Lucia Russo
Jeff Mollins
Bénédicte Rispal
OECD Economic Surveys: Estonia 2026
Abstract
The Estonian economy is well placed to benefit from artificial intelligence (AI) enhancing innovation and productivity. AI adoption has accelerated rapidly both among households and businesses and now exceeds the EU average. Current AI use remains concentrated in administrative and support functions, with limited deployment in innovation, production and core operational activities. Unlocking AI’s full potential will require supporting wider enterprise adoption and expanding workforce upskilling.
Estonia is widely regarded as a pioneer in digital governance, having leveraged its relatively small size and flexible public sector in building an advanced system of online public services. The country's early and sustained investment in digitalisation has contributed significantly to its post-transition economic convergence and the emergence of a competitive and outward-looking Information and Communication Technology (ICT) sector that accounts for 8.5% of value added in the economy, one of the largest shares among OECD countries. ICT investment remains strong, with average growth of almost 19% in the past decade (Figure 4.1).
Source: OECD National Accounts database and OECD calculations.
Despite these achievements, the country’s digital strengths have not fully translated into robust productivity growth across sectors. While average productivity growth has been strong over the past two decades, it has decelerated before the pandemic, and more recent developments have been affected by the downturn amid high inflation. At the aggregate level, labour productivity reached around 89% of the EU average in 2021 but fell to 79% in 2025 (Figure 4.2). The scope for further increases in employment rates, which are already above those of other Baltic countries and the OECD average, is limited and so further gains in living standards will need to come mainly from productivity growth.
Source: Eurostat and OECD Productivity database.
In early 2025, the government set up an advisory body, the Economic Growth Council, bringing together public and private sector representatives to advise on policies to foster growth, productivity and innovation. The Council has submitted over 700 suggestions for improving business regulations, notably for streamlining planning procedures, increasing labour market flexibility and easing regulations for foreign workers. The role of the Council this year is to start monitoring implementation of these proposals. Past OECD Economic Surveys have pointed to several areas with a scope for creating more competitive pressures by adjusting the current regulations, such as notaries, lawyers, architects and civil engineers (OECD, 2024c). Another advisory body, the Competitiveness Expert Council, was established in 2023 under the Economic Affairs Committee of the Parliament. Working together with the parliament’s permanent think tank body Foresight Centre, its work has highlighted a strategic potential in developing Estonia’s know-how in processing and manufacturing rare earth magnets, but also prevailing bottlenecks such as a lack of skilled labour, the need for wider digitalisation and reduction of bureaucracy (Foresight Center, 2025a) (Foresight Center, 2025b).
Building on its strong digital foundations and positive attitudes of the public sector towards new technologies, Estonia is well-positioned to leverage Artificial Intelligence (AI) as part of its next phase of digital transformation. Effective diffusion of AI and other advanced technologies across the economy beyond the ICT sector could enable wider productivity gains and help maintain competitiveness. Recent OECD research suggests that potential aggregate productivity gains from AI can be significant, adding 0.4 to 0.9 percentage points of annual growth in labour productivity (Filippucci, Gal and Schief, 2024). In manually intensive activities the gains are likely to be smaller than in knowledge intensive industries. They are also likely to be conditional on widespread adoption, robust digital infrastructure and clear regulations to lower overall uncertainty about the technology.
According to the 2024 OECD.AI Index which uses a composite measure to provide a holistic view of national AI ecosystems Estonia scores roughly on par with the OECD average, but above regional peers such as Latvia and Poland (Figure 4.3) (OECD, 2026). Estonia’s strengths lie in the public policy environment. The country has strong AI research per capita and an ecosystem capable of bringing innovations to the market, both in terms of investment capacity, regulatory agility, and use of AI in the public sector. This chapter describes the current state of play in terms of AI adoption and policy framework before reviewing infrastructure, investment, research potential, as well as digital and AI skills of the workforce.
OECD AI Index (2024)
Note: The index ranges from 0 to 1, where each subcomponent is scaled between 0 and 1 and then averaged.
Source: OECD, (2026[5]), OECD.AI Index.
