top of page
  • Facebook
  • Instagram
  • X
  • Youtube
  • LinkedIn
Search

Health Care's Digital Land Grab

The digital health market is growing several times faster than health spending itself. Here is how big it really is, where the money is going, and why that is both a promise and a problem for patients.

Something odd is happening in health care. Almost every rich country is complaining that it cannot afford its health system — and at the same time, investors are pouring record sums into companies that sell technology to those same systems. Both things are true, and the connection between them is the most interesting story in health economics right now.


Digital health is no longer a niche. It is a multi-hundred-billion-dollar global industry, growing at a rate that dwarfs the growth of health budgets. The question is no longer whether it arrives, but who pays for it, what it displaces, and whether patients end up better off.


Market analysts put the global digital health market somewhere between roughly $320 billion and $490 billion in 2026, and they broadly agree it will grow at 21–24% a year. MarketsandMarkets projects a rise from about $199 billion in 2025 to $573 billion by 2030. More bullish forecasters put it beyond $1.2 trillion by 2035.


That spread — a gap of $170 billion in the current year alone — is not sloppiness. It is a definitional problem, and it matters. Does a hospital's electronic patient record count as digital health? A smartwatch? A cloud contract? A telephone triage service that happens to run on software? Different firms draw the boundary in different places, and the market is sized by what vendors sell rather than by what health systems actually buy and use. For anyone trying to plan a budget, the headline number is less useful than the direction of travel — and the direction is unambiguous.


Venture funding is the sharper signal, because it tells you what people are betting on rather than what they are billing for. Rock Health's tracking of US digital health startups shows the shape of the cycle clearly: a pandemic bubble, a hard correction, and an AI-driven rebound.


Figure 1. US digital health venture funding, 2021 to mid-2026

Annual venture capital raised by US digital health startups, $ billions. Source: Rock Health


H1 2026 covers six months only, across 244 deals — slightly ahead of the same period in 2025 ($6.4B across 245 deals).
H1 2026 covers six months only, across 244 deals — slightly ahead of the same period in 2025 ($6.4B across 245 deals).

Where the money is going

2025 was the turning point. US digital health funding reached $14.2 billion, up 35% on the year before, and companies marketing artificial intelligence took 54% of every dollar raised — up from 37% a year earlier — while commanding a 19% premium on deal size. By 2026 the label had become meaningless: Rock Health stopped classifying startups as "AI-enabled" on the grounds that everything is. Investors are no longer asking who has AI. They are asking who has something AI alone cannot provide.


Four destinations dominate. First, clinical AI infrastructure — the platforms hospitals are standardising on. Second, administrative automation, especially documentation and revenue cycle management, where the return on investment is easiest to demonstrate; private equity signed a $12 billion deal for a single revenue cycle company this year. Third, mental health, the top-funded clinical indication for seven consecutive years, driven by a demand-capacity gap no workforce plan can close. Fourth, metabolic and weight management, riding the GLP-1 wave and its consumer-facing supply chain. Wearables have quietly become a fifth pillar: Oura raised $900 million at an $11 billion valuation, Whoop $575 million at $10.1 billion.


But the most consequential trend is not sectoral. It is concentration. Mega-deals of $100 million or more have gone from taking about a fifth of all capital to nearly half in two years.


Figure 2. Share of all digital health venture capital going to $100M+ rounds

Mega-deals as a percentage of total capital deployed. In H1 2026, 19 companies raised 20 mega-deals — just over 8% of deals absorbed 45% of the money. Source: Rock Health.


Meanwhile 35% of all deals in 2025 were flat or down rounds — a market of haves and have-nots.
Meanwhile 35% of all deals in 2025 were flat or down rounds — a market of haves and have-nots.

This is a market picking a small number of winners early, which has a direct consequence for health systems: the vendors you deal with in 2030 are being selected now, by capital markets, not by procurement.

