알림 목록

Analysis Region: Gyeongsangnam-do and 18 Cities & Counties
Key Industries:  Shipbuilding, Defense, Aerospace, Machinery, Automotive Parts, Agro-Fisheries & Food, Tourism
Analysis Agenda:  Administrative AX connecting dispersed data from the province, cities/counties, and companies on the same timeline as industrial changes and policy outcomes
Golden Time Type: Data Infrastructure Opportunity + Interoperability Lock-in Risk
Version: Regional AX Golden Time Intelligence v3.2
Reference Date for Reinforcement Criteria: 2026.08.28
 

The Gyeongnam Big Data Hub Platform began operations in August 2020, providing public data, statistics, analysis examples, and visualization services for the province and 18 cities and counties. Data on the status of the manufacturing sector has also been made public by unit, including company name, address, product, industry, and reference date, establishing a foundation for verifying the spatial distribution of companies. However, public evidence linking factory registration data with employment, exports, orders, energy, smart factories, policy support, and business closures/suspensions based on the same company has not yet been confirmed. If new datasets for 2026–2028 are accumulated under the current structure, the number of files will increase, but the cost of tracking changes in the status of companies and industries will rise along with them.  Gyeongnam has reached the stage of an administration that possesses industrial data, but it has not yet reached the administrative AX that connects and determines industrial changes. ( Gyeongnam Big Data Hub Platform , Status of Manufacturing in Gyeongsangnam-do )

Gyeongnam received the highest rating of "Excellent" in the Ministry of the Interior and Safety's 2024 status inspection of data-based administration, and finalized a comprehensive AI administration plan for 2025 worth 35.1 billion won, comprising three major strategies and 20 tasks. The implementation list includes an AI-based integrated welfare platform, advanced report reception, and AI support for civil servant tasks. While the institutional foundation for data management and AI utilization has expanded, a runtime for jointly assessing a single industrial risk by connecting industrial data from 18 cities and counties via a common standard has not yet been confirmed. If departmental AI services are established first over the next two to three years, data definitions and authority systems will become fixed on a service-by-service basis.  The current readiness of Gyeongnam's Administrative AX is assessed as upper-middle for internal agency utilization and lower-middle for interoperability between the province and cities/counties. ( Ministry of the Interior and Safety , Gyeongnam AI Administration Comprehensive Plan )

Gyeongnam is home to 3,014 companies that have established smart factories, a manufacturing AI data center worth 23.3 billion won (2025–2026), and the Changwon National Industrial Complex AX demonstration project worth 22.2 billion won (2025–2028). The structure involves the simultaneous expansion of production, quality, and equipment data from manufacturing firms alongside administrative data on businesses, employment, support, and the regional economy. However, the data combination method for aggregating industrial risks while protecting business information, the common enterprise identifier, and data quality standards by city and county have not been disclosed. If manufacturing AI infrastructure and administrative AI are finalized as separate systems between 2026 and 2028, companies will face duplicate submissions, while the administration will be left with redundant support and errors in performance evaluation.  Currently, the "Golden Time" is the balance between Data Infrastructure Opportunity and Interoperability Lock-in Risk. ( Yonhap News , Gyeongsangnam-do Provincial Council Major Business Report )

Gyeongnam's industries are dispersed across Changwon (machinery, defense, and nuclear power); Geoje (shipbuilding); Sacheon (aerospace); Gimhae and Yangsan (machinery and automotive parts); Jinju (agri-food and biotechnology); and Tongyeong (fisheries and tourism). Companies within each industry are not confined to a single city or county but are spread across multiple regions along their supply chains. Administrative data is separated based on business location and the department in charge, which does not align with actual supply chain units. There is no evidence showing how changes in prime contractor orders impact the employment, power, and revenue of partner companies in specific cities or counties. If industry-specific shocks accelerate over a period of two to three years, corporate risks will already have spread to other regions by the time city and county-level statistics are finalized.  While Gyeongnam's industries operate as a wide-area supply chain, administrative data remains confined to the basic administrative level.

