Healthcare providers are sometimes known for taking a wait-and-see approach to new technology, not wanting to risk exposing patient data or make changes to care until solutions have been fully vetted and validated.
Yet, 53% of healthcare organizations are actively implementing generative AI strategies, and another 32% have devised AI strategies but have not yet begun implementation, according to the 2025 Nutanix Enterprise Cloud Index (ECI) Report for Healthcare. The Nutanix-commissioned global research study surveyed 1,500 IT and DevOps/platform decision-makers, including those in healthcare.
“There’s a lot of fear about finding the right use case,” Kurt Telep, a Nutanix Field CTO focused on healthcare, told The Forecast.
“A lot of AI products in healthcare today are point products, and people are worried about making the wrong investment.”
Here are the ECI Report’s most important findings about what benefits healthcare leaders expect AI to provide, what challenges they expect to face, and what tech investments will get them where they need to be.
The Promise of Increased Productivity
While hospital billboards trumpet AI's diagnostic capabilities, Telep says much of the near-term value lies in helping clinicians and other employees perform their work more quickly.
“The thing that most people are excited about is the opportunities to increase clinician efficiency, whether it’s through ambient audio transcription, summarization of medical records, or even billing and coding.”
According to the ECI Report for Healthcare, 59% of healthcare respondents say they expect generative AI to increase productivity (more than any other benefit), while 51% say they expect to see increased automation and efficiency. Other areas of excitement include increased innovation (cited by 45%), customer retention and support (46%), and employee onboarding (40%).
“Hospitals are still very concerned with improving the clinician experience and patient experience,” said Leah Gabbert, marketing director for global industry solutions at Nutanix.
“So, when AI can optimize either one of those, healthcare leaders are going to pay attention. We’re seeing a massive consumerism of healthcare, where patients have the ability to make decisions about where they receive care. If it’s easy to make an appointment, easy to see your bill, easy to pay, patients will choose that experience every time.”
Protecting Patient Data
The top AI-related challenge faced by healthcare organizations is data privacy and security. Thirty-nine percent of healthcare respondents identify the privacy and security concerns of using LLMs as their number-one challenge, more than any other factor. And 96% of respondents say their organization could be doing more to secure their generative AI models and applications.
“Generative AI is rewriting the rules of data governance,” Gabbert said. “Healthcare leaders know the stakes, and they’re calling for smarter, more secure infrastructure to match the pace of innovation.”
Healthcare providers, Telep says, have a “healthy fear” about using public large language models (LLMs), even for uses that don’t include private patient data. To protect their data, he says, healthcare providers must adopt solutions that give them end-to-end control.
“When you own the whole pipeline, all of those things can be controlled,” he says. “As soon as you lose control – whether that’s due to an off-premises tool, or you’re farming your data out, or you’re injecting another tool into the middle – you’re significantly and exponentially increasing the risk.”
Expected ROI
There has been some pushback against the idea that AI can help organizations save money and open up new revenue sources, including a report from MIT stating that 95% of AI pilots produce “zero return.” However, healthcare leaders are far more optimistic, according to IT professionals surveyed for the ECI Report for Healthcare. Fifty-seven percent say they expect to make a gain on their generative AI investments over the next 12 months, and 69% expect a positive return over the next one to three years.
“One key component of the return-on-investment numbers is labor savings,” said Telep. “There’s also the expectation that a reduction in the individual clinician’s non-patient-facing workload will increase the number of patients they will be able to see per day, driving an increase in revenue.”
Gabbert notes that expectations are sometimes inflated before organizations actually implement new technologies, with leaders recalibrating after they see them in action.
“A lot of people are still at the peak of inflated expectations with AI,” she said. “Then, there’s a natural drop-off. We expect it all to level out eventually.”
Infrastructure for AI Success
Containerization is crucial for supporting generative AI workloads, and the ECI Report shows that 71% of healthcare organizations are already using containers to run these workloads. Still, 79% healthcare respondents feel their current IT infrastructure requires improvement to fully support cloud-native applications and containers, and 59% find cloud native and container-native application development challenging.
“To me, this report validated that we’re going to continue to see AI growth within healthcare,” Telep said. “The idea that no one in healthcare is leveraging container-based workloads is a little bit of a misnomer; it is just that these workloads are often not being managed or driven by the primary IT management organization.”
“Healthcare providers are not going to build the infrastructure until the vendors tell them that they have to have it,” Telep added. “I think AI is going to be what drives that change.”
Editor’s note: See more highlights from the ECI Report in this Nutanix blog post, Healthcare's GenAI Future Hinges on Trust, Talent, and Infrastructure.
Calvin Hennick is a contributing writer. His work appears in BizTech, Engineering Inc., The Boston Globe Magazine and elsewhere. He is also the author of Once More to the Rodeo: A Memoir. Find him on LinkedIn.
© 2025 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.