Deepnoid

Deepnoid is a KOSDAQ-listed AI solutions company operating across two business pillars: medical AI and industrial AI.

The industrial AI segment supplies airport security screening solutions ('Deep Security') and manufacturing-line defect and foreign-material inspection solutions ('Deep Factory') to customers including Korea Airports Corporation and display, semiconductor, and battery manufacturers.

The company has disclosed that most of its current revenue comes from industrial AI solutions supplied to airports, data centers, and manufacturers, with Deep Security and Deep Factory serving as its core industrial products.

The medical AI segment comprises more than twenty solutions, including the intracranial aneurysm diagnostic aid DEEP:NEURO, the chest-disease aid DEEP:CHEST, the pulmonary nodule information tool DEEP:LUNG, and the spinal-disease diagnostic aid DEEP:SPINE.

The centerpiece of its recent growth strategy is M4CXR, a large-language-model-based generative AI reading solution that analyzes chest X-ray images for more than 41 abnormal findings and automatically generates a preliminary report in an average of 2.3 seconds.

The domestic medical imaging reading market is estimated at roughly KRW 1 trillion, comprising an insurance-covered market of KRW 460 billion, a health-checkup-center market of KRW 500 billion, and an outsourced reading market of KRW 120 billion.

In terms of competitive positioning, listed domestic medical AI peers such as Lunit, VUNO, and JLK compete in a similar image-reading-assistance market, and several of these peers, like Deepnoid, have faced the expiration of pre-tax loss exemption periods granted under the technology special listing framework.

Deepnoid has outlined plans to expand beyond reading assistance into a 'medical AI agent' model that connects broader hospital workflows.

Its customer base is split between tertiary general hospitals on the medical side and public institutions and manufacturers on the industrial side, so order cycles from both pillars simultaneously influence overall results.

Deepnoid's annual revenue contracted from KRW 3.18 billion in 2022 to KRW 1.93 billion in 2023, surged to KRW 10.85 billion in 2024, then fell back to KRW 7.47 billion in 2025.

Operating losses widened every year regardless of revenue swings, from KRW -6.16 billion in 2022 and KRW -6.73 billion in 2023 to KRW -10.10 billion in 2024 and KRW -10.56 billion in 2025.

Net loss attributable to owners similarly grew from KRW -5.33 billion in 2022 to KRW -10.75 billion in 2025, with the 2025 operating margin at -141.3%.

On a quarterly basis, 2Q25 showed signs of improvement with revenue of KRW 3.67 billion and an operating loss of only KRW -0.23 billion, but this reversed sharply in 3Q25 as revenue fell to KRW 2.18 billion and the operating loss widened to KRW -3.04 billion.

Revenue then bottomed at KRW 658 million in 4Q25 and KRW 357 million in 1Q26, even as operating losses expanded further to KRW -3.72 billion and KRW -3.98 billion in those same quarters.

Revenue partially rebounded to KRW 1.31 billion in 2Q26, but the operating loss of KRW -3.30 billion still far exceeded the revenue base, illustrating a cost structure where fixed expenses flow directly through to the bottom line as revenue fluctuates.

The trailing four-quarter (3Q25-2Q26) net loss attributable to owners totaled KRW -14.32 billion, already exceeding the full-year 2025 net loss of KRW -10.75 billion.

On the balance sheet, the December 2025 rights offering lifted equity from KRW 19.04 billion in 2024 to KRW 28.48 billion in 2025 and reduced the debt ratio from 67.9% to 42.0%, but operating cash flow remained negative for four consecutive years from 2022 to 2025 (ranging from KRW -3.71 billion to KRW -13.90 billion), underscoring a persistent lack of cash-generating capacity.

The domestic medical AI image-reading market, combining the insurance-covered segment, health-checkup centers, and outsourced reading, is estimated at roughly KRW 1 trillion, while the generative AI automated-report segment that goes beyond reading assistance remains at an early stage.

Conventional CNN-based reading-assistance solutions have been criticized for facing low reimbursement rates and a limited addressable market, and Deepnoid has stated it targeted this gap by becoming the first in Korea to obtain device approval for an LLM-based generative reading solution.

Competitors such as Lunit, VUNO, and JLK are also pursuing domestic and overseas markets with their own imaging AI products, and several of these peers, having listed under the same 2020-2021 technology special-listing cohort as Deepnoid, are sequentially facing the expiration of pre-tax loss exemption periods related to administrative-issue designation.

In the industrial AI space, demand for automated quality inspection in airport security and semiconductor, display, and battery manufacturing lines remains steady, but the project-based order structure produces large quarter-to-quarter revenue swings.

At the policy level, government-funded programs such as the Ministry of Health and Welfare's AI-based clinical system support project for regionally responsible hospitals are helping to seed expanded supply to tertiary general hospitals.

However, the medical AI industry as a whole is characterized by long lead times to commercialization—covering reimbursement code inclusion, clinical validation periods, and hospital-by-hospital adoption procedures—leaving uncertainty around revenue recognition timing as a common industry-wide risk.

On July 13, 2026, Deepnoid held a media day outlining its M4CXR commercialization strategy and stated it expects related revenue to begin in the fourth quarter following hospital-by-hospital performance verification.

Following device approval from the Ministry of Food and Drug Safety in late June 2026, M4CXR has since been supplied under contract to Pusan National University Hospital in August and to Hwasun Chonnam National University Hospital on August 31, expanding clinical adoption at tertiary general hospitals.

Both contracts share a common origin in the Ministry of Health and Welfare's AI-based clinical system support project for regionally responsible hospitals.

The company also stated it completed the regulatory approval process for its agent-AI-based software medical device 'AxonFlow' on August 6, and it plans to unveil a medical-specialized foundation model called 'MedZero' by year-end while expanding applicable modalities to CT and MRI.

At the media day, CEO Choi Woo-sik cited a target of "KRW 4-5 billion in revenue per product" as an indication of the scale expected from future service expansion.

The company disclosed plans to deploy a substantial portion of its December 2025 rights-offering proceeds into medical AI R&D, clinical trials, and sales and marketing, and to pursue overseas approvals in stages, starting with clinical trials in Vietnam and the Philippines, followed by the United States and Singapore in 2026 and Europe in 2027.

These plans, however, remain company-stated targets and ongoing procedures, and the actual timing and scale of revenue recognition may vary depending on hospital-level adoption speed and reimbursement policy.

In the industrial AI segment, additional order wins from airports and manufacturers remain another variable for annual revenue.