The headline number looked like a recovery. According to Rock Health’s year-end report, US digital health startups raised $14.2 billion in 2025, up 35% from $10.5 billion in 2024 and the highest total since 2022. After a long post-2021 downturn, investors were writing cheques again.
Look one layer down and the picture changes. Rock Health noted that if you removed the nine companies that raised the most, total 2025 funding would fall below 2024’s. The number of deals fell, from 509 to 482. Twenty-six megadeals of $100 million or more took 42% of all the money, the highest share since 2021. Rock Health titled its report “a tale of two markets,” and that is the most accurate summary of where health venture capital went last year.
It went to a small number of companies, mostly selling AI, mostly selling to people who run clinical workflows. Everyone else was fighting over what was left.
Where the money went
Three patterns stand out in the 2025 data.
AI took the majority. Companies that Rock Health classified as AI-enabled made up half of all deals but captured 54% of funding, up from 37% the year before. They also raised larger rounds, with an average deal size roughly 19% higher than non-AI companies. At Series C, the premium was 61%.
Workflow beat everything else. Clinical and non-clinical workflow tools, the category that covers ambient documentation, coding, prior authorisation and similar back-office automation, took 39% of total funding. The fastest valuation stories came from here. Abridge, which builds ambient clinical documentation, doubled its valuation during the year, to $5.3 billion. OpenEvidence, a clinical information tool, multiplied its valuation roughly six times during the year.
Consumer wellness came back. Fitness and wellness companies raised $2.0 billion across 44 deals, rising from eighth to third among funded categories. Oura’s $900 million round did much of the lifting on its own.
Why the money concentrates
Concentration is not an accident of 2025. It follows from how venture funds now work. The largest funds have more capital to deploy than ever, and it is impractical for them to write many small cheques. They need a few large positions, and the easiest place to put a large cheque is a company that already looks like a winner. Every big round makes the next one easier to raise, because it signals to customers, recruits and other investors that this company will survive.
That dynamic has a name in the industry: kingmaking. Capital itself becomes a competitive weapon. A workflow company with a fresh nine-figure round can hire faster, discount harder and outlast a rival with a better product and a thinner bank account. For buyers, this can be reassuring, since the vendor is less likely to vanish. For the market, it means the winners are partly chosen by who raised first, not only by who built best.
It also changes what a smaller investor can do. Early-stage funds that cannot follow on into huge rounds find their stakes diluted in the companies that succeed, and are left holding more of the ones that stall. That is part of why the middle of the market, between promising seed companies and anointed leaders, felt so thin last year.
Why workflow AI, and why now
The logic is not hard to see. Ambient documentation and administrative automation solve a problem health systems will pay for today: clinician time and administrative cost. They do not require a randomised trial to prove clinical benefit. They slot into existing electronic records. And their value can be counted in minutes saved per visit, which is the kind of return a chief financial officer can put in a spreadsheet.
Compare that with a company trying to improve outcomes in chronic disease. It needs to show clinical effect, convince a payer to reimburse it, and wait years for the savings to appear, usually in someone else’s budget. In a market that rewards speed to revenue, the workflow company wins.
That is rational. It is also a quiet shift in what health venture capital is funding. A larger share of the money is going to tools that make the existing system run more efficiently, and a smaller share to tools that change what care is delivered or to whom.
The spreadsheet behind the scribe
The appeal of ambient documentation to a health system is easy to model, which is part of why it sold. Take an illustrative case: a system with 1,000 ambulatory physicians, each of whom spends, say, an hour or more a day on documentation outside clinic time. If a scribe tool saves even a fraction of that, the gain can be counted either as extra visits per day or as reduced burnout and turnover. Replacing a single physician who leaves can cost a system a large sum in recruitment and lost revenue, so a modest drop in turnover can justify a substantial licence fee on its own.
These are illustrative numbers, and real-world results vary. Early evaluations of ambient tools have reported time savings and better clinician experience, with more mixed evidence on productivity. But the point is structural. The buyer can build a business case from their own data in a week, without waiting for a trial. Very few other kinds of health technology can say that.
The counterargument: this is where the pain is
It would be wrong to treat the workflow boom as a failure of imagination. Clinician burnout is a real and expensive problem, and documentation burden is one of its most frequently cited causes. Administrative cost is a large share of health spending in the United States. Tools that give clinicians back time may do more for patients, indirectly, than many products that claim clinical outcomes and never demonstrate them.
