2026.03 · Pebblous Data Communication Team
Reading time: ~20 min · 한국어
Executive Summary
On March 4, 2026, Korea's Presidential National AI Strategy Committee (국가인공지능전략위원회) declared 2026 as the "AI G3 Leap Year" and unveiled a total AI budget of 9.9 trillion KRW (~$7.1B USD) across 41 government ministries and 741 AI-related programs — roughly tripling the previous year's spending. MSIT (과기정통부) accounts for 5.1 trillion KRW (51%) and MOTIE (산업통상부) for 1.7 trillion KRW (17%). This report analyzes the accompanying Budget Program Guide (533 projects, 5,296 pages) and identifies 25 key projects that directly align with Pebblous's core technologies.
Our methodology involved keyword-based filtering (data, synthetic, LLM, agent, Physical AI, quality, preprocessing, manufacturing AI, etc.) across the table of contents of the 533 projects in the Budget Program Guide, followed by an in-depth evaluation of 94 relevant project descriptions on a 5-tier suitability scale. DataClinic is applicable to 12 projects, PebbloSim to 8, and Data Greenhouse to 6.
The combined budget of the top 7 priority projects amounts to approximately 50 billion KRW (~$36M USD), with MOTIE's manufacturing AI project cluster as the primary target. 15 out of 17 key projects are newly launched for 2026, requiring immediate action on consortium formation and proposal preparation.
1. Analysis Overview
Korea's Presidential National AI Strategy Committee designated 2026 as the "AI G3 Leap Year" and allocated a total of 9.9 trillion KRW across 41 ministries for AI programs (CBS NocutNews, 2026.03.04). Key allocations include 2.1 trillion KRW for AI computing resources, 300 billion KRW for DeepTech/AI startup funds, and 600 billion KRW for the AX Sprint Project. This analysis covers the Budget Program Guide (533 projects, 5,296 pages) released alongside the announcement.
Pebblous Core Capabilities
Pebblous is the only player that integrates Data OS + Quality Assessment + Simulation Generation into a single platform. This differentiates us from NVIDIA (infrastructure level), Applied Intuition (autonomous driving-specific), and MOSTLY AI (no physics simulation).
| Technology Area | Solution | Key Capabilities |
|---|---|---|
| Data Quality Management | DataClinic | Data diagnosis, cleansing & quality certification (Data Imaging, ISO/IEC 5259) |
| Synthetic Data Generation | PebbloSim | Synthetic data for Physical AI (Neuro-Symbolic Hybrid World Model) |
| Autonomous Data Operations | Data Greenhouse | Agentic AI-based autonomous data operating system |
| Data Analytics | PebbloScope | Data visualization and analytics tool |
Our existing clients include global companies such as Hyundai Motor, LG Electronics, and LG Uplus. Pebblous was selected as the lead company for MSIT's (Ministry of Science and ICT, 과학기술정보통신부) 'Global Big Tech Development Program' (글로벌 빅테크 육성 사업), currently executing a 3-year, 6 billion KRW (~$4.3M USD) project. In 2025, we achieved 115% year-over-year revenue growth and over 50% SG&A reduction.
Methodology
We first filtered the 533 projects in the Budget Program Guide by keywords (data, synthetic, LLM, agent, Physical AI, quality, preprocessing, manufacturing AI, etc.), extracted the full descriptions of 35 highly relevant projects, and then identified an additional 59 missed projects through supplementary analysis, finalizing a total of 25 key projects.
2. Top Priority Projects (5-Star)
These 7 projects directly align with Pebblous's core technologies and promise high synergy upon participation.
2-1. Data Preprocessing Automation for Industrial AI (산업AI용 데이터 전처리 자동화 기술개발)
This project develops preprocessing automation technology to make industrial data AI-trainable. Key deliverables include a low-code/no-code preprocessing automation process modeler and preprocessing libraries for specific industrial domains (safety, facilities, quality, energy/environment, process logistics).
Fit Rationale: DataClinic's core data diagnosis and cleansing capabilities are precisely what preprocessing automation requires, and Data Greenhouse's autonomous data operations feature perfectly matches the "automation" keyword. Validation through existing manufacturing clients like Hyundai Motor and LG is also feasible.
2-2. Medical Data Synthesis & Digital Medical Product Development (의료데이터 합성기술 및 디지털의료제품 개발)
A long-term project to solve medical data privacy issues (security, personal information protection, medical law) through synthetic data generation technology and develop digital medical products using the results.
Fit Rationale: PebbloSim's Neuro-Symbolic Hybrid World Model is ideally suited for generating "synthetic data that preserves disease attributes." The symbolic simulation ensures medical accuracy through logical consistency, while the neural generation model produces realistic representations. Generating medical synthetic data free of Physical Hallucination is PebbloSim's core differentiator.
