Dr Kola Tytler (MBBS MBA MRCGP) | 15 July 2026 | 10 min read
Neko Health, the Stockholm-based preventive health company co-founded by Spotify's Daniel Ek and Hjalmar Nilsonne, has closed a $700 million Series C funding round, reportedly valuing the company at around $7 billion. The round was led by Lightspeed Venture Partners and co-led by O.G. Venture Partners, with existing investors Atomico, General Catalyst and Lakestar joined by new backers including Liberty City Ventures and Michael Dell-backed BDT & MSD. The company says it has completed more than 100,000 assessments across the UK and Sweden, and plans to open its first US clinic in New York later this year, backed by a US waitlist reported at over 300,000 signups. That is an unusually large raise for a company that, at first glance, runs consumer health clinics rather than software.
Why the size of this round is worth pausing on
Most of the medical AI funding cycle over the past two years has concentrated on companies that primarily process existing information: clinical search tools, ambient documentation assistants, referral and coding automation, and administrative workflow support. Neko sits in a different category. It does not depend primarily on data that already exists in a medical record. It generates new physiological and imaging data itself, through proprietary hardware, in a 60-minute clinic visit priced at £299 in the UK, and then applies AI and clinician interpretation on top of it. Investors backing a $7 billion valuation are not simply betting that Neko's software is good. They are betting that owning the full stack, sensor, data, algorithm and clinician, produces something a purely software-based competitor cannot easily replicate.
The first phase of generative AI in healthcare
The wave of medical AI investment that preceded this one was built almost entirely on interpreting or generating information that already existed somewhere: clinical literature, patient records, consultation audio, referral text. OpenEvidence built a clinical search and reasoning tool used, by its own account, by a large share of US physicians. Ambient scribes including Heidi, Tandem and TORTUS turned spoken consultations into structured notes. iatroX and similar platforms built citation-grounded clinical reference and decision-support tools. All of these are valuable, and several now carry very large valuations of their own. But none of them create new clinical data about the patient in front of them. They organise, retrieve or document data that already exists or is already being generated elsewhere.
What makes Neko structurally different
Neko's assessment uses proprietary optical, cardiovascular and dermatological sensors, built in-house, alongside blood analysis, to capture what the company describes as millions of data points per visit. Its leadership has been explicit that owning the sensing hardware, not just the software layer on top of it, is central to the model: newer devices in its clinics, referred to internally as Derma-2, Echo-2 and Spectrum-2, are designed and manufactured by the company itself, and the pace at which Neko can add new detection features depends on that vertical control rather than on a third-party device supplier's roadmap. That is a meaningfully different business to run than a chat interface sitting on top of a large language model, and it is considerably harder and slower to copy.
The broader implication for how medical AI gets categorised
It is becoming useful to think about medical AI in terms of what kind of data it touches, rather than simply how sophisticated its model is. Some products understand and retrieve medical knowledge. Some capture and structure the clinical encounter itself. Some interpret imaging or physiological signals that already exist. And a smaller, more capital-intensive category, of which Neko is currently the most visible example, actually generates the physiological data in the first place, at the point of contact with the patient, and then interprets it end to end. Each category faces different competitive dynamics, different regulatory exposure, and different capital requirements, and conflating them, treating every well-funded health AI company as competing in the same race, tends to obscure more than it reveals.
A balanced read on what this proves, and does not
The raise is genuine evidence of investor confidence in AI-native preventive health as a commercial category, at a scale that would have looked implausible for a body-scanning company even eighteen months ago. It does not, on its own, demonstrate that broad, largely asymptomatic preventive screening improves mortality or population-level outcomes; that is a separate, harder question that commercial success does not answer, and one worth returning to directly rather than assuming. What the round does show clearly is that the more interesting frontier in medical AI may not be a better chatbot. It may be systems that combine hardware, proprietary data generation, AI interpretation and clinician judgement into something closer to a complete healthcare delivery model than a software feature.