The Gyeongnam Big Data Hub makes data available across various fields, including population, industry and employment, environment and weather, transportation, welfare, and tourism, while city and county data such as manufacturing, social enterprises, cooperatives, water usage, and accommodation establishments are also registered. Although the scope of the data is broad, update cycles, address standards, company name notation, missing values, and methods for preserving historical records vary by dataset. There is a gap between searchable data and combinable data. If annual, monthly, and ad-hoc data for 2026–2028 are added based on different criteria, it will become difficult to reconstruct the chronological order of industrial changes.  While the scale of Gyeongnam's data has expanded, there is significant variation in readiness for combination across datasets.

Gyeongnam Province's industrial support is fragmented across the provincial government, 18 cities and counties, Gyeongnam Technopark, the Gyeongnam Investment & Economic Promotion Agency, research institutes, universities, chambers of commerce, and industry-specific organizations. While a single company may participate in multiple projects—such as smart factories, exports, employment, R&D, energy, and finance—there is no evidence that the project numbers and corporate identification systems are identical across these institutions. As the number of institutions and support programs increases, the status of a single company becomes dispersed across multiple records. If support expands over two to three years, a greater risk than duplicate benefits is the inability to determine which combination of support actually produced results.  Although the structural scale of Gyeongnam's administration is large, the corporate-level evidence chain remains incomplete.

In 2024, Gyeongnam’s GRDP reached 151.2 trillion won, an 8.5% increase from the previous year, ranking third nationwide. While a rebound in the manufacturing sector and exports in defense, shipbuilding, and automobiles drove regional growth, GRDP figures by city and county do not immediately reflect current industrial conditions due to discrepancies between the base year and publication dates. A time gap exists between regional growth and the status of individual industrial complexes and companies. If changes in global orders, tariffs, and supply chains accelerate between 2026 and 2028, finalized statistics may be useful for ex-post evaluations but will be too late for early warnings.  Although Gyeongnam accurately compiles industrial scale, it lacks a structure capable of immediately identifying industrial inflection points.

With 3,014 companies having established smart factories, the number is the second highest in the nation after Gyeonggi Province, and the Manufacturing AI Data Center plans to provide GPUs, analysis, and demonstration environments to small and medium-sized enterprises (SMEs) 24 hours a day. While the volume of internal corporate data generation and analysis capabilities are expanding, the method of the administration directly collecting raw corporate data is hampered by issues regarding trade secrets, security, and ownership. A gap is being identified where data openness is equated with industrial analysis. If projects are designed based on the premise of centralized collection over a period of two to three years, the number of participating companies will decrease, leaving only data from large corporations and leading firms. Although  manufacturing data is abundant, its potential for administrative utilization is limited by trust and authority structures rather than the volume of data held.

Gyeongnam received an "Excellent" rating in the 2024 Data-Based Administration evaluation. However, in the same evaluation of 679 institutions nationwide, while the average score for management systems was 77.8 points, data sharing scored 61.8 points and analysis and utilization scored a low 56.3 points. This indicates a structure where an excellent management system does not automatically guarantee actual inter-agency sharing or policy utilization. The detailed scores for Gyeongnam's "Excellent" rating and the disparities among the 18 cities and counties are not fully verifiable in the publicly available data. If the evaluation grades for 2026–2028 become the target, procedures and management systems will be optimized before the actual resolution of industrial problems. Although  an "Excellent" rating was confirmed in the evaluation, the industrial data connectivity performance of the 18 cities and counties is subject to separate verification. ( Ministry of the Interior and Safety Evaluation Results )

The assessment unit of industrial administration is shifting from annual statistics and individual support projects to real-time status changes in companies, supply chains, and regions. When a decrease in orders, changes in electricity usage, withdrawals from employment insurance, a decline in export customs clearances, and business closures are linked with civil complaints, industrial risks can be detected ahead of confirmed statistics. While Gyeongnam possesses the source data for each signal or has the institutions to access it, it lacks evidence combined across the same company, industrial complex, or timeframe. If signal linkage is delayed for two to three years, crises appear only as targets for post-hoc support by individual departments.  The structural change in Administrative AX has moved from a stage of collecting more data to a stage of identifying multiple weak signals as a single risk.

