Katie Adams , 2025-12-14 16:27:00
When I stepped onto the show floor at the Radiological Society of North America’s annual meeting this month in Chicago’s McCormick Place, I was immediately overwhelmed by the more than 700 vendors exhibiting the latest in imaging technology. Once I explored a bit, something else struck me — nearly every company I spoke with was collaborating with Nvidia in one way or another.
Silicon Valley-based tech firm Nvidia — until now best known for making advanced graphics processing units (GPUs), or chips, for gaming — is becoming one of the hottest names in healthcare. Over the years, it has developed an unrivaled hardware and software tech stack that has become the backbone of deep learning in healthcare. Through the strength of this technology, it has forged partnerships with not only health tech companies but also drugmakers, hospitals, research institutions and digital health startups.
Consider this — in 2025 alone, Nvidia has teamed up or expanded its relationships with Mayo Clinic, Mass General Brigham, Eli Lilly, GE HealthCare, Philips and Verily. So what makes Nvidia different from the other tech companies that have tried to carve out a place in the sun within the world of healthcare?
The past decade has seen multiple tech players founder. One notable example is Google’s health division — the tech giant struggled to scale its clinical AI tools for years and ended up dismantling the business unit in 2021. IBM Watson Health, which fell apart the same year, is another instance in which high expectations for healthcare AI ran up against the realities of complex clinical workflows, siloed data and slow provider adoption. Amazon also has forayed into healthcare with primary care services that it ultimately shuttered, though the company is still hoping to crack the healthcare code through its pharmacy service and One Medical acquisition.
The difference seems to be that Nvidia understands the limits of its role within healthcare.
“We’re not trying to become a healthcare company. We are focused on bringing the most important breakthroughs in computing and making them accessible to the entire healthcare ecosystem,” said Kimberly Powell, general manager of healthcare at Nvidia.
While IBM Watson Health ads on TV were a common occurrence before it crumbled and sold its parts to a private equity firm, it is unlikely you will see Nvidia tout its AI and data analytics capabilities on TV any time soon. However, by focusing on enabling industry players, both big and small, the company is carving out a unique and increasingly indispensable role in healthcare.
Deep tech expertise meets practical use cases
Powell sees Nvidia as a builder of platforms for healthcare stakeholders. The company provides hospitals, device makers, drug manufacturers and researchers with the computing power and AI tools they need to innovate, she said.
Healthcare organizations know their domains better than Nvidia ever could, so the company focuses on equipping them with the technology they need to turn their ideas into real-world tools, Powell explained.
Nvidia entered healthcare in the mid-2000s, starting with GPU-accelerated computing for medical imaging. Powell noted that radiology was the first and most obvious application because scanners are essentially sensors producing massive datasets that Nvidia’s GPUs are uniquely suited to process.
The company has since expanded its work in the healthcare field, applying AI and accelerated computing to drug discovery, surgical robotics and clinical workflow automation.
“You see us popping up all over the place because we will go really deep. We’re not a huge company, but we’ll go really deep with our amazing thought leaders and domain experts. Then we’ll translate that into tools that make it much more approachable for the rest of the industry to take advantage,” Powell declared.
The company — which employs about 36,000 people — hires domain experts in fields like radiology, robotics and computational chemistry to act as translators between healthcare problems and computational solutions, she pointed out.
Powell said this approach has been critical in enabling smooth collaboration with partners in highly technical areas like drug discovery or cancer care because it ensures that Nvidia’s technology is applied in ways that truly address real-world clinical and scientific needs.
Making imaging faster and smarter
Nvidia has been working with GE HealthCare for nearly a decade to develop imaging tools to support timelier diagnoses and simpler clinical workflows, and last week, GE announced a new suite of imaging products powered by Nvidia’s technology infrastructure — including a CT system, MRI machine, cardiac ultrasound device and digital mammography tool.
These products use Nvidia’s computing platforms to accelerate image processing and visualization. Essentially, Nvidia’s computing technology makes it much easier for devices to handle the massive and complex datasets that are typical of modern imaging, said Sergio Calvo, GE HealthCare’s global general manager of theranostics.
He said GE is working with Nvidia to build a next-generation PET scan system as well.
“What they’re doing for us is game changing for the future of PET. Instead of the image just being just a static picture, it’s going to be a picture with multiple frames — so you can create a movie of the PET scan,” Calvo explained.
This improvement will allow clinicians to see how tracers — the radioactive substances used in PET scans to highlight metabolic or molecular activity — move through the body over time, giving way for more dynamic imaging that can improve diagnosis and guide more personalized treatments, he stated.
He added that GE is working with Nvidia to develop devices that perform autonomous ultrasounds and X-rays, but these are in the prototype stage.
Another medtech giant, Philips, also uses Nvidia’s computing to power its imaging and clinical decision support tools. The company relies on Nvidia’s software and AI applications in addition to its hardware, noted Shez Partovi, Philips’ chief innovation and strategy officer.
He said Philips and Nvidia are jointly developing foundation models that enable machines like MRI scanners to perform automated reasoning. In the future, instead of a human analyzing MRI images, the scanner itself may understand what it’s seeing and make autonomous decisions, similar to how self-driving cars operate, Partovi remarked.
“A car has a display, but the human usually does the driving. Now today, the car is driving on its own. So we asked the question, with Nvidia, what would it look like if an MRI [machine] was driving on its own?” he stated.
With the radiologist shortage showing no signs of slowing down, the goal is to make MRI scanning less reliant on human intervention, Partovi said.
