Two Years (or less) to a Patient: What We Want From Med-Tech

In September I promised a deeper post on each of our four focus areas. This is the one on med-tech, and it's the one I get the most questions about, usually some version of: "You're a health system fund writing $250K to $2M checks. Why would you touch regulated devices?"

It's a fair question. Devices are slower and more capital-hungry than software, and the clearance process is unfamiliar. Here's why we're doing it anyway, and what we need to be true before we say yes.

Why med-tech, why now

Three reasons.

The exit market woke up. EY's Pulse of the MedTech Industry (Oct 2026) put med-tech financing at a record $63B for the year ending June 30, up 87%, and M&A at roughly $130B, the most in over a decade. J.P. Morgan counts $64.1B of med-tech M&A across 141 deals through September. Medtronic paid $585M for CathWorks in February and led a $700M round for Cornerstone Robotics in September. When the big companies are buying, early-stage devices have somewhere to go.

The money isn't reaching the early stage. Most of EY's $63B was IPOs and debt. On the venture side, later-stage rounds took 70% to 87% of each quarter's med-tech dollars this year (NewMarketPitch, Oct 2026), and PitchBook has been flagging a "deterioration of funding possibilities for early-stage medtech companies" since last fall. Seed and Series A device companies are under-fed. That's where we invest.

The regulatory moat is real. The FDA's AI-enabled device list passed 1,614 authorizations by mid-2026, with 335 in 2025 and 181 in the first half of this year, 96% of them through 510(k). A clearance isn't a patent, but it is a head start measured in months that software alone can't buy. In September I said every pure-software company has to show us two of four moats. A regulated device arrives with one of them.

What we look for

Six things. Few devices meet all of them, and that's by design: we want the handful we can actually help.

1. A patient is touched by the product in our service area within 24 months

Our footprint is the Appalachian Highlands: 29 counties across Tennessee, Virginia, North Carolina and Kentucky, most of them rural. We invest because something can be deployed and validated there, and the value we add is tied to that. So we work backward from a Ballad patient two years out:

• 510(k) fits. Median time from receipt to decision was 126 calendar days across 3,197 clearances in 2025 (Rhizome's analysis of FDA records). Most of what we back will be here, or already cleared.
• De Novo fits when the submission is in or close. The FDA granted 27 in 2025 at a median of 299 days, and about a quarter of recent grants took more than a year (Complizen, Oct 2026).
• PMA (Pre-Market Approval) rarely fits. The review goal is 180 FDA days, but with the clinical program the road to market usually runs one to three years, and the capital need is beyond our check. We'd look at a post-pivotal PMA with a syndicate that can carry it.

Two things we keep in mind. FDA days pause every time the agency asks a question, so we plan in calendar days. And CDRH's reviewer bench shrank in 2025; De Novo grants fell from 47 to 27. We underwrite to how long things actually take.

2. It reaches people where there is no specialist

The Chartis 2026 Rural Health State of the State report is the backdrop for everything we do. Nationally, 417 rural hospitals are vulnerable to closure and 41% are operating in the red. In Tennessee the share of vulnerable rural hospitals rose from 44% to 61% in a single year. More than 300 rural hospitals have stopped delivering babies, more than 300 have dropped general surgery, and more than 450 no longer offer chemotherapy.

Behind every one of those numbers is a family driving 60 minutes to the nearest cardiologist, or not going. The devices we love bring the point of care, or the specialist's judgment, into that county: a sensor a patient wears at home, a diagnostic a primary-care nurse can run, an AI read that lets one specialist cover six sites.

There's a tailwind here, too. The $50B Rural Health Transformation Program awarded Tennessee $206.9M and Virginia $189.5M for FY26, and technology that improves rural access is exactly what states said they'd fund.

3. It dramatically lowers the cost of care, or lifts outcomes at the same cost

Better isn't enough. We're looking for a step change in one of two directions:

• Cost. The same outcome in a lower-cost setting, with a lower-cost clinician, or in fewer visits. Home instead of hospital, nurse instead of physician, one encounter instead of three.
• Outcome. A clear improvement in a metric our service lines already track, with cost held flat. Earlier detection, fewer readmissions, fewer transfers out.

We also want to see the evidence. A JAMA Health Forum study (Lee et al., Aug 2025) looked at 950 FDA-cleared AI devices: 60 of them accounted for 182 recalls, 43% of those recalls came in the first year, and devices with no reported clinical validation averaged 3.4 recalls each, compared with about two for validated ones. The 510(k) pathway often doesn't require human data. We'll ask for it anyway, and if it doesn't exist yet, our footprint is a good place to generate it together.

4. The people who use it would miss it

Ballad is 22 hospitals, hundreds of sites, one EHR, a supply chain, biomed, IT security and a value analysis committee. A device that needs a new workflow at every site rarely makes it, however good the science.

The AMA's 2026 physician survey matches what we hear in our own hallways: 81% of physicians now use AI in practice, 85% want a say in adoption decisions, and 88% call safety and efficacy validation critical. Clinicians are open, and they want to be part of the choice. So before we invest, we put the device in front of the nurse, tech or physician who'd actually use it and ask two questions: would you miss this if it went away, and what would need to change for you to use it every day?

In practice, low friction means it runs on equipment we already own or adds little, needs one integration rather than five, and can be taught in minutes.

5. A moat beyond the clearance

The clearance is the first moat. We look for a second, and in devices it's usually one of these:

• Data. The device generates a longitudinal dataset competitors can't buy.
• Distribution. You have some secret sauce to get widespread adoption.
• Hardware plus software. A sensor and the algorithm trained on it, where neither works without the other.
• Reimbursement. A CPT code, an NTAP or an established coverage pathway.

6. Other health systems at the table

We'd rather not be the only operator on the cap table. A syndicate that includes other health systems tells us another CMO already said yes, and gives the company a second place to validate and a second logo for the next round. Rock Health's H1 2026 report named strong partnerships and network effects as one of the four things investors now screen for; in a device round, a health-system co-investor list is the clearest version of that signal.

What this leaves

A narrower funnel than most device investors: cleared or near-cleared diagnostics, sensors and AI reads in cardiovascular, cancer, neuroscience, urology, orthopedics, general surgery, and other critical service lines, that move care closer to home, that someone here has already said they'd use, and that other systems are already backing. We think it's also the funnel that produces companies we can genuinely help.

What to bring us

1. Your FDA status in calendar days: where you are, what's left, who's done the pre-sub.
2. The county you'd deploy in first and the clinician who'd run it.
3. The cost or outcome number, and the human data behind it.
4. Your second moat.
5. Who else on your cap table runs a hospital. (we can help make connections if we get excited about your product)