In early July, the CEO of Palantir sat in a CNBC studio and grew agitated enough that the anchor paused to tell him he sounded angry. Alex Karp had come to discuss a partnership with Nvidia. What he delivered instead was closer to a sermon. The serious customers, he argued, want to own it all, the compute, the models, the data, the edge, rather than rent their future from anyone.
It is tempting to dismiss this as a vendor talking his book, and it partly was. Some of his sharpest jabs at the AI labs are contested. But the core of it deserves a hearing, because it lands at a peculiar moment for one of the least glamorous corners of American health care.
Fifty Billion Dollars, and a Five-Year Clock
Rural health is about to receive the largest infusion of money in its history. The Rural Health Transformation Program will allocate $50 billion to the states over five years, and the first funding opportunities are opening this summer, with some deadlines measured in weeks. The program's stated ambitions are broad: prevention, workforce, new care models, and technology. Its unstated deadline is not. In 2030, the money stops, and everything built with it will have to keep working in communities that could not afford it the first time.
The history of one-time money in poor places follows a familiar arc. The grant arrives; the consultants arrive with it; systems are purchased; and five years later, the subscriptions lapse and the capacity quietly evaporates. The federal government has now added a twist: states must document results annually to keep the funds flowing. The reporting burden lands hardest on the smallest institutions, the county health department, the critical access hospital, and the clinic with one IT person who also fixes the printers.
Why Renting Fails Rural Health
And yet rural health may need artificial intelligence more than any sector in the country, because its defining problem is capacity. More than a hundred rural hospitals have closed in the past decade. The communities that remain are older, sicker, and farther from care, served by clinicians stretched across impossible geography and administrators drowning in reporting requirements written for institutions ten times their size. You cannot recruit your way out of that arithmetic. AI is the first technology to address arithmetic itself. It can draft the grant report, close the referral loop, watch the chronic disease registry, and hand the community health worker a manageable day, which is to say it can return hours to a workforce that has none to spare. For rural health, AI is less a luxury than a lifeline. Which is precisely why it matters who controls it.
The record of adoption so far suggests how matters as much as whether. MIT researchers found that ninety-five percent of corporate AI pilots produced no measurable return. What distinguished the exceptions was unglamorous. They treated the model as one component among many. They kept their data where they could see it. They retained the freedom to switch models, an option they rarely used but refused to surrender, and they kept a person accountable at the end of every automated process.
Ownership Has a Name: Sovereignty
There is a name for this posture. People have started calling it AI sovereignty, and it is less a technology than a stance. The sovereign organization uses the best models available and signs away control of none of them. This is not a quarrel with the model makers. Their work is remarkable, and the best of them will even let you run their models inside your own walls. It is a quarrel with a default, the arrangement in which your data, your context, and your hard-won judgment quietly come to live on someone else's terms. The distinction from privacy is worth drawing. Privacy concerns about who may see your data. Sovereignty concerns who shapes how your institution thinks, decides, and remembers. A rural health system that pours its patient records and its accumulated judgment into a platform it does not control has traded away something it cannot easily buy back.
The Five Tenets of Rural Health Sovereignty
Sovereignty is not complicated, and it does not take a technologist to demand it. It comes down to five things a state can own and include in every contract.
- Own the data. Your records, and everything an AI learns from them, stay in your hands, on terms no vendor can change.
- Own the choice of model. Use the best AI for each job and stay free to switch, so no single company holds the leverage.
- Own the integrations. The system integrates with what you already run, so your tools work together rather than trapping you in one.
- Own the decisions. A person stays accountable for every automated call. No black boxes.
- Own what remains. The capability outlasts the grant, because sustainability is the whole point.
The Choice Every State Makes This Quarter
What makes this moment unusual is that the powerful have already made their choice. The Pentagon's software partner will not surrender control of its models. The sophisticated enterprises treat vendor lock-in as an existential risk. The only open question is whether the institutions with the least margin for error, the ones holding the medical records of sixty million rural Americans, will be offered the same terms, and whether they will know to ask.
This is the work we do at Tapp Network with more than 24 state agencies and 1,000 organizations. We built the Rural Health Transformation Operating System, announced at the Rural Health Summit with the Rural Health Coalition, to show it can be done. Five years from now, the money will be gone. Someone will walk into a county health office and ask what remains. The answer should be everything. The system, the data, and the knowledge are still there, still working, still owned by the communities it serves. Learn more about RHT reporting.
