Korea’s AI Voucher Programme (AIVP) supports the adoption of artificial intelligence by SMEs, medical institutions and micro-businesses, while also fostering the growth of domestic AI solution providers. Overseen by the Ministry of Science and ICT and administered by the National IT Industry Promotion Agency, the programme operates through a central digital platform that manages supplier registration, demand-supply matching, application procedures, evaluation and project monitoring. Since its launch in 2020, it has supported more than 1 000 projects and progressively expanded its scope to cover a wider range of firms and delivery tracks. Evaluation evidence points to positive effects on employment, sales and firm performance, particularly in service sectors. The programme highlights the value of combining flexible cost-sharing, centralised digital delivery and supplier-market development to support AI adoption among smaller firms.
Korea’s AI Voucher Programme
Abstract
|
|
Objectives |
|
|
|
Delivery arrangements |
|
|
|
Budget |
|
|
|
Outreach |
|
|
|
Targets |
|
|
|
Closest UK Counterpart |
|
Introduction
Korea’s AI Voucher Programme (AIVP) is a government-led policy programme that supports the widespread adoption of artificial intelligence (AI) solutions among small and medium-sized enterprises (SMEs), medical institutions, and micro-businesses, with the dual objective of lowering adoption barriers and creating new market opportunities for AI developers. The programme was first introduced in Korea’s National AI Strategy in 2019 and launched in 2020, with the goal of fostering the growth of AI providers and accelerating the digital transformation of industry in line with the rapid pace of AI-driven change.
Delivery arrangements
The AIVP is overseen by the Ministry of Science and ICT (MSIT) and administered by the National IT Industry Promotion Agency (NIPA), which serves as the dedicated implementing agency. Its basic structure is that the government allocates vouchers to selected demand-side firms (buyers), which use them to purchase AI solutions from registered supply-side firms (vendors); the supply firms then request reimbursement from NIPA. The programme operates on an annual basis, with projects typically supported for a period of around seven months, placing emphasis on facilitating the initial adoption of AI solutions. Centrally, implementation is managed through NIPA’s SMART platform, which supports the full project cycle, including supplier registration, demand-supply matching, calls for applications, evaluation of consortia, funding agreements, implementation monitoring, and outcome tracking.1
The delivery structure, centred on a digital platform that matches demand and supply firms, shares similarities with the Smart Manufacturing Innovation Initiative (SMII). However, unlike SMII, where regional centres act as field-level delivery agents and local governments co-lead certain projects, in AIVP their role is largely limited to outreach and promotion. This reflects a design difference: SMII often involves customised solution development, whereas AIVP focuses on purchasing and applying pre-developed AI solutions. To bridge potential gaps with firms in the field, complementary measures are in place. In principle, demand firms are expected to select supply firms directly from the registered pool and form a consortium before applying to the programme. For firms that face difficulties in this process, NIPA provides matching support by recommending suitable suppliers and operates a dedicated call centre to assist applicants throughout the entire procedure.
At its inception, the AIVP did not distinguish between different programme tracks.2 Reflecting the fast-changing AI technology and market landscape, the programme has continually adjusted its structure and eligibility rules to broaden participation and strengthen its policy impact. In 2023, the scheme was restructured into five tracks: General, Medical, AI Semiconductor, Micro-business, and Global, and by 2025 the General and Medical tracks were integrated, leaving four operational tracks: General, AI Semiconductor, Micro-business, and Global, each with distinct objectives and delivery arrangements.
In 2025, while the AIVP follows a common structure in which demand firms adopt AI solutions from registered supply firms, its four tracks are clearly differentiated in their objectives and operational modalities.
The General track is the most inclusive, covering SMEs, mid-sized firms, and medical institutions across diverse industries. Its primary aim is to stimulate broad-based AI adoption, with government funding concentrated on the application of registered solutions, and evaluation criteria focusing on firms’ readiness and commitment to digital upgrading.
The AI Semiconductor track is distinct in requiring consortia to include not only demand firms and solution providers but also domestic AI semiconductor producers and cloud service operators. With higher project ceilings (up to KRW 230 million – approximately USD 159 300 – compared to a maximum of KRW 200 million under the other tracks, and including earmarked support for semiconductor infrastructure), this track is designed less for diffusion than for validating AI services based on indigenous semiconductor technologies.
