YIDU TECH (02158) first profitable financial report interpretation: AI medical care from "investment narrative" to "structural realization"

When YIDU TECH (02158) released its performance for the fiscal year 2026, the market's most immediate reaction focused on three key numbers: the first full-year profit (reaching 78.77 million yuan, exceeding the profit forecast range), a 14.6% year-on-year revenue growth, and positive operating cash flow in the second half of the year. While these numbers are important, focusing solely on whether the company is profitable could overlook the true significance of this financial reportit is not just a simple company performance report, but a snapshot of the ongoing structural changes in the industry. Looking at the historical experience of the capital market, what truly changes the valuation system is never "profit occurring," but rather "how profit occurs." In other words, the transition from losses to profits for some companies involves cyclical repair, while for others it involves structural restructuring, and YIDU TECH is closer to the latter this time. To understand this, it is necessary to place it back within the context of the entire medical AI industry. The "first phase" of medical AI has ended: from project-driven to system competition. Over the past five years, the medical AI industry has roughly followed a typical path: algorithmic companies entering hospital information scenarios, large models driving product upgrades, and AI gradually entering clinical processes. The industry has been lively, with continuous financing and new concepts emerging. However, regardless of how the form changes, the underlying commercial structure has not fundamentally changedmost companies still rely on "project-based revenue". What is the essence of project-based revenue? One delivery corresponds to one revenue, one acceptance corresponds to one conclusion, and the next project starts from scratch. This model means that it cannot form true scalable leverage, and financially it exhibits three typical characteristics: - Income growth but long-term profit lagging behindbecause human resources need to be reinvested every time a project is undertaken; - Gross profit margin fluctuationdue to significant differences in customization levels between different hospitals and systems; - High dependence of cash flow on project pacecash flow inflow when acceptance is concentrated, pressure during idle periods. These three characteristics have been common in the financial performance of most medical AI companies in the past few years. YIDU TECH has also been in this stage in recent years. However, the change in the 2026 fiscal year is that these three indicators have begun to show directional changes in the same year: stable income growth, rising gross profit margin, and positive cash flow in the second half of the year. These improvements occurring simultaneously within the same fiscal year cannot be explained by simple seasonal fluctuations. This is a crucial turning pointthe medical AI industry is beginning to transition from "project competition" to "system competition". System competition is about selling not just a piece of software or an algorithm module, but the "ability to operate continuously"the ability to self-iterate with changes in hospital business, to learn continuously from data in multiple hospitals, and to serve more scenarios without increasing marginal costs. This is why, as the industry enters this stage, differences between companies suddenly widen: the competition variable has shifted from "whether projects can be won" to "whether systems can work continuously and generate quantifiable clinical and management value". In this regard, YIDU TECH's financial report actually serves as a signal: the industry may be entering the second phase at a faster pace than expected by the market. Comparative perspective: YIDU TECH is more like the realization of a "medical operating system" When comparing global AI medical companies horizontally, an interesting hierarchical structure can be observed. The first type is tool-based AI companies, such as imaging AI and single-disease auxiliary diagnosis tools. These companies solve the "recognition" problemfinding lesions in medical images, identifying abnormalities in electrocardiograms. They provide a point-like ability, with clear value but a clear ceiling: the unit price of each software is limited, and it is difficult to naturally expand from one disease to another. The second type is process-based AI companies, focused on optimizing clinical workflows and enhancing hospital information system AI. These companies solve the "efficiency" problemreduce the amount of writing for doctors, decrease the amount of data entry for nurses, accelerate medical record circulation. They provide linear improvements, valuable but susceptible to substitution competitionbecause hospital information system (HIS) providers and electronic medical record (EMR) providers may integrate AI functionality into their products at any time. The third type is platform-based AI companies, aiming to form a unified data and decision-making system across hospitals, pharmaceutical companies, and insurance companies. These companies solve the "decision-making" problemhelping doctors make more precise judgments in complex cases, assisting pharmaceutical companies to avoid unnecessary detours in research and development, aiding insurance companies in managing risks better in product design. They provide a systemic ability, with the deepest moat, but also the longest verification cycle. YIDU TECH's position lies in the transitional zone between the second and third categorieshaving already possessed the underlying architecture of platform companies (unified data governance, unified model base, unified scenario adaptation framework), but still undergoing the final transition from "project delivery" to "platform scaling". From the structure of the financial report, the three main businessesAI for Medical, AI for