2-3. Manufacturing AI Data Value Chain Development (제조AI 데이터 벨류체인 구축)
A project supporting data collection, evaluation, and certification for traditional manufacturing sectors including casting, mold-making, precision machining, heat treatment, and welding. Core goals include establishing AI training dataset quality metrics and manufacturing data standardization/compatibility verification.
Fit Rationale: Pebblous's core philosophy that "data quality determines AI performance" is identical to this project's raison d'etre. Building quality evaluation and certification systems using DataClinic and applying automatic labeling technology are directly applicable.
2-4. Manufacturing-Specialized Large-Scale AI Service Development & Validation (제조특화 초거대 제조AI 서비스 개발 및 실증)
A project to build manufacturing data collection systems and develop/validate large-scale manufacturing AI models for manufacturing and partner companies within industrial complexes.
Fit Rationale: The performance of large-scale AI models is directly dependent on training data quality. As an industrial complex-based project directly related to the Hyundai Motor supply chain, consortium formation leveraging existing client relationships is highly feasible.
2-5. Thermal Process-Specialized Manufacturing AI Foundation Model (열공정특화 제조AI 파운데이션 모델 개발)
An AI foundation model development and validation project for thermal process manufacturing (casting, welding, heat treatment, etc.), featuring a relatively large budget.
Fit Rationale: Foundation model development requires large-scale, high-quality training data, and synthetic data augmentation (PebbloSim) is particularly effective for thermal process data characteristics. Integration with thermal process companies within Hyundai Motor's industrial complex is a natural fit.
2-6. AI Autonomous Manufacturing SDM Platform Development (AI 자율제조 SDM 플랫폼 기술개발)
A project to develop an autonomous manufacturing Smart Digital Manager (SDM) platform, building data-driven autonomous decision-making systems for manufacturing processes.
Fit Rationale: The concept of Data Greenhouse (autonomous data operating system) closely parallels the SDM platform. Real-time sensor data quality management and autonomous data cleansing can serve as core SDM modules.
2-7. Manufacturing Data Standards & AI-Enabled Carbon Neutrality (제조데이터 표준·AI 활용 탄소중립 지원 기술개발)
A project for developing a product lifecycle Manufacturing Carbon Footprint (MCF) platform and AAS (Asset Administration Shell)-based core process carbon reduction technology.
Fit Rationale: Carbon footprint calculation accuracy depends entirely on manufacturing data quality. AAS standard-based data standardization falls within DataClinic's scope, and the project is directly linked to the automotive industry's EU carbon regulation compliance, creating strong synergy with Hyundai Motor.
3. Highly Recommended Projects (4-Star)
Projects with strong relevance to Pebblous technologies, accessible through consortium participation or service provision.
| Project Name | Ministry | Budget | Pebblous Opportunity |
|---|---|---|---|
| World Best LLM Data Utilization Support (World Best LLM 데이터 활용 지원) | MSIT (과기정통부) | 30B KRW | LLM training data quality verification, 95% quality achievement service |
| Industrial AI Agent Technology Development (산업AI 에이전트 기술개발) | MOTIE (산업통상부) | 6B KRW | Directly aligned with Data Greenhouse Agentic AI technology |
| Virtual Convergence-Based Physical AI Core Technology (가상융합기반 피지컬AI 핵심기술) | MSIT (과기정통부) | 5.06B KRW | PebbloSim synthetic data, virtual-to-real validity evaluation |
| Collaborative Intelligence Physical AI SW Platform (협업지능 피지컬AI SW플랫폼) | MSIT (과기정통부) | 76.7B KRW | Data quality management subsystem participation |
| National Data Infrastructure Data-X (국가 데이터 인프라 Data-X) | MSIT (과기정통부) | 7B KRW | Data space quality standards & certification mechanisms |
| AI Innovation Safe Data Utilization (AI혁신 데이터 안전 활용 지원) | MSIT (과기정통부) | 5B KRW | Synthetic data generation within safety zones |
| Data-Driven Industrial Competitiveness Enhancement (데이터기반 산업경쟁력 강화) | MSIT (과기정통부) | 11.96B KRW | Core company for data processing & quality management infrastructure |
| AI Integrated Voucher (AI통합바우처) | MSIT (과기정통부) | 80.4B KRW | Register as data quality management service provider |
| Industrial AI Solution Validation & Deployment (산업AI 솔루션 실증·확산) | MOTIE (산업통상부) | 12.8B KRW | Validation participation with existing clients |
4. Watchlist Projects (3-Star)
Projects without direct technology match but with potential for Pebblous technology utilization. Includes large-scale newly launched projects discovered through supplementary PDF analysis.
