Deal count barely moved this half. Average round size did, and that changes what a fundable pitch now has to prove.
US digital health startups closed their strongest first half in three years. That headline number hides an uneven recovery, and the shape of it matters more to a founder raising right now than the total itself.
Growth this half came almost entirely from existing categories getting bigger checks, not new categories breaking through for the first time. That distinction is the whole story.
Round sizes grew even though deal count didn't
Startups raised $7.4B across 244 deals in H1 2026, per Rock Health, up from $6.4B a year earlier, a roughly 15.6% increase. Deal count held almost flat year over year.
Divide it out and average round size rose from around $26.2M to about $30.3M, a 16% jump. That means essentially all of the extra billion dollars went into existing-sized deals getting bigger, rather than more startups getting funded in the first place.
Three companies alone, eMed ($200M), Nourish ($100M) and Midi Health ($100M), pulled in $400M between them, more than 5% of the entire half's funding total concentrated in a single clinical indication.
Clinical workflow software startups captured close to 48% of both deals and capital over the past year, according to one market tracker, making it the clear anchor category of this cycle.
For context, that's roughly the same share of total US digital health funding, $14.2B in 2025, that AI-enabled companies alone captured last year, a 54% share, according to Rock Health, suggesting this concentration pattern isn't new, just intensifying.
"Good but undifferentiated" is now the hardest position
Three companies in one clinical category took over 5% of an entire half-year's digital health funding total. That is not broad-based growth, it is concentration.
This is a barbell market. Capital is flowing to startups with category-defining scale, like the clinical workflow leaders, or to startups with a genuinely defended niche, like the weight management platforms riding Medicare's new GLP-1 economics, covered later in this issue.
"Good but undifferentiated" is now the hardest position for a startup to raise from, riskier than either extreme, because it doesn't give an investor a clean story to tell their own limited partners.
A founder sitting in the middle of that barbell, solid product, no category-defining scale and no defended niche, is finding this half meaningfully harder than the headline $7.4B suggests. The gap between a strong pitch and a fundable one has widened even as the market has technically recovered.
Where the barbell's other end still has room
Watch median deal size against mega-round share through the second half of the year. If the gap keeps widening, expect the bar for a mid-sized Series A or B to keep rising even as headline totals look healthy.
Also track which categories outside clinical workflow and metabolic care manage to break into next quarter's top funded list, since that is where the barbell's other end still has room, and where a genuinely differentiated pitch is most likely to land a check this year rather than next.
Finally, watch whether deal count itself starts moving in the second half, since a rise in deal count alongside rising average size would be the clearest signal that the recovery is finally broadening rather than concentrating further.