 

"Digital health" bundles together technologies that have almost nothing in common economically. Administrative automation — ambient documentation, coding, scheduling, claims — has the clearest business case and the lowest clinical risk; it is also where adoption is fastest. Diagnostic and decision-support AI promises the largest clinical gains and carries the heaviest evidential and regulatory burden. Virtual care and remote monitoring sits in between: systematic reviews find telehealth cost reductions in the range of a few hundred to a few thousand dollars per event and consistent, modest clinical benefit in conditions like diabetes and hypertension, but the results are highly context-dependent. And consumer platforms — wearables, apps, direct-to-consumer services — are growing fastest of all while sitting largely outside the health system's control. Among mental health and weight management startups that raised money in the first half of 2026, 64% sell directly to consumers.

 

The pressure on health systems

OECD countries spent 9.3% of GDP on health in 2024; in 16 of them the figure is at least 10%, and public health spending now absorbs around 15% of all government expenditure. The OECD expects public health spending to rise by a further 1.5 percentage points of GDP by 2045 — and names technological change, alongside ageing and rising expectations, as a principal driver. The WHO projects a global shortfall of about 11 million health workers by 2030. So systems face rising cost pressure and falling labour availability simultaneously. Digital tools are the obvious answer. They also generate three distinct pressures of their own.


Cost-effective is not the same as cash-releasing. A technology can improve outcomes per pound spent and still increase the total bill.


The first is financial. Most digital health savings are avoided future costs — an admission that did not happen, a complication caught early. Those are real welfare gains, but they rarely appear as money a finance director can redeploy this year. Worse, better detection often finds more disease, and easier access often finds more demand. Convenience can be its own cost driver.


The second is evidential. NICE's Early Value Assessment route was designed to get promising technologies into the NHS quickly. Of 103 technologies assessed, 57 were conditionally recommended — with mental health (40%) and oncology (20%) dominating. "Conditionally" is doing a lot of work in that sentence: it means adopt while generating evidence, and the evidence often does not arrive. Health systems are accumulating a portfolio of half-evaluated tools.


The third is regulatory and operational. The EU AI Act's core obligations for high-risk systems — which explicitly include clinical AI — apply in full from August 2026, bringing conformity assessment, technical documentation and human oversight requirements. The European Health Data Space adds another governance layer. Meanwhile, as well-funded vendors expand their roadmaps to own more of the "operating layer" of care, hospitals increasingly find themselves buying overlapping products from competing suppliers, then paying to integrate them.


What patients actually get — and who gets left behind


For patients, the upside is genuine and often underrated by clinicians. Booking without a phone queue. A repeat prescription in ninety seconds. Continuous glucose data instead of a fingerprick every few days. Therapy in three weeks rather than nine months. Access to a specialist opinion without a day off work and a two-hour drive. These are meaningful improvements in the experience of being ill, and they are precisely the improvements health systems have historically been worst at delivering.


The direct-to-consumer boom also shifts power. Patients increasingly arrive already monitored, already informed, sometimes already treated. That changes the consultation, and not always comfortably.

But the distributional picture is uncomfortable. In the UK, roughly 4.5 million people have never used the internet, and 94% of them are over 55. A 2026 survey found that one in three people aged 75 and over could only see their doctor by booking digitally — and the same proportion said they felt cut off from care. Qualitative research published in the BMJ this year found that digitisation delivered convenience for some patients while creating new risks and new forms of exclusion for others. The people most likely to need health care are, on average, the least likely to be able to access it digitally.


This is the equity trap at the heart of the digital health boom. Investment flows toward those who can pay and engage — insured, employed, app-literate consumers with GLP-1 prescriptions and sleep-tracking rings. The clinical need sits disproportionately elsewhere. Left alone, a rapidly growing digital health market will improve care for people who were already doing relatively well, and health inequality is the difference between those two curves.


The technology is not the interesting variable any more; it works, roughly, and it is coming. What matters is the institutional response. So when the next digital health product arrives with a compelling pitch, three questions do most of the work. Does it release a resource that can actually be redeployed, or only avoid a hypothetical future cost? What is the evidence, and who paid for it? And which patients does it reach — the ones with the greatest need, or the ones easiest to serve?


A market growing at 23% a year will not wait for good answers. But health systems that insist on them will get a very different decade from those that do not.


Sources


 
 
bottom of page