The Micro-business track targets very small firms and sole proprietors. Here, supply firms apply on behalf of groups of at least ten micro-businesses, delivering turnkey solutions that include hardware. Micro-businesses are exempted from matching contributions, while supply firms bear the private share of costs.
Finally, the Global track is oriented towards internationalisation. In this track, supply firms apply independently, with overseas partners serving as demand clients. Projects must demonstrate verified contracts and are required to deliver internationally recognised outputs such as proof-of-concept pilots, certifications, or clinical trials.
Source: National IT Industry Promotion Agency (NIPA), adapted by the authors (2025)
Budget
The AIVP began in 2020 with a government budget of KRW 59.9 billion (USD 41.5 million), followed by KRW 56.0 billion (USD 38.8 million) in 2021, and KRW 98.0 billion (USD 67.9 million) in 2022. Budgets then decreased to KRW 70.0 billion (USD 48.5 million) in 2023 and KRW 42.5 billion (USD 29.4 million) in 2024, before reaching KRW 27.6 billion (USD 19.1 million) in 2025. In other words, the programme’s budget increased from around KRW 60 billion at launch to nearly KRW 100 billion in 2022, before being substantially scaled back to less than half of its peak level by 2024. According to discussions during the National Assembly audit of ICT-related agencies at the end of 2023, the government assessed that, after more than four years of implementation, the AIVP had largely achieved its initial market-creation objectives. Against the backdrop of broader fiscal constraints, the budget was therefore reduced, while new funding was allocated to other AI-related initiatives, including programmes focusing on large-scale (foundation) AI models and alternative pathways for AI adoption diffusion.
The programme generally follows a co-funding principle in which the government covers up to 80% of total project costs for SMEs and 70% for mid-sized firms, with the remainder covered by participants. For firms led by young entrepreneurs aged under 39, or for regionally based firms, the private share is further reduced to lower entry barriers. Matching requirements can be met through both cash and in-kind contributions, allowing SMEs to reduce their effective cash burden to as little as 2-5% of project costs. Funding ceilings vary across tracks, and in most cases, demand firms can receive vouchers of up to KRW 200 million per project (approximately USD 138 500). In the AI Semiconductor track, the ceiling is higher, at KRW 230 million (approximately USD 159 300), with KRW 30 million earmarked specifically for semiconductor infrastructure. Project budgets are subject to further adjustment following detailed review and approval by NIPA.
Outreach
Since its launch in 2020, the AIVP has supported a growing number of firms while steadily expanding the scope of eligible participants. In 2020, the programme funded 225 projects with participation restricted to SMEs, and the number of registered supply firms reached 553. In 2021, the scope of beneficiaries was broadened to include mid-sized firms, with 207 projects selected and the supplier pool expanding to 991 firms. In 2022, the programme provided its largest support to date, funding 350 projects including 75 projects in a newly established medical track, while the number of registered supply firms grew to 1 553. By 2023, eligibility was further expanded to include micro-businesses and overseas partners, with 251 projects supported and the supplier pool exceeding 2 093 firms.
In 2025, the programme budget is set at KRW 27.6 billion, with approximately 130 projects expected to be selected. Over the five years since its inception, the AIVP has supported more than 1 000 projects across SMEs, mid-sized firms, medical institutions, micro-businesses, and global partners, while also fostering the growth of domestic AI providers through the rapid expansion of the registered supplier pool.
Independent external evaluation
An impact evaluation conducted by the Gyeonggi Research Institute (GRI) 3 provides robust empirical evidence on the effects of the AI Voucher Programme. The study built a firm-level panel dataset by linking information on AI voucher beneficiaries with the Korea Data Exchange (KDX) database of companies operating in-house R&D laboratories, which contains detailed financial and employment records. This dataset includes annual employee headcounts from 2015 to 2024 and financial information such as sales from as early as 2001, enabling long-term trend analysis. Methodologically, the study applied difference-in-differences (DID) and two-way fixed effects (TWFE) models to compare beneficiary firms with non-participating firms, controlling for firm heterogeneity and common time trends.