| Project Name | Ministry | Connection Point |
|---|---|---|
| AI AGENT Leading Nation Initiative (AI AGENT 선도 국가 사업) | MSIT (과기정통부) | Data Greenhouse, agent technology |
| Human-AI Collaborative LAM Development & Global Validation (인간-AI 협업형 LAM 개발·글로벌 실증) | MSIT (과기정통부) | Large-scale new project (66.7B KRW), data-driven LAM |
| Physical AI Leading Technology Development (피지컬AI 선도기술개발) | MSIT (과기정통부) | PebbloSim Physical AI synthetic data |
| Multi-Institutional Multimodal Federated Learning Medical AI (다기관-멀티모달 연합학습 의료AI) | MOHW (보건복지부) | Medical synthetic data demand (~9B KRW) |
| AI Safety Trustworthy AI Technology Development (AI Safety 신뢰 AI 기술개발) | MSIT (과기정통부) | Data quality = AI safety (7.95B KRW) |
| AX Innovation Enterprise Creative Technology Development (AX 혁신기업 창의 기술개발) | MSIT (과기정통부) | New, targeting AX innovation companies (7.5B KRW) |
| AI Open Source Ecosystem Development (AI 분야 오픈소스 생태계 조성) | MSIT (과기정통부) | New (10B KRW), data tools |
Additional Discovery Areas (59 Projects)
A comprehensive re-review of the full table of contents uncovered 59 additional related projects that were missed in the initial analysis. Key areas include:
Autonomous Driving & Automotive
Direct application of PebbloSim synthetic data, Hyundai Motor linkage
Robotics & Autonomous Agents
PebbloSim robot simulation training data
Digital Twin
PebbloSim + DataClinic simulation
Healthcare, Bio & Privacy
Solving privacy challenges through synthetic data
5. Pebblous Technology-to-Market Mapping
Analysis of the number of applicable projects per solution shows DataClinic with the broadest coverage.
12 Projects
DataClinic
Manufacturing data preprocessing, quality evaluation, standardization, certification. MOTIE manufacturing AI project cluster is the primary target.
8 Projects
PebbloSim
Medical data synthesis, Physical AI synthetic data, LLM data augmentation. Surging demand for synthetic data.
6 Projects
Data Greenhouse
Agentic AI, autonomous manufacturing (SDM), industrial AI agents. Projects with the "autonomous" keyword.
5 Projects
PebbloScope
Data visualization & analytics. Supplementary tool in large data infrastructure projects (Data-X, big data construction).
6. Budget-Ranked Project Overview
Of the 17 key projects, 15 are newly launched in 2026 and only 2 are continuing programs (WBL, Manufacturing-Specialized Large-Scale AI). This means open call-for-proposal opportunities are available. The table below is sorted by budget size.
| Budget | Project Name | Rating | New/Cont. |
|---|---|---|---|
| 80.4B KRW | AI Integrated Voucher (AI통합바우처) | 4-Star | New |
| 76.7B KRW | Collaborative Intelligence Physical AI SW Platform (협업지능 피지컬AI SW플랫폼) | 4-Star | New |
| 30B KRW | World Best LLM Data Utilization Support (World Best LLM 데이터 활용 지원) | 4-Star | Cont. |
| 12.8B KRW | Industrial AI Solution Validation & Deployment (산업AI 솔루션 실증·확산) | 4-Star | New |
| 11.7B KRW | Thermal Process Manufacturing AI Foundation Model (열공정특화 제조AI 파운데이션 모델) | 5-Star | New |
| 11.96B KRW | Data-Driven Industrial Competitiveness Enhancement (데이터기반 산업경쟁력 강화) | 4-Star | New |
| 9.2B KRW | AI Autonomous Manufacturing SDM Platform (AI 자율제조 SDM 플랫폼) | 5-Star | New |
| 7.68B KRW | Manufacturing-Specialized Large-Scale AI Service (제조특화 초거대 제조AI 서비스) | 5-Star | Cont. |
| 7.21B KRW | Manufacturing Data Standards & Carbon Neutrality (제조데이터 표준·탄소중립 기술) | 5-Star | New |
| 7B KRW | National Data Infrastructure Data-X (국가 데이터 인프라 Data-X) | 4-Star | New |
| 6B KRW | Industrial AI Agent Technology Development (산업AI 에이전트 기술개발) | 4-Star | New |
| 5.06B KRW | Virtual Convergence Physical AI Core Technology (가상융합 피지컬AI 핵심기술) | 4-Star | New |
| 4B KRW | Manufacturing AI Data Value Chain (제조AI 데이터 벨류체인 구축) | 5-Star | New |
| 3.2B KRW | Data Preprocessing Automation for Industrial AI (산업AI용 데이터 전처리 자동화) | 5-Star | New |
| 2.8B KRW | Medical Data Synthesis Technology (의료데이터 합성기술) | 5-Star | New |
7. Strategic Action Plan
A 4-phase strategy for executing against the 25 key projects on a timeline basis.
