As hospitals and health systems expand AI, technology leaders are encountering costs that extend beyond purchasing and implementing the tools, from unpredictable usage fees to the infrastructure and workforce needed to keep them running.
Healthcare executives have told Becker’s that although AI offers opportunities to improve efficiency and patient care, its long-term financial implications remain difficult to predict as organizations move from pilots to widespread adoption.
Here are four expenses health systems are confronting as they scale AI:
1. AI maintenance and monitoring costs
Rochester, Minn.-based Mayo Clinic has found that maintaining AI systems can be considerably more expensive than anticipated.
Micky Tripathi, PhD, Mayo Clinic’s chief AI implementation officer, told Becker’s in August that the health system had 128 clinical AI solutions in practice and nearly 500 in its pipeline.
“The cost of maintaining these systems is far higher than I think any of us in the industry really thought,” Dr. Tripathi said.
Unlike traditional software, AI models require continued monitoring as underlying data changes and model performance shifts. Dr. Tripathi said Mayo does not have enough specialized engineers to develop new applications while also maintaining existing ones.
He pointed to ongoing computing expenses, specialized staffing and data infrastructure as three major cost pressures associated with scaling AI, and said health systems must account for those ongoing needs, not just initial deployment costs.
2. EHR infrastructure costs from patient-facing AI
Health systems are also beginning to absorb infrastructure costs generated by patients connecting their medical records to consumer AI platforms.
At Charlotte, N.C.-based Advocate Health, the number of patients linking their MyChart accounts to ChatGPT increased more than 7,000% from July to August, according to Bobbie Byrne, MD, executive vice president and CIO.
The connections generate application programming interface calls against Advocate’s Epic EHR, creating computing expenses even though the health system did not enable the integration itself.
“There is some cost to us based on the volume of patients and the number of requests each patient enters into [ChatGPT],” Dr. Byrne told Becker’s in October.
She said Advocate could absorb the current expenses but expressed concern about continued growth.
Epic has since outlined plans to support Fast Healthcare Interoperability Resources notifications that would reduce unnecessary requests from connected applications. Seth Howard, Epic’s executive vice president of research and development, discussed the change Oct. 21 at the company’s Open@Epic developer conference and said it would help health systems scale these connections.
3. Unpredictable AI consumption fees
Unlike traditional software contracts with fixed subscription prices, some AI tools charge organizations based on how extensively they are used.
At Omaha, Neb.-based Children’s Nebraska, Ryan Cameron, executive vice president and chief information and innovation officer, has raised concerns about the financial uncertainty associated with token-based pricing.
Tokens are chunks of text that large language models read and generate, and vendors often bill by the token, meaning expenses can increase as more employees use AI or applications handle increasingly complex tasks.
Mr. Cameron told Becker’s in October that hospitals should consider maintaining a 5% to 10% contingency budget when adopting consumption-based technologies, particularly if utilization exceeds expectations.
Still, Children’s Nebraska has not experienced a major unexpected AI bill that forced it to discontinue a tool, Mr. Cameron said, though it remains early in its implementation. The health system is monitoring adoption and consumption costs as it expands AI.
Philadelphia-based Jefferson Health is also taking steps to prevent spending from escalating. Luis Taveras, PhD, executive vice president and chief digital and information officer, told Becker’s in August that the organization has adopted a tiered approach that directs employees toward different AI tools based on their needs. Employees with routine questions are pointed to Google, while those who need more move up to Microsoft Copilot tiers with fixed pricing. Only employees who need more advanced capabilities get access to frontier models billed by the token.
4. Rising hardware prices driven by AI demand
The AI boom is also raising prices on technology purchases that go beyond AI itself.
At Shelbyville, Ind.-based Major Health Partners, Don Barton, chief technical officer and director of supply chain management, told Becker’s in May that growing demand for AI infrastructure is increasing the cost of enterprise technology.
“The rapid expansion of AI infrastructure and data centers has significantly increased demand for servers, GPUs, memory and storage components, driving up prices for nearly all enterprise hardware and IT-related purchases,” he said.
Mr. Barton said those pressures are making it more expensive for hospitals to modernize their IT infrastructure and cybersecurity environments.
Becker’s 3rd Annual Fall Chief Pharmacy Officer Summit is the top tier event that the people leading the future of pharmacy convene. This 3-day event is built specifically for Chief Pharmacy Officers and senior pharmacy leaders. The sessions examine the workforce, financial and clinical priorities shaping pharmacy today. Becker’s is where candid discussions flourish and new ideas thrive. Becker's 3rd Annual Fall Chief Pharmacy Officer Summit
