Healthcare AI is entering a reality check phase, with investors now expecting startups to prove clear financial and operational value.
Here are five soundbites from interviews with investors last week at the ViVE conference in Los Angeles about how they are thinking about ROI for healthcare AI tools.
The CFO test
Health systems’ tight finances mean that AI purchases increasingly require CFO approval, which means there needs to be a solid case for financial returns, said Larry Cohen, CEO of Health2047, the American Medical Association’s venture studio.
“You can see it in the way people talk about their products and try to pitch what they’re doing. Now, everybody just doesn’t talk about how much easier it is or how much happier everybody is. Now, they’re coming around to financial ROI,” he declared.
In today’s market, Cohen said, AI tools have to show they can pay for themselves.
The long game
Many healthcare AI companies are still in the early innings when it comes to proving financial returns, particularly for tools that assist with clinical use cases, pointed out Uma Veerappan, vice president at Flare Capital Partners.
In the near term, she believes demonstrating adoption could matter more than showing immediate ROI for some startups.
“With the early stage nature of many of these companies, I don’t necessarily believe that ROI needs to show within the first six or so months of deployment. Rather, I think the ability to show you have a sticky product is what’s really important. After a longer period of time is probably when financial and clinical ROI become more table stakes,” Veerappan remarked.
ROI depends on the tool
ROI looks different depending on the kind of AI tool, noted Rachel Feinman, managing director of Tampa General Hospital Ventures.
Some tools, such as ambient clinical documentation scribes, deliver ROI indirectly rather than through immediate financial gains, she pointed out. For example, health systems may adopt these tools to stay competitive in recruiting physicians, even if they don’t increase visit volume or short-term revenue.
“Already, health systems and clinics are starting to advertise that they have ambient listening for providers, and providers are putting that at the top of their list as it relates to where they want to go — that’s where they want to be,” Feinman explained.
At the same time, she said AI that automates manual administrative work — especially for tasks like revenue cycle automation and data abstraction — typically offers clearer and more immediate financial returns.
ROI-driven business models
Vig Chandramouli, partner at Oak HC/FT, said that he is seeing more AI startups shifting away from traditional SaaS pricing toward models tied directly to outcomes, such as charging per completed task or transaction. This approach helps customers feel they are paying only for measurable results, he stated.
“Companies are getting forced into pricing at three to five times ROI. Value capture is front and center on almost every pitch,” Chandramouli declared.
He warned, however, that transaction-based pricing can create hidden risks. If a company underestimates the true cost of delivering each successful transaction, the unit economics can quickly erode.
In order for the shift to ROI-based pricing to work, Chandramouli thinks startups will have to carefully unpack each contract — clarifying what counts as a successful action, what fees are fixed versus variable and how realistic projected revenue is.
Photo: Malte Mueller, Getty Images