While generative AI rapidly automates individual tasks for public officials, such as writing reports, responding to civil complaints, and searching for data, determining industrial policy requires standardized data and verifiable computational processes. Although Gyeongnam’s AI administration plan includes AI to support public officials' work, the structure for preserving the source, reference date, change history, and basis for judgment of industrial data is separate. A gap exists between document-generating AI and evidence-based industrial administration. If generative AI is deployed first between 2026 and 2028, reports of the same format will be mass-produced from different source materials, making it more difficult to detect errors.  The risk of Gyeongnam’s administrative AX lies not in the lack of AI adoption, but in the AI ​​describing unconnected data as if it were consistent facts.

Manufacturing AI data centers analyze corporate production, quality, and facility data, while administration manages corporate registration, employment, export, and support history. Since these two types of data differ in sensitivity and purpose of use, concentrating them in a single repository increases risks related to security, liability, and participation. There is a gap in equating industrial data connectivity with centralized storage. If a data replication-centric structure remains fixed for two to three years, both security breaches and the costs of managing data recency increase. The appropriate evaluation axis for industrial data connectivity across 18 cities and counties is not data centralization, but the possibility of distributed verification regarding common questions.

While city and county administrations handle the same businesses, roads, farmland, tourist sites, and welfare targets, they differ significantly in organizational scale, data expertise, and system budgets. This structure makes it difficult for Changwon, Gimhae, and Yangsan, as well as county regions, to establish AI and data systems at the same pace. There is a lack of publicly available comparative evidence regarding data quality, update delays, and analysis personnel across cities and counties. If a competition for in-house implementation takes place between 2026 and 2028, larger cities will possess sophisticated data systems, while smaller counties will rely on external solutions.  Administrative AX is a technology that reduces regional disparities, yet it is also a technology that widens administrative gaps depending on differences in readiness.

The Gyeongnam AI Administration Comprehensive Plan consists of three major strategies, 20 tasks, and a budget of 35.1 billion won, and includes an AI-based integrated welfare platform, a report reception system, and a civil servant work support platform. While the service scope of administrative AI has expanded to include welfare, safety, and internal operations, the integrated judgment of industrial data across 18 cities and counties has not been verified as an independent runtime. A gap remains between administrative service AI and industrial policy AI. If service-specific platforms are finalized first for 2026–2028, only APIs and conversion tasks will increase without a common data model.  Currently, AI administration is in the stage of business and service innovation, not yet at the stage of a metropolitan industrial operating system.

The Gyeongnam Manufacturing AI Data Center will be conducted on a scale of 23.3 billion won from 2025 to 2026, and the Changwon National Industrial Complex AX demonstration will be conducted on a scale of 22.2 billion won from 2025 to 2028. The Manufacturing AI Convergence Foundation and the AX Lab also support corporate data analysis and solution development. While manufacturing data analysis hubs are being established, the extent to which benefits have spread to companies in 17 cities and counties outside of Changwon, as well as industry-specific data standards, have not yet been verified. If user companies become concentrated among leading companies in smart factory implementation over the next two to three years, non-digital companies and companies in rural areas will remain outside the data ecosystem.  Although manufacturing AI infrastructure is being built, the results of its diffusion across 18 cities and counties remain unverified.

The Gyeongnam Big Data Hub enables the search, download, and visualization of public data from the province, cities, and counties, and also operates big data analysis, contests, and training programs. While accessibility to public data and points of application exist, it is difficult to track instances where industrial risk detection results led to departmental project changes, budget adjustments, or on-site inspections in a consistent format. A runtime gap remains between analysis and administrative decisions. Even if the number of contests and analysis reports increases between 2026 and 2028, utilization performance cannot be repeatedly verified unless actual changes in decision-making are recorded.  The platform has reached the data provision stage but falls short of the decision tracking stage.

