GPUs are also used to train major language models which can improve the experience of providers and patients. This type of AI is used to clinical documentation And medical chatbots.
Experts say the faster and greater computing power of GPUs has made them essential for continued advances in AI in healthcare. “If you’re building a speech synthesis model, a protein design model, or a computer vision model, GPUs can train and run those models more efficiently at scale,” says Kelleher.
“Central processing units are great for sequential tasks,” says Lynch. “But AI requires parallel processing capabilities, and GPUs are better at that.”
Healthcare Semiconductor Supply Chain
Jensen Huang, CEO of NVIDIA, recently describe demand for the company’s Blackwell GPU chip is being called “crazy.” In October 2024, a Morgan Stanley analyst reported that these GPUs were “reserved for 12 months”. Strong demand from hyperscalers, especially Amazon Web Services, Microsoft Azure And Google Web Platform maybe have contributed to the global chip shortageaccording to CDW.
Jon McManus, vice president and head of data, AI and development for Advanced healthcaredescribes the advanced semiconductor supply chain as “fragile” because it relies on “thousands of suppliers having to work in perfect harmony” to produce the chips.
“You already had a weakened industry due to the upheaval caused by the COVID-19 pandemic, followed by this incredible demand placed on it,” says McManus.
Lynch agrees, but thinks the growing interest in AI would have led to a chip shortage even without the pandemic. “My interpretation of the market is that we would have seen this slowdown due to the increase in demand for AI workloads,” he says. “It’s difficult to make more chips because we’ve never seen such demand for raw materials.”
RELATED: These four factors make AI PCs attractive to healthcare professionals.
Impact of chip shortage on healthcare
The majority of health systems use cloud services. According to TechTarget, a Forrester 2023 report on the cloud in healthcare notes that 73% of healthcare organizations partner with multiple public cloud providers.
McManus says organizations like Sharp Healthcare are gaining access to GPU-based AI capabilities through these partnerships. “That’s why I think the healthcare industry isn’t really feeling the impact of the chip shortage, because most of the big cloud providers have already purchased GPUs in bulk.”
He theorizes that health systems with technical debt or that still primarily use on-premises servers could struggle to deploy AI if the chip shortage persists in the long term. If large cloud providers didn’t have enough GPUs, organizations late in the AI game might not be able to secure these partnerships.
“Unless you have local AI development talent, many of these AI capabilities are almost exclusively available through cloud partnerships,” says McManus. “Health systems that are not yet doing anything in this space may struggle to access it. »
EXPLORE: When is the cloud suitable for organizations deploying AI?
How to get around the chip shortage
To get around the chip shortage, one option is to use AI resources from large cloud computing companies to train large language models or run analytics before bringing the proprietary data back on-site. “I don’t need to buy GPUs and keep them on-site in my own data center,” says Lynch. “I can basically rent them on Amazon, Microsoft Azure and others, and I’m only charged for the data I actually use.
McManus recommends working with multiple vendors to ensure the organization always has access to cloud computing services and the AI capabilities it needs. Another option is to try an alternative such as Google Tensor Processing Unitsdesigned to handle AI workloads.
Risk mitigation should be part of any organization’s planning, McManus adds. “In the event of a significant shortage of GPUs, the economic consequences could be profound. At some point, if there is a limit, be prepared to put that AI ambition on hold until the market can stabilize.