Most healthcare organisations pour money into GPUs and assume better AI will follow. But compute is only half the equation.
In this Q&A, PEAK:AIO CEO Roger Cummings makes the case that data infrastructure deserves just as much attention as processing power
Q: Artificial intelligence is quickly becoming part of everyday healthcare. What are the biggest challenges that hospitals face as they adopt AI?
Roger Cummings, CEO of PEAK:AIO: Healthcare organisations are generating and analysing more data than ever before.
Medical imaging and electronic health records create enormous amounts of data that often must be available instantly.
Many organisations have invested heavily in GPU technology to power their AI initiatives, and the ability to move data to those systems quickly has become just as important as the AI itself.
If the data cannot be delivered in a reasonable amount of time, even the most powerful AI infrastructure will not perform well.
What we’ve learned at PEAK:AIO is that successful healthcare AI is about building infrastructure around how clinicians and researchers work in the field, so the right information is available exactly when they need it.
Q: Many healthcare leaders focus on computing power when planning their AI initiatives. Why should they also be thinking about data infrastructure?
Roger Cummings
Cummings: AI is like a high-performance engine, very powerful, but it still needs fuel to run.
For AI, the fuel is data. As AI models become larger and healthcare data continues to grow, organisations need infrastructure that can continuously deliver information without creating delays.
Successful, helpful AI in the healthcare sector requires both powerful compute and a data platform that can keep pace.
Too often, organisations think AI performance is determined by compute alone.
The best outcomes are when compute and data work together. If one outpaces the other, organisations never realise the full value of their AI investments.
Q: Healthcare organisations are under pressure to improve efficiency while controlling costs. How can modern data infrastructure help?
Cummings: IT teams have always balanced performance and budget, and healthcare is no exception. Organisations want to extract the most value from the technology they have invested in.
When infrastructure can deliver data more efficiently, organisations can improve how they use their AI resources and, ideally, support more users and apps without continually adding hardware.
Healthcare doesn’t need more complexity. It needs infrastructure that is easy to manage and delivers the best performance for clinicians, researchers, and patients.
That is the ultimate goal — helping lower the cost of operating AI while improving performance.
Q: Looking ahead, how do you see healthcare AI evolving over the next several years?
Cummings: Healthcare is moving beyond AI as a standalone tool and toward AI becoming part of everyday clinical operations — it is already well on its way, and it will help prioritise needs, support diagnoses, and improve administrative efficiency.
As that happens, healthcare organisations will need infrastructure that can scale without becoming more complicated.
The technology should stay in the background. Clinicians should not have to think about IT infrastructure.
They should have the information they need when they need it.
Finally, the technology supporting AI should be able to grow as demand grows, allowing clinicians to focus on patient care rather than on whether the underlying systems are keeping up.
Ultimately, organisations will need infrastructure that allows data to move quickly and reliably wherever it is needed.
Q: If there’s one misconception healthcare leaders have about AI infrastructure today, what is it?
Cummings: From our perspective at PEAK:AIO in supporting AI initiatives in collaboration with King’s College London and other major NHS Trusts, adding more GPUs does not automatically lead to better AI performance.
Compute is important, but AI only performs as well as the data feeding it. As healthcare organisations expand their AI initiatives, the ability to deliver data efficiently becomes just as important as the AI processing power itself.
The organisations seeing the greatest success are looking at AI as an end-to-end solution, where compute, data, and infrastructure all work together.
When those pieces are aligned, healthcare organisations can get far more value from their AI investments.
Q: How has working with healthcare organisations changed the way PEAK:AIO thinks about its technology?
Cummings: One of the biggest lessons we have learned is to start with the healthcare challenge, not the technology.
Early on, we spent time listening to clinicians and healthcare organisations about the problems they were trying to solve rather than telling them what technology they should buy.
Whether it is helping radiologists review images faster or enabling medical research, the goal is not to build better AI storage.
The goal is to remove technology barriers so healthcare professionals can spend more time focused on patients and less time waiting on systems.
Bio
Roger Cummings, CEO of PEAK:AIO
Roger is a seasoned entrepreneur and business leader with a distinguished track record of driving growth, innovation, and market leadership. Over his career, he has successfully guided five early-stage companies through highly successful acquisitions, raising over $1 billion in funding to fuel their global expansion.
Specialising in application infrastructure and AI/ML technologies, Roger has consistently identified emerging opportunities and built organizations that establish market dominance in rapidly evolving industries.
Roger is the CEO of PEAK:AIO, a company at the forefront of enabling enterprise organizations to scale, govern, and secure their AI and HPC applications. Under his leadership, PEAK:AIO delivers cutting-edge software-defined data solutions, transforming commodity hardware into high-performance storage systems for AI and HPC workloads.
Before PEAK:AIO, Roger was the CEO of FitStack, a University of Wisconsin spin-out focused on DevOps intelligence. Prior to that, he led Evidence IQ, an AI/ML company specializing in evidence-based data intelligence.
In addition to his executive roles, Roger has co-authored several papers on go-to-market strategies, operational excellence, and AI application infrastructure, reflecting his thought leadership in the field. Roger’s career exemplifies a commitment to building innovative solutions that empower businesses to thrive in the age of AI and advanced computing.