In the Ministry of the Interior and Safety's 2024 evaluation, the average for central administrative agencies was 85.2 points, for metropolitan and provincial governments 85.3 points, and for public enterprises and quasi-governmental institutions 88.2 points, indicating that management infrastructure was generally at a high level. On the other hand, the average for analysis and utilization across all agencies remained at 56.3 points, and data sharing at 61.8 points. The common bottleneck in South Korea's data administration lies not in data possession, but in inter-agency sharing and the actual utilization of data for decision-making. Even Gyeongnam Province  's excellent rating is not evidence that it has automatically escaped this nationwide structure. As evaluation systems and agency platforms expand over the next two to three years, the gap between formal standards and actual interoperability widens. While Gyeongnam currently holds a top-tier domestic management position, it remains unverified as a leading region for industrial data runtime.

Other metropolitan and provincial governments, such as Gyeonggi-do's Data Dream and Chungbuk's Big Data Hub, also provide public data, visualization, civil complaint, traffic, and environmental analysis. Possessing a platform itself has shifted from being a differentiating competitive advantage for metropolitan administration to a basic infrastructure. Gyeongnam's distinctiveness arises when connecting manufacturing data from shipbuilding, defense, aerospace, and machinery with administrative evidence, but the results of corporate data participation and integration remain limited. If the competition remains focused on platform functions between 2026 and 2028, Gyeongnam's unique asset of manufacturing agglomeration will not be transformed into administrative competitiveness.  Gyeongnam's comparative advantage lies not in public data platforms, but in the potential to connect industrial data to policy decisions.

The Seoul Metropolitan Government disclosed that 63% of its employees utilized generative AI by 2025 and revealed budget savings compared to individual subscriptions in the operation of common services. While Gyeongnam is promoting AI administration plans and utilization training, it lacks evidence disclosing civil servant usage rates, reduced work hours, error rates, and improvements in policy decisions using the same criteria. A gap exists between the scale of the plan and the verified results. If usage rates and outcomes are not disclosed for two to three years, AI administration is compared solely based on whether or not projects have been introduced. Although  the scope of Gyeongnam's administrative AI plan is broad, measurable operational performance is still in the early stages. ( Seoul Metropolitan Government )

Gyeongnam is simultaneously pursuing 3,014 smart factories, a manufacturing AI data center worth 23.3 billion won, an AX demonstration industrial complex worth 22.2 billion won, and 20 AI administration projects worth 35.1 billion won. The structure involves separate investments in manufacturing data generation, computing infrastructure, and administrative AI services. However, there is no evidence linking data and performance across projects based on the same company, industry, city, or county. Once these three investment axes are completed independently between 2026 and 2028, even if infrastructure expands, it will be impossible to determine which specific combinations have transformed productivity, employment, and exports. For  Gyeongnam's AX investment, the disconnect in evidence between projects poses a greater risk than the scale of the investment itself.

The initial customized demonstration support for the Manufacturing AI Data Center targets 30 small and medium-sized enterprises (SMEs) in the province, while there are 3,014 companies in Gyeongnam that have established smart factories. These 30 companies represent approximately 1% of the total number of smart factory operators, and the proportion is even lower when considering all manufacturing companies. There is a scale gap between the depth of the initial demonstration and its widespread adoption across the entire industry. If the reproducibility conditions of successful models are not preserved as data over a period of two to three years, the diagnosis, refinement, and development process must be restarted for every subsequent company.  The time-sensitive risk for Manufacturing AI in Gyeongnam lies not in demonstration failure, but in the fact that the cost of repeating the demonstration is not decreasing. ( Hankook Ilbo , Yonhap News )

There are 20 administrative AI projects, 18 basic local governments in Gyeongnam, and major industries are dispersed across various regions. Although the number of projects and the number of cities and counties appear similar, there is no evidence that a single project spreads equally to all of them. A gap is identified where project units are confused with regional units. Once provincial government-centered projects are completed over the next two to three years, it becomes highly likely that cities and counties will rebuild the same functions using separate budgets and systems.  The number of metropolitan AI projects does not reflect the level of diffusion across the 18 cities and counties.