That argument largely holds. The concern is not that workflow tools are being funded. It is the balance. When one category takes 39% of the money, and the premium for describing yourself as AI rises, founders working on harder clinical problems face a steeper climb for capital. Some of the most important opportunities, in chronic disease, mental health and care for populations that are not profitable to serve, need patient money and a willingness to wait for evidence. That kind of money was scarcer in 2025 than the headline total suggests.
The platform risk hanging over workflow
There is also a risk inside the winning category. Workflow tools live on top of electronic health records, and record vendors can build similar features into their own products. When the platform owner decides a feature is core, the independent vendor’s differentiation narrows, and the price it can charge falls. Scribes and coding assistants are exactly the kind of capability that could become a standard feature rather than a separate purchase.
The companies raising at high valuations know this. Their bet is that speed, product quality and customer relationships build a lead that platform owners cannot easily close. Some will be right. But investors paying premium prices for workflow companies are also taking a view that those features will not be commoditised soon. That is a bet, not a certainty.
The incentive to put “AI” in the pitch deck
When AI-enabled companies command a premium on deal size, every company has a reason to describe itself as AI-enabled. Some of that is genuine. Some of it is relabelling. Investors know this and try to screen for it, but the incentive is powerful, and it distorts how companies describe what they actually do.
For health systems evaluating vendors, that matters. The label tells you less than it used to about the product. The useful questions are older ones: what problem does this solve, what evidence shows it works in a setting like mine, and who is accountable when it fails?
The companies stuck in the middle
The less-reported part of the 2025 story concerns the companies that did not raise. Rock Health counted more than 600 companies that last raised venture money in 2021 or 2022 and had seen neither new funding nor an exit since. Many were valued at peak-market prices and now face a choice between a down round, a distressed sale or shutting down.
Exits did reopen, partially. Five digital health companies went public in 2025, including Hinge Health and Omada Health, after only two public exits in the previous three years. But Hinge listed at a valuation of about $3 billion, less than half of its $6.2 billion private valuation from 2021. Mergers and acquisitions rose 61% to 195 deals, and Rock Health highlighted several that were sales below what the companies had raised or been valued at. SteadyMD, for example, sold for $25 million after raising around $40 million.
This is consolidation, not collapse. It does mean that a generation of companies founded in the 2020-2021 boom will be absorbed, often cheaply, by better-capitalised competitors.
What this means outside the US
Rock Health’s data covers US startups, and it would be a mistake to read it as a global picture. But US venture patterns tend to shape what gets built and sold elsewhere. If American capital concentrates on workflow automation for large US health systems, fewer well-funded companies will be building for public health systems, low-resource settings or the chronic disease burden that dominates most of the world.
What this means for the people making decisions
For founders outside the favoured categories, the implication is uncomfortable but useful: the 2025 market rewarded fast, measurable return to a paying buyer. A company working on clinical outcomes can still raise, but it needs to show a path to revenue that does not depend on waiting years for savings to appear. Partnering with risk-bearing providers, who capture the savings directly, is one route.
For health system executives, abundant capital in workflow AI means vendors will be plentiful and aggressive. The discipline is to buy on measured results in their own setting, to negotiate pilots with clear exit terms, and to consider what happens if the vendor is acquired or fails. Many of the 2021 cohort that are now selling cheaply had customers who depended on them.
For public funders and philanthropies, the concentration of private capital is a signal of where the gap is. If venture money is flowing toward efficiency for large US systems, the areas it is avoiding, such as low-resource settings and long-horizon clinical outcomes, are where public and philanthropic capital can make the most difference.
The things to watch over the rest of this year are whether the AI premium holds as more companies adopt the label, whether record vendors move deeper into workflow features, and whether any of the stranded 2021 and 2022 companies find buyers at prices that let their investors recover anything at all.
The takeaway
Health venture capital did not so much recover in 2025 as concentrate. More money flowed to fewer companies, more than half of it went to AI-enabled startups, and workflow tools were the single biggest category. That may deliver real value to overstretched clinicians. It also means the market is betting heavily on efficiency, and much less on the harder, slower work of changing health outcomes.
Is the concentration of health investment in workflow AI a sensible bet on real pain points, or a narrowing of what innovation is allowed to look like? Tell us what you think in the comments.
