The chipmaker's second annual industry snapshot of more than 600 healthcare and life sciences professionals finds the sector has decisively crossed from experimentation into execution — and the financial returns are real.
---
Healthcare AI has spent years promising transformation while delivering mostly pilot programs. NVIDIA's second annual "State of AI in Healthcare and Life Sciences" survey, published February 24, suggests the experimentation phase is over. Among more than 600 healthcare and life sciences professionals surveyed, 70% said their organizations are actively deploying AI — up from 63% in 2024 — and the majority of executives say that deployment is now visibly moving the financial needle.
The headline finding is deceptively simple: 85% of management respondents said AI is helping increase annual revenue, and 80% said it is helping reduce costs. Those are not projections. They reflect what organizations are already seeing in production environments, from radiology departments using AI to accelerate image reads to pharmaceutical research teams applying large language models to literature review and biomarker identification.
The Numbers Behind the ROI Claims
The survey's strength is in the segment-level detail. In medical technology, 57% of respondents said they are seeing measurable return on investment from AI deployed for medical imaging — a use case where radiologists use AI to flag areas of concern on scans and move through higher volumes of images more efficiently. In pharmaceutical and biotechnology, 46% said drug discovery and development ranked among their top ROI-generating applications.
The differentiation by segment matters because the use cases look very different across the industry. Digital healthcare organizations identified virtual health assistants and chatbots as their top ROI driver, cited by 37% of respondents. Payers and providers — the segment that includes hospitals, primary care practices, and insurers — pointed to administrative task automation and workflow optimization, named by 39%.
AI budget growth tracks the satisfaction with these returns. Eighty-five percent of respondents said their AI budgets would increase in 2026, with 46% expecting growth of more than 10%. On revenue specifically, 44% of all management respondents said AI has already pushed annual revenue up by more than 10%, with small companies reporting even sharper gains — 56% of small-company respondents said revenue growth exceeded 10%.
The Embedding Problem
The survey's expert commentary cuts through the positive headline numbers with a more nuanced view of what separates organizations that are actually capturing returns from those still chasing them.
"Scaling generative AI in healthcare starts with focusing on real clinical and operational problems, rather than the technology itself," said Dr. Annabelle Painter, clinical AI strategy lead at Visiba U.K. "The organizations seeing impact are those that embed AI into existing workflows instead of layering AI on top as a separate tool."
That distinction — embedding versus layering — reflects a broader pattern visible in the data. The survey found that the top AI use cases for the entire industry are clinical decision support (cited by 42% of respondents), medical imaging (38%), and administrative workflow optimization (38%). These are all applications where AI slots into an existing process rather than replacing it wholesale. The radiology use case is illustrative: AI does not displace the radiologist; it highlights areas of concern so the radiologist can work faster and more consistently.
Painter's second observation may be more important for long-term outcomes. "Healthcare organizations that successfully integrate AI are those that explicitly fund and prioritize evaluation as a core operational function, ensuring AI delivers measurable improvements in safety, quality and patient care over time." The organizations struggling, the data suggests, are those that deploy and move on rather than treating ongoing measurement as part of the investment.
Generative AI and the Agentic Turn
The composition of what healthcare organizations are actually running has shifted sharply in a single year. Generative AI and large language models now rank as the most common workload, cited by 69% of respondents — up from 54% in 2024. That 15-percentage-point jump in one year is notable in any industry; in healthcare, where regulatory caution typically slows adoption, it signals genuine momentum.
New to this year's survey: agentic AI. Forty-seven percent of respondents said they are using or assessing AI agents, including 22% who have already deployed them. In pharma and biotech specifically, 55% said they are using agentic AI for literature review and analysis, and nearly half are deploying agents for drug discovery and biomarker identification tasks — the kind of high-volume, high-precision information work that previously consumed significant researcher time.
John Nosta, president of NostaLab, a healthcare think tank, offered a pragmatic forecast for near-term impact. "Over the next 12-18 months, the most visible and scalable impact of AI will come from logistics and administrative streamlining," he said. "That's where adoption curves are already steep — scheduling, documentation, coding, utilization management and care coordination."
Nosta's framing implicitly tempers some of the survey's more expansive claims about drug discovery. The financial returns from radiology and administrative AI are already being captured at scale; the returns from AI-accelerated drug discovery, while real, operate on a longer cycle tied to clinical development timelines.
Open Source as Strategic Infrastructure
Eighty-two percent of survey respondents said open source software and models are moderately to extremely important to their AI strategy — a figure that reflects how the healthcare industry is approaching the build-versus-buy question. Open source frameworks give organizations the ability to fine-tune models on proprietary clinical or genomic data without surrendering that data to a third-party vendor.
Nosta addressed the tension between open exploration and regulated deployment directly. "Open models will shape the intellectual field. They are essential for exploration and for keeping the field honest. But in clinical environments where safety, liability and accountability are nonnegotiable, proprietary systems will remain necessary for validation, integration and trust. The key insight here is that discovery will be open, and deployment will demand stewardship."
That framing captures something the headline adoption numbers do not. At 70% active AI usage, healthcare has reached a level of deployment where the governance question is now more pressing than the adoption question. The survey found that 40% of respondents said HIPAA, FDA approval processes, and GDPR compliance strongly influence their agentic AI implementation strategies — a reminder that the sector's regulatory constraints are not an afterthought but a structural feature of how AI actually gets used.
What the Survey Does Not Settle
NVIDIA sells the infrastructure that runs most of these AI workloads, which makes its annual survey a document worth reading carefully. The respondent pool of more than 600 professionals skews toward organizations already engaged enough with AI to have opinions on it. Organizations that have not yet adopted AI, or that tried and stepped back, are not well-represented in the data.
The ROI figures are also self-reported — organizations are assessing their own financial outcomes without a standardized methodology, which means the 85% revenue-improvement figure likely reflects a wide range of definitions and measurement approaches. Whether a $50,000 efficiency gain in a single radiology department counts as "AI increasing annual revenue" is a judgment call that different respondents will make differently.
What the survey does establish clearly is direction. AI adoption in healthcare is accelerating across every segment tracked, generative AI has become the dominant workload within a single year, and agentic AI is arriving faster than most industry observers expected. The ROI is real enough that 85% of organizations are planning to spend more. The harder work — embedding AI into clinical workflows, funding rigorous evaluation, navigating regulatory requirements — is now the problem the industry has to solve at scale.
---
Sources: NVIDIA Blog (blogs.nvidia.com), Healthcare Digital (healthcare-digital.com), Digital Watch Observatory (dig.watch), Blockchain News (blockchain.news)
"The organizations seeing impact are those that embed AI into existing workflows instead of layering AI on top as a separate tool."— Dr. Annabelle Painter, Clinical AI Strategy Lead, Visiba U.K.