Key findings are as follows:
Employment: Beneficiary firms experienced an average increase of about 6 percentage points in employment, statistically significant and persisting up to three years post-adoption.
Sales and productivity: Sales growth of around 30% relative to non-beneficiaries was observed, with especially strong effects in non-manufacturing sectors (services, distribution). Productivity effects were not immediate but emerged more clearly after one to two years.
Sectoral heterogeneity: Gains were concentrated in non-manufacturing industries, suggesting that service-oriented SMEs benefited most from AI-enabled customer and operational solutions.
Sustainability: Employment and sales impacts persisted for at least three years, indicating durability rather than one-off effects.
Limitations: The study notes potential self-selection bias, since early adopters are often more innovation oriented. However, the use of a long firm-level panel (2015-2024 employment; 2001-2024 financials) combined with DID and TWFE specifications mitigates this concern.
Between 2020 and 2023, more than 1 000 SMEs supported by the AI Voucher Program were evaluated using NIPA’s management system data and official statistics, showing consistent improvements across multiple performance indicators. Reported productivity gains averaged 37.6% in 2020, 38.5% in 2021, 40.2% in 2022, and 45.0% in 2023, while quality improvements measured through defect-rate reductions reached 41.4%, 41.9%, 42.3%, and 43.5% over the same period. Unit costs declined steadily by 34.7%, 35.0%, 35.6%, and 36.2%, and on-time delivery improved by 17.2%, 17.8%, 18.0%, and 18.3%. In terms of broader economic impact, by 2023 voucher-supported firms generated KRW 430.3 billion in linked sales, KRW 48.4 billion in linked exports, and KRW 69.7 billion in retained investment, alongside the creation of 864 full-time equivalent jobs. The program also strengthened the ecosystem, with the number of registered AI solution providers expanding from 553 in 2020 to 2 034 in 2023. Follow-up evaluations confirmed broadly similar outcomes across cohorts, demonstrating consistent effectiveness and impact at scale.
The following main lessons can be distilled from the study of this programme.
Lowering entry barriers through cost-sharing flexibility. By covering up to 80% of project costs for SMEs and recognising in-kind contributions, the programme enabled participation by firms with limited financial capacity. This flexible matching mechanism reduced effective cash burdens to as little as 2-5% and proved critical in stimulating adoption among smaller and resource-constrained enterprises.
Evolving programme structure to reflect technology and market change. The introduction and subsequent adjustment of tracks (General, Medical, AI Semiconductor, Micro-business, and Global) demonstrate how delivery models can be adapted to emerging priorities, from healthcare to semiconductor ecosystems and internationalisation. This flexibility allowed the programme to remain relevant in a rapidly evolving AI landscape.
Centralised delivery supported by digital infrastructure. The SMART platform ensured transparent and standardised management of the full project cycle, while complementary measures such as matching services and a call centre helped bridge the gap for firms unfamiliar with AI procurement. This combination of centralisation with practical field support improved accessibility and programme coherence.
Measurable economic and employment impact. Independent evaluations showed sustained increases in employment (around 6 percentage points) and sales growth (about 30%) among beneficiary firms, with the strongest effects in service industries. These results highlight the importance of aligning digital support programmes not only with manufacturing but also with service-sector SMEs that can benefit significantly from AI adoption.
Ecosystem development through supply-side participation. The rapid growth of registered AI solution providers (from 553 in 2020 to over 2 000 in 2023) shows that voucher schemes can simultaneously stimulate demand and nurture domestic AI supply markets, reinforcing broader innovation ecosystems.
Impact differs across sectors, suggesting a dual-track approach. The AIVP experience, when considered alongside the Smart Manufacturing Innovation Initiative (SMII), shows that the simple adoption of pre-developed solutions is effective in service industries, whereas manufacturing often requires customised design and integration. This highlights the value of combining both approaches and indicates that digital adoption policies should leverage high spillover effects in services while accommodating tailored needs in manufacturing.
Short project duration catalyses initial adoption but requires follow-up support. The seven-month project cycle demonstrates that voucher schemes can effectively trigger low-level adoption of AI. However, sustaining these gains requires firms’ clear understanding of their needs, complemented by consultation during matching and aftercare services to support continued use and scaling.
