Despite the increased use of Artificial Intelligence (AI) in the medical field, it is reportedly 'not profitable.' Some advised the need for new challenges to overcome systematic limitations.

During the KHF2025 K-Digital Healthcare Summit on the 18th, Park Yong-min, a lead researcher at LG AI Research, suggested that for healthcare AI to be profitable, companies need to direct their business from hospitals to pharmaceutical companies and adopt new performance-based business models.

According to Park, three factors contribute to the reality that AI is 'not profitable' in healthcare: an oversaturated market, a structural limitation centered on on-premise systems, and AI's role as a supplemental tool detached from the core business.

Park said, "The medical AI solution market has become so saturated, like a camera app market that anyone can easily jump into, that it's difficult to create a differentiated value." He pointed out that "With the rapid advancement and open-sourcing of AI technology in the last year or two, the difficulty of development has decreased, making it even harder to generate profits."

Furthermore, Park explained that because hospitals, for safety reasons, insist on installing software on their own servers rather than in the cloud, the revenue for software companies is extremely low.

For example, the software license accounts for only about 10% of the cost, while the remaining 90% covers ancillary expenses such as hardware installation and maintenance personnel, creating a structure where the software developer's share is minimal.

Another barrier preventing the active use of this technology in the field is the lack of perceived intrinsic value in AI.

While many medical AI companies sell their AI models, the argument is that because the products are treated as auxiliary tools separate from the core business, the value customers perceive is unclear.

Park explained, "AI itself doesn't make money; the core business that embraces AI must be profitable." He added, "If medical AI is not linked to a hospital's core revenue, it becomes difficult for the decision-makers to justify spending money on it."

'Reimbursement and implementation delays' have been hurdles for medical AI…request prompted for a change in perspective

Park also analyzed that structural obstacle in the industry are a significant factor in the lack of profitability for medical AI companies, citing a reimbursement system that does not translate technological advancements into increased revenue.

"The system is structured so that AI created with technology from two years ago or the latest technology gets the same amount of money," Park stated. "Even innovative AI is facing difficulty in generating additional revenue."

According to analysis by Park, the lack of incentives for technological advancement weakens companies' motivation to invest, resulting in a proliferation of similar technologies on the market.

Park also pointed out that the process of adopting technology in hospitals is inefficient. To integrate an AI solution into a hospital's Electronic Health Record (EHR) system, it must undergo complex decision-making procedures, including review by each clinical department and the hospital's IT committee.

"Assuming it takes about six months to adopt a specific solution, that's a long enough time for updated technology to emerge," Park said. "As a result, a technology gap can open up before both the medical professionals (the users) and the developers (the suppliers) can even feel the effects of the adoption."

Related to this, the importance of shifting the business direction was emphasized.

Park stated that AI's value would be recognized adequately if it were integrated into the value chain of pharmaceutical companies, which have larger budgets, rather than being limited to hospital services.

For example, a health management app that collects a patient's daily data could be used as a tool to collect digital biomarkers for clinical trials or as a companion app for patient monitoring after a new drug is launched. Such usage would significantly increase its value and a pharmaceutical company's willingness to pay.

"Even if it's the same technology, changing perspective from the current field to new drug development would open up a wider world," Park urged for a shift.

Concluding his presentation, Park emphasized, "Medical AI must find a path to profitability." He pointed out that medical AI companies can only build sustainable businesses by adopting realistic strategic adjustments and fostering inter-industry collaboration, while acknowledging structural limitations.