Although Gyeongnam received an excellent grade for data-based administration, the difference between its management system score of 77.8 points and its analysis and utilization score of 56.3 points in the national evaluation was 21.5 points. A structural gap exists between the agency's management capabilities and its actual utilization capabilities. Since detailed scores and distributions by city and county for Gyeongnam are unavailable, it is impossible to determine whether the same gap exists. Even if the excellent grade is maintained from 2026 to 2028, actual AX performance will stagnate unless the number of cases where industrial policy decisions are modified based on data increases.  Evaluation scores reflect preparedness but do not replace the ability to solve industrial problems.

The biggest gap is that the 18 cities and counties cannot identify the same company, business, region, or time in the same way . Data for factory registration, business owners, employment, exports, support projects, civil complaints, and energy each use different numbers, addresses, and reference dates. A wide-area Master ID, which automatically tracks the connection of companies with the same name as well as address changes, corporate divisions, and factory relocations, has not been verified. As the volume of data increases between 2026 and 2028, errors in manual matching and duplicate detection will increase.  The primary gap in the connection of Gyeongnam's industrial data is the common identification system, not the AI ​​model.

The second gap is the timing. GRDP, statistical yearbooks, and corporate status data are updated annually or on specific reference dates, whereas employment, electricity, civil complaints, and exports are on a monthly or near-real-time basis. Combining data with different update cycles without a reference point causes the causes and effects of industrial changes to become intertwined. There is a lack of evidence to verify latency and revision history for each dataset at a glance. If an AI learns from data that has not distinguished time lags for two to three years, it mistakes past states for current risks.  The accuracy of Gyeongnam Administration AX is limited by time-axis alignment rather than the amount of data.

The third gap lies in the record of administrative decisions. While data analysis results and reports remain, records of which judgments the responsible departments changed and how they modified budgets, projects, and inspections are not preserved in a standardized format. Without a connection between evidence and decisions, identical analyses are repeated, and the causes of failure are not accumulated. As AI supporting civil servant tasks becomes more widespread between 2026 and 2028, the volume of document generation will increase, but decision history will become more fragmented.  The biggest operational gap of Gyeongnam Administrative AX is a structure where it is not the data itself, but rather what was changed using that data, that cannot be tracked.

There is the Gyeongnam Big Data Hub, a manufacturing AI data center, 3,014 smart factories, and provincial, city, and county administrative information systems. The basic elements of public data, enterprise data, and computing infrastructure are each established. However, common enterprise identifiers, metadata, authorization, and quality standards connecting them have not been identified. If individual infrastructures are upgraded first between 2026 and 2028, interconnection costs will increase.  Infrastructure readiness is assessed as high, while interoperability readiness is assessed as low.

While provinces and major cities possess specialized organizations for data, ICT, and industry, small counties face limitations in manpower and budget. Evidence comparing the number of data personnel, analytical experience, turnover, and reliance on outsourcing across 18 cities and counties is lacking. Even when the same platform is provided, a gap remains in utilization capabilities that vary by region. If reliance on private solutions expands over the next two to three years, the data structures and knowledge of smaller cities and counties will accumulate with external firms.  Although expertise exists at the provincial level, readiness for expansion at the city and county levels is uneven.

There are existing laws such as the Data-Based Administration Act, public data systems, personal information protection and security regulations, and directions for establishing AI ethics and generative AI security standards in Gyeongnam. While the legal and security framework is being formed, standards regarding consent, purpose, retention period, and result disclosure for using corporate business data for administrative analysis remain separate. An institutional gap persists between public data and industrial secrets. If clear participation conditions are not established between 2026 and 2028, companies will not provide their most sensitive and valuable data.  The public administration data system is rated as upper-middle, while the public-private industrial data trust system is rated as lower-middle.

The Gyeongnam Big Data Hub enables residents and businesses to search for and download public data, and also operates analysis training and contests. While data access has extended to external users, there is a lack of consistent evidence tracking cases where corporate productivity, sales, and employment have improved through the use of open data. A gap remains between data provision and utilization results. If download and view counts remain as representative indicators for two to three years, actual economic effects cannot be verified.  What has been bestowed upon the residents is the right to data access, while the dissemination of problem-solving outcomes remains unconfirmed.

