watch for measures of delivery, which is what has always been missing.
The signals that would indicate real progress are concrete. Whether the companies that hit the early "minimum viable product" deadlines actually move patient data between rival systems, not just within their own. Whether patients report the friction easing, fewer repeated forms, records that follow them, apps that can pull their full history with consent. Whether the reputational mechanism develops teeth, meaning CMS publishes not just who joined but who delivered and who did not, so that the list becomes a scoreboard rather than a guest book. And whether the major EHR vendors, whose cooperation is decisive because they hold most of the records, actually open their systems in ways that cost them their stickiness advantage, which is the truest test because it is the one that runs against their own interest.
Until those show up, the pledge count is a measure of willingness to be associated with a popular goal, not evidence the goal is being reached.
How to read it
The fair synthesis is that this is a reasonable strategy aimed at a genuinely hard and genuinely important problem, and that its central bet is unproven. Interoperability matters, patients are real losers under the current fragmentation, and trying a collaborative approach after mandates underdelivered is a defensible response rather than a naive one, especially with better technology now available. There is a real chance the combination of industry buy-in, reputational competition, and matured tools accomplishes more than regulation did.
There is also a real chance it produces a large number of signed pledges and modest actual change, because the incentive to keep data siloed, the thing that defeated the mandates, is untouched by a voluntary promise and arguably weakened by the absence of any penalty. Both outcomes are live, and which one arrives depends on execution and enforcement details that the pledge language itself leaves open.
The useful posture is patient skepticism. The goal is worth pursuing, the approach is worth trying, and the announcements so far measure enthusiasm rather than results. The number that matters is not how many companies pledged. It is how many patients stop filling out the same clipboard twice, and that number will not be visible for a while yet. This week's stocktaking is a progress report on intentions. The progress that counts is still ahead, and the history of this particular problem counsels holding applause until the records actually move.
Primary sources
- CMS's Health Technology Ecosystem overview and Early Adopters pages for the initiative's goals, the pledge categories, and the framing that 15 years of regulation had not delivered a connected experience.
- CMS's April 2026 press release for the First Wave launch, the Medicare App Library, and the figure of more than 700 pledging organizations.
- Hogan Lovells for the description of pledges as two-to-five-sentence, largely aspirational and flexible commitments, and the May 2026 addition of the Electronic Prior Authorization category.
- Wilson Sonsini for the July 2025 program launch, the four ecosystem objectives, and the first-quarter-2026 framework targets.
- The American Hospital Association for the March 31 minimum-viable-product deadline; background on the 2009 HITECH Act's roughly $35 billion in EHR incentive spending and the 21st Century Cures Act information-blocking provisions reflects widely published health-policy history.