Korea’s AI Voucher Programme shows that a voucher model can lower SME entry barriers while creating a trusted supplier marketplace. Three features are informative for the UK: a vendor-neutral catalogue of pre-validated solutions and implementers; guidance to match firms with appropriate tools; and a small-ticket co-funding that recognises in-kind contributions, keeping cash outlays manageable for micro and cash-constrained firms. A central digital platform for application, matching, contracting and monitoring underpins transparency and consistent delivery.
Evaluation evidence indicates durable gains in employment and sales, with especially strong effects in services, suggesting that early, pre-developed applications can yield quick productivity benefits outside the ICT sector. The programme’s segmentation into tracks also offers a general lesson: differentiating pathways for distinct SME profiles (including micro-business cohorts) could reduce transaction costs and improves fit, while supplier participation scales the domestic market and aftercare support helps sustain use beyond initial adoption.
For the UK, the practical takeaway is to treat introductory AI support as a managed customer journey rather than a generic grant: guidance, a trusted catalogue, milestone-based funding tied to verified use, and clear guardrails.
This work is published under the responsibility of the Secretary-General of the OECD. The opinions expressed and arguments employed herein do not necessarily reflect the official views of the Member countries of the OECD.
This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.
The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.
Photo credits: © Cover © Bardhok Ndoji/iStock by Getty Images Plus.
© OECD 2026
Attribution 4.0 International (CC BY 4.0)
This work is made available under the Creative Commons Attribution 4.0 International licence. By using this work, you accept to be bound by the terms of this licence (https://creativecommons.org/licenses/by/4.0/).
Attribution – you must cite the work.
Translations – you must cite the original work, identify changes to the original and add the following text: In the event of any discrepancy between the original work and the translation, only the text of original work should be considered valid.
Adaptations – you must cite the original work and add the following text: This is an adaptation of an original work by the OECD. The opinions expressed and arguments employed in this adaptation should not be reported as representing the official views of the OECD or of its Member countries.
Third-party material – the licence does not apply to third-party material in the work. If using such material, you are responsible for obtaining permission from the third party and for any claims of infringement.
You must not use the OECD logo, visual identity or cover image without express permission or suggest the OECD endorses your use of the work.
Any dispute arising under this licence shall be settled by arbitration in accordance with the Permanent Court of Arbitration (PCA) Arbitration Rules 2012. The seat of arbitration shall be Paris (France). The number of arbitrators shall be one.
← 1. Ministry of Science and ICT (MSIT) & Information and Communication Industry Promotion Agency (NIPA). (2025). AI Voucher Support Program: Official Announcement. Public Notice No. 2025-0166, February 12, 2025. Available at: https://www.msit.go.kr/bbs/view.do?bbsSeqNo=96&nttSeqNo=3180221&sCode=user (in Korean).
← 2. Korea SME & Startup Support System (BizInfo). (2025).AI Voucher Support Programme: Call for Proposals (AI Semiconductor Track). Ministry of SMEs and Startups & Information and Communication Industry Promotion Agency (NIPA). Available at:https://www.bizinfo.go.kr/web/lay1/bbs/S1T122C128/AS/74/view.do?pblancId=PBLN_000000000104808 (in Korean).
← 3. Gyeonggi Research Institute (GRI). (2023). An Analysis of Employment Effects of Policy Changes: Focusing on the AI Voucher Program. Gyeonggi Research Institute Research Report. Available at: http://nsp.nanet.go.kr/plan/subject/detail.do?nationalPlanControlNo=PLAN0000048827 (in Korean).
← 4. Information and Communication Industry Promotion Agency (NIPA). (2023). AI Voucher Performance Workshop and 2024 (Preliminary) Programme Briefing Materials. NIPA. Available at: https://www.nipa.kr/home/2-1/15112 (in Korean).
← 5. Information and Communication Industry Promotion Agency (NIPA). (2024, December 4). AI Voucher Performance Workshop and Briefing on the 2025 Implementation Plan. NIPA. Available at: https://www.nipa.kr/home/2-1/15744 (in Korean).