AI development without the heavy lifting
Companies are increasingly partnering with Nvidia to boost their AI capabilities because the company provides the infrastructure needed both to train AI models and to deploy them in real-world applications, said Tom Valent, chief business officer at clinical decision support startup Aidoc. Working with Nvidia has allowed his company to focus on algorithm development rather than the underlying computing pipeline, he noted.
In Valent’s view, Nvidia has built an ecosystem that helps both startups like Aidoc and medtech giants like Philips and GE move their AI concepts from the lab to the clinic.
This is true for provider organizations as well. For instance, Mayo Clinic has been working with Nvidia this year to speed up the development of AI tools for digital pathology.
Mayo has tackled the challenge of analyzing massive, gigapixel slide images by leveraging Nvidia’s hardware and software platforms, stated Matt Redlon, vice president of digital biology within Mayo’s digital pathology unit.
Mayo and Nvidia are co-developing tools that can handle these extremely large images efficiently, focusing on the computing and software techniques needed to process, store and analyze the slides while also enabling AI models to extract clinical insights, Redlon explained.
Nvidia was “the obvious partner” for the work because the company owns “what has become the standard platform for training AI models,” he added.
To him, the partnership is less about buying a finished product and more about gaining access to the tools and infrastructure that allow Mayo’s researchers and clinicians to focus on solving real clinical problems.
By providing both the hardware and the optimized software layers, Nvidia enables the health system to experiment at a much quicker pace, Redlon said.
Restraint and humility
Some of Nvidia’s most notable collaborations this year have emerged in the life sciences space.
For example, Verily started integrating Nvidia’s AI software and GPU infrastructure into its research platform, and the two companies are working together to create more powerful multimodal models using large biomedical datasets.
The goal is to create more foundational models that combine genomic data, imaging, clinical notes and lab results to spot patterns that no single dataset could reveal on its own. This could lead to earlier discoveries in areas like disease prediction, drug development and personalized care, explained Myoung Cha, Verily’s chief product officer.
“It’s a great partnership. I think they bring more than computing — there’s also this desire they have to really catalyze the ecosystem. I think they see the potential for AI in healthcare, and a lot of what they’re doing is really to push the envelope, to push into the frontier, and to accelerate this whole AI ecosystem across healthcare,” Cha remarked.
He noted that Nvidia understands a fundamental reality of healthcare AI: data isn’t easily accessible and doesn’t live openly on the internet the way consumer data does. So the company’s leadership is fully aware that healthcare data is fragmented across systems and governed by strict privacy rules.
This challenge also presents the opportunity — this is precisely why there is a large untapped potential for AI in healthcare, Cha said.
He pointed out that Nvidia also recognizes that healthcare requires a “last mile” layer, which its partners like Verily can provide. This focus and restraint differentiate Nvidia from past tech giants that overreached, like IBM Watson Health, Cha stated.
Another leader from one of Nvidia’s life sciences partners — Ben Mabey, chief technology officer at Recursion Pharmaceuticals — agreed with Cha’s take that Nvidia’s success stems from its commitment to building the horizontal compute layer rather than a healthcare business. Mabey’s company works with Nvidia to shorten drug discovery timelines.
Companies like IBM and Amazon began by trying to sell higher-level services and products, like disease-specific AI solutions, population health management tools, and even primary care. Nvidia, on the other hand, sells computing and software infrastructure — which biotech companies, tech startups and providers desperately need to train and run modern AI tools, Mabey explained.
“The real fundamental question is do you believe accelerated computing can have a big impact in healthcare? If so, Nvidia is one of the few places you can really get that from,” he stated.
It’s not as if Nvidia has no rivals in the space, though. Other companies offer accelerated computing platforms, including Google, AMD and Intel, but Nvidia is “really the one game in town,” when it comes to delivering this infrastructure at scale, Mabey said.
The hurdles between potential and impact
Nvidia’s ability to stay humble has been its secret weapon, according to Pankit Bhalodia, partner at consultancy West Monroe. Nvidia’s goal seems to be to help other companies to reach end results or products, rather than trying to do everything themselves — which is unlike many other tech disrupters, he said.
Bhalodia also highlighted that Nvidia has found a sweet spot in healthcare by concentrating on areas that have historically been slow and costly, like drug discovery and complex imaging workflows. However, partnerships alone don’t guarantee an AI revolution — execution matters. There are still some sticky challenges that Nvidia has yet to overcome during its foray into healthcare, Bhalodia added.
Data readiness gaps still show up as a consistent problem. Many healthcare organizations are still not fully prepared to organize and clean their data, which is essential for AI to work effectively. Nvidia’s role is to provide the computing power and software tools that can process and analyze this data — but Bhalodia explained that the healthcare industry still doesn’t have a clear method to bridge the data readiness gap and fully take advantage of AI.
There is also the ever-present challenge of workforce adoption. Even with the next generation of AI tools, clinicians and other staff need training and confidence to integrate these technologies into their daily workflows, or the potential benefits will go unrealized, he noted.
But that responsibility perhaps falls more on healthcare stakeholders rather than Nvidia itself.
Ultimately, Nvidia’s success in healthcare comes from its willingness to play a supporting role rather than trying to own the entire ecosystem. By providing the computing power and AI development infrastructure that healthcare organizations need, it’s helping the industry turn ideas into actual tools — even if the company’s real-world impact is still contingent on the greater healthcare industry overcoming the challenges Bhalodia described.
But this erstwhile gaming company has successfully morphed into a computer vision company with applications in myriad industries. And along the way, it has become the most valuable company in the world and a crucial partner in healthcare’s AI transformation.
Photo: JHVEPhoto, Getty Images
