Oath Surgical

There is a familiar move in healthcare technology, and it goes like this: you build a piece of software, you point it at a hospital, and you spend the next several years discovering that the hospital's twelve other pieces of software would rather not talk to yours. It is a large market and a miserable business, and most companies accept the misery as the price of entry.

Oath Surgical looked at that arrangement and made a different, more expensive decision. Instead of selling software into somebody else's operating room, it built its own. Three of them, in Portland, Oregon, described as "AI-native" - which is a way of saying the walls and the wiring and the software were designed together, on purpose, at the same time. Then it put an operating system on top called OathOS. The company is, in the language of its founder, trying to rebuild outpatient surgery "from the ground up," and the notable thing is that it means the ground quite literally.

This is unusual. Vertical integration is a phrase people use a lot and practice rarely, because it requires doing two hard things instead of one. Oath is running a chain of surgery centers and writing a clinical AI platform, and the bet is that neither works nearly as well without the other. If you own the physical environment, you don't have to beg legacy systems to cooperate - you just don't install them.

What it actually is

Strip away the framing and there are two products stapled together. The first is a set of ambulatory surgical centers - premium, standalone, outpatient facilities where the pitch is safe, cost-effective surgery outside the hospital. The second is OathOS, and OathOS is the part that gets people excited.

In the operating room, OathOS behaves as an ambient AI. It watches and listens - surgical video, audio, and data streaming off the devices in the room - and it does so in real time. The stated goal is "zero-documentation" surgery, which is a slightly heroic way of describing the removal of a very real burden: the hours surgeons and their teams spend typing up what everyone in the room already knows happened. OR-mounted smart interfaces give the team live visibility into the procedure while the automation runs quietly in the background, and the ambition stretches across the whole arc of a case - from referral, through the operation, to recovery.

Whether zero documentation is achievable or merely directional is beside the point for now. The interesting claim is architectural: that surgery is one of the least instrumented, most improvised environments in modern medicine, and that treating it like software infrastructure is overdue.

The founder who quit the cap table

Oath Surgical was founded by Oliver Keown, a physician who spent roughly fifteen years around the frontier of surgical technology - most visibly as head of Intuitive Ventures, the venture arm of Intuitive Surgical, the company that makes the da Vinci robots. This is a relevant biographical detail because it tells you what kind of company Oath is trying to be. Keown spent years writing checks into surgical-tech startups and business-model experiments. At some point he apparently concluded that the deal worth doing was the one he would have to build himself, and left investing to run it.

There's a certain honesty in that. An investor deciding the best opportunity on his desk is the company he'd have to found is, if nothing else, a strong signal. It also explains the shape of Oath's thesis - value-based care, surgeon ownership, and a suspicion of retrofit solutions - which reads less like a software roadmap and more like a set of opinions about how surgery should be owned.

The money

Oath emerged from stealth in May 2025 with OathOS and its owned centers, backed by around $10 million. Five months later, in October 2025, it raised an oversubscribed $24 million Series A led by FPV Ventures, with participation from McKesson Ventures and existing backers OSE, Black Opal Ventures, Tau Ventures, and Rogue VC. That brought total funding to roughly $35 million.

"Oversubscribed" is a word that appears in nearly every funding announcement and means slightly less each time, but the investor list is worth reading closely. McKesson Ventures is strategic money - the kind that comes attached to a company that moves a meaningful fraction of American healthcare supply. FPV led. The rest followed their earlier checks. The capital, the company said, would go toward expanding into new specialties like oncology, scaling the national network, and advancing OathOS.

Then NVIDIA showed up

In January 2026 Oath announced a partnership with NVIDIA - a name that usually turns up in data centers, not surgery centers. The collaboration brings NVIDIA's spatial AI infrastructure to Oath's operating rooms to power OathOS with real-time analysis of surgical video, audio, and operational data, and to run agentic workflows during a case. The framing was deliberate: OathOS as a "multimodal, ambient clinical intelligence platform," and the operating room as an infrastructure layer to be built rather than a place software visits.

Partnerships like this can be substance or garnish. What makes this one more legible than most is Oath's ownership model. Because it controls its own centers, it can actually deploy heavy compute and camera infrastructure end-to-end, rather than negotiating it past a hospital's IT department one firewall at a time. That is, again, the whole argument: owning the room is what makes the AI installable.

Products & the surgeon-owned network

Does the model hold?

The skeptical read is straightforward. Owning surgery centers is capital-intensive, slow, and regulated; three of them in one city is not yet a revolution. Building a serious clinical AI is also hard, and doing both at once is the kind of ambition that can quietly become two struggling companies wearing one logo. "Value-based care" has been the answer to American healthcare's cost problem for two decades and remains stubbornly difficult to scale.

The optimistic read is that all of those are the same objection, and Oath's answer to it is the integration itself. You can't scale value-based surgery on top of systems you don't control, the argument goes, so you build the centers to control them, and you write the software because nobody else's fits. Three centers isn't the product; the template is the product. If it works, the fourth through fortieth are copies.

Which of those readings is correct is, at the moment, genuinely unknown - and that is what makes Oath one of the more interesting experiments in healthtech right now. It is a company that took the most annoying constraint in its industry, interoperability, and tried to make it disappear by refusing to depend on anyone else's stack. That is either the expensive way to fail or the only way to win. We will find out one operating room at a time.

Timeline

FAQ