Life Sciences, AI for Careare not simply business splits, but more like three different system interfaces: providing data entry and clinical decision paths at the hospital end, enhancing research and development efficiency and evidence generation at the pharmaceutical end, and offering population scale and risk pricing at the insurance end. These three interfaces share the same foundation: YiduCore. This brings about a crucial change: the company no longer relies on a single industry cycle but resonates across industries. This is very evident in the financial reportall three business segments are growing simultaneously, rather than one line driving the overall growth. The true value of YiduCore: not the scale of processed data, but the "computationalization of clinical decision-making" If the first two parts represent "operational changes", then YiduCore represents the "underlying variable" that determines long-term competitiveness. The market often discusses the scale of data processed by YiduCorenearly 9 billion medical records, 1.32 billion patient visits. These numbers indeed constitute a barrier to entry, but from an industry observation perspective, the more critical issue is another one: can medical AI convert "clinical experience" into "computable rules"? This matter is much more important than the scale of data processing itself. Because the medical industry is fundamentally an experience-driven systemevery judgment made by a doctor is based on guidelines, experience, past cases, and the complex individual differences of patients. For AI to truly enter this system, it must resolve a fundamental contradiction: how to convert highly unstructured medical knowledge into structured decision logic? In other words, not to have AI memorize medical textbooks but to have AI understand the complete chain from the patient's "chief complaint" to "diagnosis" to "treatment plan" to "prognosis management". The significance of YiduCore lies precisely in this. It is not just a data platform but is doing three more fundamental things: structuring the process of disease occurrence and developmentfrom symptom onset to test results to diagnostic classification to treatment pathways, forming a dynamic evolving model; standardizing clinical pathwaysnot rigidly copying guidelines but combining actual treatment plans taken by doctors in the real world with corresponding results; modeling diagnostic and treatment decisionsallowing AI to, in front of a specific patient, generate interpretable and traceable decision recommendations based on existing evidence and data. This is also why medical AI and general large models are fundamentally different in underlying logic. General AI solves the "representation problem"how to generate fluent natural language, how to organize existing knowledge, how to imitate human conversational styles; whereas medical AI solves the "decision-making problem"whether the conclusion is correct for a real patient, whether the suggestion can be applied, and whether the judgment is supported by evidence-based reasoning. The former emphasizes being "like a person", while the latter emphasizes being "trustworthy". Establishing "trustworthiness" requires long-term accumulation of real clinical data, deep collaboration with top hospitals, and repeated verification in real diagnostic and treatment scenariosa process that cannot be quickly assembled through computational power and parameters. Key change in the capital market: Medical AI is transitioning from "story-based pricing" to "structure-based pricing" If we summarize this financial report from the perspective of the capital market, it actually provides an important signal: medical AI is transitioning from PS logic to tiered pricing logic. In the past, the market mainly looked at three indicators: revenue growth ratebelieving in "scale first, profit later"; number of projectsused to demonstrate market coverage and customer acquisition capability; technological conceptsbig models, intelligent agents, multimodal, whichever is trending. Under this pricing logic, the differences in valuation between companies were not significant, as all companies were telling the same "industry space is huge" story. However, the future market will gradually shift to another set of indicators: cash flow capabilitynot just "how much revenue is collected," but also "the efficiency of cash recovery"; penetration rate of medical systemsnot just "how many hospitals have been entered," but also "how deeply they are used at the department level"; cross-scenario reuse capabilityhow many practical payment demands YiduCore's capability has generated in multiple scenarios; integrated capabilities across the medical, pharmaceutical, and insurance endswhether all three payment parties simultaneously recognize the value of the same data foundation. Under this new logic, the differences between companies will quickly widen. Because the essence of the medical AI industry has never been a technological race but a competition of system penetrationwho can establish unified data and decision standards across hospitals, pharmaceutical companies, and insurance companies first, will have the ability to define industry rules. The value of the YIDU TECH 2026 financial report lies in the evidence it provides of "system penetration" taking place. Looking back at this 2026 fiscal year performance, its industry significance can be summarized in three ways: the first profit validates that the business model can be established, the positive cash flow confirms that the system is beginning to have self-circulating capability, and the synergistic growth of multiple businesses verifies the existence of platform reuse capability. The simultaneous occurrence of these three things in the same financial report is not common in the medical AI industryit means that the industry is transitioning from "whether AI can be developed" to the second phase of "how AI can change the medical system". YIDU TECH is an early sample being validated in this phase.