Phase 1: Immediate ~ 1 Month
Priority Project Selection & Strategy Development
- Finalize 3-4 core targets from the 7 five-star projects
- Pre-confirm joint participation intent with Hyundai Motor & LG
- Contact program officers at implementing agencies (KIAT, NIA, etc.)
Phase 2: 1~3 Months
Consortium Formation & Proposal Preparation
- Sign MOUs with existing clients (Hyundai Motor, LG)
- Secure university and research institute partners
- Draft technical proposals (DataClinic + PebbloSim focus)
Phase 3: 3~4 Months
Proposal Submission & Business Development
- Finalize proposals aligned with each project's call-for-proposal schedule
- Budget estimation and allocation planning
Phase 4: Post-Award
Project Execution & Expansion
- Deploy core personnel
- Leverage project outcomes for additional program entry (vouchers, validation programs, etc.)
8. PebbloSim Technical Differentiators & Project Matching
A summary of how PebbloSim's technical differentiators, confirmed through blog analysis, match government project requirements.
Neuro-Symbolic Hybrid World Model
Medical data synthesis (disease attribute preservation), Physical AI (physics law compliance)
Vector-to-Param
Precision targeting of data gaps -- supplementing sparse data for manufacturing AI foundation models
Multimodal Synchronization (RGB, Depth, LiDAR, Radar)
Autonomous driving and robot training data generation
Physical Hallucination Elimination
AI Safety, trustworthy AI validation
Digital Twin Engine
Applicable to all 5 digital twin-related projects
EU AI Act / ISO 42001 Audit Evidence
Manufacturing data standards & carbon neutrality (EU regulatory compliance)
Frequently Asked Questions (FAQ)
What is the total scale of Korea's 2026 AI budget?
Korea's 2026 AI budget totals 9.9 trillion KRW (~$7.1B USD), roughly tripling the previous year. 741 AI programs span 41 government ministries, with MSIT accounting for 5.1 trillion KRW (51%) and MOTIE for 1.7 trillion KRW (17%). Of the 533 projects detailed in the Budget Program Guide, 25 directly align with Pebblous's core technologies.
How many projects can Pebblous participate in?
Our in-depth analysis identified 7 top priority (5-star), 9 highly recommended (4-star), and 7 watchlist (3-star) projects, totaling 25 key projects. Additionally, we discovered 59 related projects with potential opportunities.
Which projects are the best fit for DataClinic?
DataClinic is applicable to 12 projects. MOTIE's "Data Preprocessing Automation for Industrial AI" (산업AI용 데이터 전처리 자동화 기술개발) and "Manufacturing AI Data Value Chain Development" (제조AI 데이터 벨류체인 구축) are the most direct matches, as their core requirements align with data diagnosis, cleansing, and certification.
Which government projects are suitable for PebbloSim?
PebbloSim is applicable to 8 projects. Key targets include Medical Data Synthesis Technology (MOTIE), Virtual Convergence-Based Physical AI Core Technology (MSIT), and projects related to autonomous driving, robotics, and digital twins. The Neuro-Symbolic Hybrid World Model's Physical Hallucination elimination is the core differentiator.
What is the ratio of new vs. continuing programs?
Of the 17 key projects, 15 are newly launched programs for 2026, and only 2 are continuing (World Best LLM, Manufacturing-Specialized Large-Scale AI). The overwhelming majority being new programs means call-for-proposal opportunities are wide open.
Which project has the largest budget?
The AI Integrated Voucher (AI통합바우처) at 80.4B KRW is the largest, followed by Collaborative Intelligence Physical AI SW Platform (협업지능 피지컬AI SW플랫폼) at 76.7B KRW, and World Best LLM Data Utilization Support at 30B KRW. Among 5-star projects, the Thermal Process Manufacturing AI Foundation Model (열공정특화 제조AI 파운데이션 모델) has the largest budget at 11.7B KRW.
What source material was this analysis based on?
This analysis comprehensively examined the "2026 AI Fiscal Programs Status" document (533 projects, 5,296 pages) published by Korea's National AI Strategy Committee (국가인공지능전략위원회). A three-stage methodology was applied: keyword-based primary filtering, in-depth extraction analysis of relevant project descriptions, and comprehensive TOC re-review for supplementary discovery.
References
- National AI Strategy Committee (국가인공지능전략위원회), "2026 AI Budget Programs Guide" (533 projects, 5,296 pages), released 2026.03.05
- "AI budget 9.9 trillion KRW this year... government releases 741 programs", CBS NocutNews, 2026.03.04