The Manufacturing AI Data Center plans to provide customized analysis, solutions, and demonstrations to 30 initial SMEs. While support reaches the actual sites of these companies, the pathway for expansion to all 3,014 smart factories and non-digital manufacturing enterprises is still in its early stages. There is a gap between supporting leading companies and achieving widespread industrial diffusion. If customized development costs for individual companies are repeated between 2026 and 2028, many SMEs will be unable to operate independently once the support program ends. Although  the on-site application of Manufacturing AI has begun, the level of self-reliance and diffusion remains unverified.

While city and county data is disclosed on the provincial platform, the quality runtime—the time required for companies and residents to request corrections and track the results—vary depending on the dataset. A gap remains where updating and accountability are not consistently displayed after disclosure. If outdated information on companies and facilities—dated by two to three years—is used for AI search and policy analysis, incorrect administrative targets are repeatedly generated.  Although data disclosure has increased, user control over data quality remains limited.

Changwon is home to a concentration of national industrial complex AX demonstrations and manufacturing AI data centers, while Gimhae, Yangsan, Geoje, and Sacheon also possess large-scale industrial enterprises. In contrast, rural areas are centered on data related to agriculture, tourism, small-scale manufacturing, and lifestyle services, and lack real-time equipment data and specialized personnel. This structure translates into differences in data generation volume and analytical capabilities. If manufacturing-centric AX becomes the standard for the entire Gyeongnam region within two to three years, seasonal, climate, and resident population data from agricultural and fishing villages will be relegated to a lower priority.  The administrative AX gap in Gyeongnam widens not only based on city size but also on the extent of the digital footprints left by industries.

Data from 18 cities and counties differs in categories, file formats, update cycles, and scope of disclosure, even within the same field. Directly comparing detailed data from large cities with annual aggregated data from rural areas results in varying analytical precision. Regions with abundant data may be overrepresented as having many problems or high performance. If the AI ​​analysis for 2026–2028 learns to interpret this difference as actual regional disparities rather than differences in data volume, resource allocation bias will become entrenched.  The data gap among cities and counties goes beyond a statistical quality issue and poses a risk of spatial bias in AI administration.

For models developed at the Changwon manufacturing AI hub to be applied to Geoje shipbuilding, Sacheon aviation, Gimhae machinery, and military food processing, retraining is required due to differences in processes, equipment, and data structures. Evidence proving the potential for wide-area diffusion of a single successful model is still limited. A gap remains between hub concentration and regional application. If reliance on the central hub increases between 2026 and 2028, regional problem definitions and field knowledge will remain outside the scope of the model.  The spatial diffusion of manufacturing AI in Gyeongnam is more likely to be stalled by industry-specific data translation rather than computing access.

The Gyeongnam Manufacturing AI Data Center will be established in stages from 2025 to 2026, the Manufacturing AI Convergence Infrastructure from 2024 to 2026, the AX Demonstration Industrial Complex from 2025 to 2028, and 20 AI administration projects starting in 2025. This is a phase where major data and AI systems are designed, built, and disseminated within the same two to three-year period. There is currently no evidence that common identifiers and metadata have been aligned in advance. Once system-specific data models are finalized between 2026 and 2028, subsequent connections will become complex transformation projects rather than new development.  Currently, the "Golden Time" is not the period for AI adoption, but rather the time when data structures remain fixed for an extended period.

Data from 3,014 smart factories is already accumulated within company-specific solutions and equipment, while new manufacturing AI businesses generate additional data. If the format and ownership of historical data cannot be linked to the current system, the continuous timeline used for AI training is severed. There is a lack of evidence regarding data succession between existing smart factories and new AX businesses. If only data from new businesses is standardized over the next two to three years, the smart factory data, into which years of investment have been made, will remain as a disconnected asset. The irreversibility of the Golden Time lies not in losing data, but in losing the link that allows for comparison between the past and the present.

