Five of Epic's competitors just refused discovery in the Texas antitrust case — in unison — and the court is now assembling something that has never existed: an under-oath map of how EHR data control actually works. The public v0 of that map is already sitting in a CMS file: every hospital that attested to the Promoting Interoperability program and every certified product it reported. Pick a vendor, hover a state, and drag the sample-size slider to watch small-state "dominance" dissolve.
4,593 hospitals attested to CMS's Promoting Interoperability program in 2023 and named the certified EHR technology they run. Epic alone was reported by 40% of them; Epic plus Oracle Cerner covers 62 cents of every hospital dollar's worth of buildings — and that's the undercount, because this file weighs a 25-bed critical-access hospital the same as an 800-bed academic center.
4,593
hospitals attesting, PY2023
40.0%
reported Epic (1,836 hospitals)
61.7%
Epic + Oracle Cerner combined
34/56
states + territories where Epic is the #1 vendor
The explorer
Every state, one dot — concentration vs. market size
Each dot is a state (or territory). Horizontal position: how many hospitals attested there (log scale). Vertical: the share reporting the selected vendor. Red dots are states where that vendor is #1. Hover for the full vendor breakdown; the slider hides small markets so you can see which "dominance" is real and which is an artifact of counting seven hospitals.
Share of attesting hospitals by state · EPIC
min hospitals/state≥ 1 · 56 shown
–
national share of attesting hospitals
–
states/territories where #1
–
highest-share state (in view)
–
hospitals reporting this vendor
Small-n check: at ≥1, Delaware shows Epic at 57% — of seven hospitals. Drag the slider to ≥50 and the outliers that survive are the real story: Wisconsin 73%, Minnesota 80%. That's what home-field advantage looks like in the vendor's own backyard.
The national stack — who the 4,593 hospitals report
Shares overlap slightly: a hospital can report certified products from more than one developer (e.g., an Epic hospital running a Cerner lab module), so the seven tracked vendors plus "other/none" are counted per-hospital, not per-product.
The Texas paradox: the state doing the suing is one of Epic's weakest markets by building count — 97 of 389 attesting Texas hospitals (25%) report Epic, behind its national share, with MEDITECH (76) and Oracle Cerner (69) close behind. The lawsuit isn't about how many buildings run Epic. It's about what happens to the data of the patients inside them — which is exactly the layer this public file cannot see.
Critical lens
What this map can't show — and why the subpoenas matter
Buildings are not beds, and beds are not records. This file counts each hospital once. Epic's clients skew large — academic centers, big systems — so its share of beds and patient records runs well above its 40% share of buildings. A concentration map weighted by patients would be redder. Nobody publishes that map; discovery might.
It's a census of the compliant. Only hospitals attesting to CMS Promoting Interoperability appear — 4,593 of roughly 6,000 US hospitals. The missing ~1,400 are disproportionately small, rural, or non-participating, so the small-vendor tail here is an undercount. And it's all self-reported, for program year 2023.
Market share was never the question. Five subpoenaed competitors — Oracle, athenahealth, eClinicalWorks, Palantir, Baylor Scott & White — each independently refused to produce documents until Epic's in-house lawyers were walled off. What they're afraid of isn't visible in any share chart: it's the norms of data access between competitors, set unilaterally, that this file's cheerful "Y" in the interoperability column says nothing about.
The public record can tell you who runs which EHR where. It cannot tell you who controls the data behind the door — the thing every builder's SMART-on-FHIR access, app-store economics, and integration timeline actually rides on. That map is being assembled right now, under subpoena, in a Texas courtroom. The OCEO hearing on August 7 decides how much of it you get to read.
clinicians.dev · an experiment by clinicians.build
Data: CMS Promoting Interoperability Program hospital attestations, PY2023: mimi_ws_1.healthit.chit_hpip, extract mimi_src_file_date=2025-05-01; MimiLabs query 6a5f19e952c62e70720f131f, run 2026-07-21. Vendor = certified-product developer reported by the hospital; hospitals may report multiple vendors, and this is not bed-, patient-, or record-weighted market share.
⚠︎ AI-generated · not reviewed by a human · verify against the linked sources before relying on it