Payments had no Stripe until it had one, and the difference is instructive. Before, a company that wanted to accept money online assembled a gateway, a processor, a bank relationship, and a compliance operation, and then ran that machinery forever as a cost of doing something else. After, it wrote a few lines of code and treated the whole apparatus as a utility. The value did not disappear; it moved into a layer beneath the businesses that depended on it. The same pattern produced Twilio beneath communications and Plaid beneath consumer finance. In each case a category discovered that it had been forcing every participant to rebuild the same regulated plumbing, and a neutral layer emerged to hold it once.
Care that runs on biosignals is a category still in the "before." A provider who wants to turn a patient's physiological signal into something a clinician can act on — and, increasingly, something a payment model will reconcile — has to solve a long chain of hard problems: capturing the signal from a proliferation of devices, decoding and normalizing it, governing whatever algorithm or model runs on it, holding the patient data compliantly, and delivering the result into a record system as structured, standards-conformant data. Most build a fraction of that chain, bolt the rest together, and maintain the seams by hand. The plumbing is rebuilt at every site, and the rebuilding is not anyone's differentiator.
Biosignal infrastructure is the name for the layer that holds that chain once, so the participants above it do not each rebuild it. Defined plainly, it is the neutral, governed path from any biosignal to the patient: capture across devices, governed execution of algorithms and AI on the signal, custody of the protected data, and delivery into clinical systems as structured FHIR with its provenance intact. It is what sits between the sensor and the chart, and its defining property is that it belongs to none of the parties it connects.
That last property is the whole of it, and it is worth being exact about why. A layer that many parties build on is only trustworthy if it does not compete with them. A device maker's cloud is a fine place to send that device's data and a poor place to send a competitor's. A vertically integrated care company that also holds monitoring data is, to the provider next door, a supplier that may become a rival. Neutrality is not a virtue a walled garden can adopt by policy, because its business is the wall. The neutral layer is structurally available only to a party whose business is the layer — which is why, across categories, the rails end up owned by a Stripe or a Plaid rather than by the largest incumbent that runs on them.
The reader inclined to dismiss this as analogy should look at what is arriving underneath care right now. Nearly half of U.S. adults own a wearable, and a majority own at least one connected health device.[1] The FDA has authorized more than 1,400 AI-enabled medical devices, three-quarters of them in a single specialty, with the count climbing each quarter.[2] Interoperability mandates are pushing structured exchange from aspiration toward obligation, with a national trusted-exchange framework now carrying more than a hundred million documents.[3] And payment is moving to reward outcomes measured from exactly this kind of real-world signal, with roughly 45 percent of U.S. healthcare payments now flowing through models that bear accountability for results.[4] The signal is multiplying, the intelligence that runs on it is multiplying, and the obligation to govern both is hardening into law. A category with those three properties does not stay in the "before" for long.
Why care specifically needs a neutral rail, more than payments or communications did, comes down to what the rail carries. A dropped message is an inconvenience. A biosignal that reaches a clinician stripped of its provenance, or a model that runs on a patient's data with no audit of what it saw or did, is a clinical and regulatory exposure. The layer beneath care therefore cannot be merely a pipe; it has to be a governed pipe — one that holds the data as the accountable party, keeps AI inside a policy envelope, and carries every value with the lineage that lets a downstream decision be trusted. Neutrality makes the layer buildable-upon; governance makes it safe to build upon. Biosignal infrastructure is the category that requires both at once.
A definition written in 2026 should also say what the layer is for right now, because infrastructure is named for its substrate and understood by its era. Payments infrastructure meant checkout before it meant billing and fraud; compute infrastructure meant websites before it meant training runs. Biosignal infrastructure, this decade, means the infrastructure health AI runs on. The actor on the signal used to be a dashboard or a nurse reading a worklist; increasingly it is a model that can speak. And the moment the actor became generative, the governed half of the layer stopped being back-office bookkeeping and became the visible product, because an agent is only as good as what it can see, and only as safe as what checks it before it speaks. The substrate did not change. What runs on it did.
We work in that category because we think it is the one the next decade of care runs on, and because it does not yet have the name that payments and communications already gave their equivalents. Naming it is not a marketing exercise. It is how a provider, a device maker, or a regulator comes to recognize that the thing they keep rebuilding by hand is in fact a layer — and that a layer, once someone holds it neutrally, is something you can simply build on.
Sources
- Rock Health, "Insights on wearables and connected devices from Rock Health's 2025 Consumer Adoption Survey," 2026 — rockhealth.com
- The Imaging Wire, "FDA Updates AI List with New Clearances" (1,451 AI-enabled devices through Dec. 31, 2025), March 11, 2026 — theimagingwire.com
- The Sequoia Project (TEFCA RCE), "New Designated QHIN, New SOPs, and More," Dec. 4, 2025 — rce.sequoiaproject.org
- AHIP, "New Survey Demonstrates Health Plans' Continued Commitment to Value-Based Care Models" (CY2024 APM data), Feb. 2, 2026 — ahip.org