Provinces, cities, and counties are introducing generative AI, business support AI, and civil complaint, welfare, and safety services, respectively. Once AI begins generating administrative documents, documents that do not automatically preserve their sources and decision histories will rapidly accumulate. Currently, the methods for linking document management with the evidence of AI-generated content have not been sufficiently disclosed. If AI documents without sources are incorporated into the record system between 2026 and 2028, it will be difficult to restore accountability for policy decisions thereafter. The end condition of the "Golden Time" is not the completion of AI services, but the moment unverifiable administrative records become commonplace.

12-1. Golden Time Application Case in Basic Local Governments ① — Changwon City

Changwon is home to a concentration of 22.2 billion won for the National Industrial Complex AX demonstration project, 23.3 billion won for the Manufacturing AI Data Center, numerous smart factories, and supply chains for defense, machinery, and nuclear power. It is the region within the province where industrial, computing, and corporate data can intersect most densely. However, there are currently no public outcomes that connect prime contractor orders, partner company production, employment, power, and policy support data at the corporate level. If the infrastructure remains limited to individual company demonstrations between 2026 and 2028, Changwon's model will not spread to supply chains in other cities and counties.  Changwon's "Golden Time" is determined not by the completion of the data center, but by whether the time it takes for changes in prime contractors to be transmitted to partner companies, employment, and the local economy can be determined early.

12-2. Golden Time Application Cases in Basic Local Governments ② — Sacheon-si and Jinju-si

Sacheon is home to the Aerospace Administration, the National Aerospace Industrial Complex, and KAI, while Jinju possesses university, research, education, and administrative functions. The structure is such that corporate, talent, and research data within the aerospace ecosystem are generated across the two cities and adjacent counties. However, there is a lack of evidence linking data on corporate registration, R&D, recruitment, housing, education, and partner companies into a common industrial zone. If Sacheon and Jinju establish separate systems and performance indicators between 2026 and 2028, the same aerospace ecosystem will be administratively separated into two distinct industries. The "  Golden Time" for Sacheon and Jinju is not the period for collecting data in one place, but rather the period during which the two administrations evaluate the same companies, talent, and projects based on the same evidence.

Once manufacturing AI data centers, smart factories, administrative AI, and city/county platforms are finalized with distinct identifiers and data models, subsequent connections involve the repetitive process of source data maintenance, APIs, transformations, and manual matching. This structure results in long-term accumulated interoperability costs exceeding the system construction costs. Currently, the total connection costs and the number of duplicate systems remain unconfirmed. If individual optimization continues for two to three years, regional integration becomes irreversible not due to technical issues, but due to conflicts over budget, responsibility, and contracts.  What is being overlooked is not the data, but the option to connect systems at a lower cost in the future.

Companies submit the same information to multiple agencies for repetitive support applications, performance reports, and status surveys. Without common corporate identification and consent-based data reuse, the administrative burden on companies increases as AX projects expand. There is no evidence regarding duplicate submission times per company or duplicate collection costs per agency. If duplicate requests expand between 2026 and 2028, data quality will deteriorate due to fatigue and perfunctory input.  What is currently being lost is the trust in corporate data participation.

If the results of administrative analysis are not linked to actual changes in budgets, projects, or inspections, the causes of success and failure are not conveyed to the next person in charge. While AI can quickly search past documents, it cannot restore unrecorded judgments or on-site context. During the two to three years of repeated civil servant transfers and the termination of outsourced projects, decision knowledge leaks out of the organization.  The irreversible loss for Gyeongnam's administration AX is organizational memory, where judgments are not accumulated, rather than data files.

Gyeongnam simultaneously possesses a Big Data Hub, 3,014 smart factories, a Manufacturing AI Data Center, AI administrative plans, and public data from 18 cities and counties. While the source assets for data connectivity already exist, they have not yet been verified by the same standards. If connectivity standards are established before the completion of the construction project between 2026 and 2028, a comparable timeline will remain without the need to rebuild the existing infrastructure.  The greatest asset currently available is not a new platform, but the interoperability of existing data.

Gyeongnam's industries—including shipbuilding, defense, aerospace, machinery, and automotive parts—are interconnected through supply chains, meaning that changes in orders and employment within a single industry are transmitted to multiple cities and counties. Currently, this propagation path is observed only partially in administrative data. Within two to three years, connecting the same companies, supply chains, and regions will establish the first baseline for comparing the conditions before and after industrial shocks.  The industrial asset currently available is the actual map of Gyeongnam's supply chain, rather than an AI model.

Since AI services for the province, cities, and counties are still being built in stages, there is still time to apply a common structure for recording sources, judgments, decisions, and outcomes. The level of standardization in actual decision history is low. Once historical data is accumulated between 2026 and 2028, it will be possible for the first time to compare which analyses and projects have changed corporate, employment, and regional outcomes.  The administrative asset currently available is not automated documents, but the continuity of verifiable decisions.

The AX response is not a structure that replicates data from 18 cities and counties into a single central repository.  It is a runtime that connects common identifiers → metadata → distributed analysis → supply chain determination → administrative decision → outcome, while keeping sensitive original corporate data under the control of the holding institution and the company.

The first application axes are the defense and machinery supply chain in Changwon and the aerospace industrial zones in Sacheon and Jinju. Signals regarding prime contract wins, partner company production and employment, smart factory and AI support, power usage, and business closures are linked on a company- and monthly basis, but only aggregated risks by city, county, and industry are disclosed externally.

The failure condition is not a decrease in the dataset. Although the platform's data and AI services are increasing, the same company is counted under multiple numbers, reference dates differ by city and county, and the source of AI judgments and actual business changes are not recorded.

Data Infrastructure Opportunity + Interoperability Lock-in Risk

Gyeongnam is not a region lacking data. It is a region where the same companies and industries are recorded by different institutions, numbers, and time points, making it impossible to connect them as a single change.

The current risk lies not in the delay in AI adoption, but in the fact that manufacturing AI, administrative AI, and city/county information systems are being completed separately without connection standards.

The period from 2026 to 2028 is a Golden Time when Gyeongnam's industrial data structure remains fixed for an extended period. Integration after this period will become a conversion and re-verification project that is more expensive than new construction.

Final verdict: Public data infrastructure is high, manufacturing data potential is high, interoperability among 18 cities and counties is low, trust in corporate participation is medium-low, the connection between Evidence and Decision is low, and technical irreversibility is in progress.

The verification sequence is Identification → Standardization → Combination → Analysis → Judgment → Administrative Decision → Field Outcome → Re-verification . If any one of the number of datasets, the number of AI tasks, or platform usage increases, it is not determined as Administrative AX.

Runtime Chain

Data Identity → Interoperability → Industrial Signal → Administrative Decision → Verified Outcome

Structural insights remaining from this analysis

The structural problem with connecting industrial data across 18 cities and counties does not lie in the fact that the data is located in multiple places. Factory registrations, employment, exports, support projects, energy, and civil complaints are originally generated by different agencies. The problem lies in the fact that each agency records the same company and the same event under different names, numbers, addresses, and reference dates.

Therefore, linking industrial data differs from the task of copying all sources to a single location. Even if sources are centralized, if identical companies cannot be identified and timelines cannot be aligned, the files will simply grow larger. Conversely, if common identifiers, metadata, and judgment questions match, supply chain risks can be aggregated without moving sensitive corporate data.

The competitiveness of Gyeongnam Administrative AX is determined not by the amount of data owned, but by the ability of 18 cities and counties to judge the same changes of the same company using the same evidence . Without this capability, AI summarizes 18 incomplete records more quickly, but once this capability is established, dispersed records also function as a single industrial runtime.

Golden Time Thesis — Gyeongnam’s Golden Time from 2026 to 2028 is not a period for collecting more data from 18 cities and counties in one place, but rather a period for identifying the same companies, businesses, regions, and times identically and recording whether that evidence led to actual administrative decisions and outcomes. Once this connection is formed, Gyeongnam’s decentralized industries become a single administrative AX supply chain. If it is not formed, as manufacturing AI and administrative AI become more advanced, the costs of connecting them and judgment errors increase together.