The share of Estonian enterprises using AI rose sharply to 23.4% in 2025, from around 14% in the year before. This places Estonia above the EU average share of 20% and positions the country among the strong performers in Central and Eastern Europe (Figure 4.4). However, the country trails Nordic economies, where uptake is substantially higher. The recent acceleration reflects a growing perception among firms that AI has become increasingly accessible and necessary for competitiveness, supported by stronger co-operation between research institutions and enterprises aimed at accelerating the commercial deployment of AI-based solutions (Statistics Estonia, 2025).
The diffusion of pre-trained and generative AI models is likely to have played an important role in reducing these barriers. Recent OECD analysis shows that business uptake has accelerated in many countries as firms increasingly rely on ready-to-use AI tools rather than developing AI systems from scratch. This has reduced the need for specialised technical expertise and large upfront investments, making AI technologies more accessible to a broader range of firms, particularly small and medium-sized enterprises (SMEs) (Kergroach and Héritier, 2025).
Adoption remains heterogeneous across firms (Figure 4.4). Large enterprises in Estonia report the highest adoption rates, with around 53% using AI. This remains slightly below the EU-27 average and well below Nordic leaders, where adoption among large firms approaches 75-80%. By contrast, around one-fifth of small firms and one-third of medium-sized firms report using AI. While this exceeds EU averages for both size classes, Estonia has relatively few large firms and many small ones, so this gap is particularly important to overall outcomes and there is substantial scope for further gains in a segment that accounts for a large share of employment.
Note: AI adoption rates across countries in small (10-49 employees), medium (50-249 employees), and large (250 or more employees) enterprises. “Total” refers to enterprises with 10 or more employees.
Source: Eurostat (2025b).
Knowledge-intensive industries lead in AI adoption, although performance relative to the EU varies across activities (Figure 4.45). AI use in the information and communication sector in Estonia (57%) is below the EU average (62%) and Nordic peers, where adoption ranges from around 80% in Denmark and Finland to 88% in Sweden. This relatively lower uptake may reflect the structure of Estonia’s ICT sector, which is characterised by a large number of small, service-oriented providers rather than large platforms or product-based firms typically found in major EU technology hubs. By contrast, professional, scientific and technical activities show strong relative performance, with 52% of enterprises using AI, well above the EU average of around 40%, consistent with the prevalence of consulting, engineering, and IT-enabled business services where AI tools are more readily integrated into existing workflows.
These sectoral patterns reflect differences in production functions and the nature of AI applications across activities. In more traditional sectors, AI adoption patterns are more mixed. Construction - typically a low-AI-intensity sector - shows relatively strong uptake compared with EU peers. This likely reflects the use of AI-enabled tools in administrative, planning, and compliance functions, supported by Estonia’s advanced digital public infrastructure, rather than widespread deployment in core production processes. Accommodation and food services, as well as transport and storage, also record adoption rates above the EU average. In these sectors fewer than one in four enterprises use AI, indicating significant scope for further diffusion. Uptake in utilities, including energy and waste management, remains below the EU average. This may reflect sector-specific constraints, such as legacy systems, regulatory requirements, and high integration costs in capital-intensive network industries.
The functional use of AI in Estonian firms broadly mirrors EU patterns, but with differences in intensity across business functions and firm sizes. AI adoption is most prevalent in marketing and sales and in the organisation of business administration or management (Figure 4.6). Estonia underperforms the EU benchmarks in innovation and production-related activities. AI adoption in R&D and innovation activities among large Estonian firms remains below the EU average, contrasting with many EU countries, where large enterprises lead AI uptake in R&D-intensive functions. This may reflect more limited in-house research capacity, weaker integration of AI into innovation processes, or the sectoral composition of large firms. AI use in core operational and technical functions such as production and logistics is also relatively lower and generally below EU benchmarks, especially among large firms. These functions typically require capital-intensive AI applications and closer integration with physical assets and legacy systems, which can slow diffusion. Overall, AI adoption primarily supports administrative efficiency rather than improvements in production.
Share of enterprises with 10 or more employees (2025)
Source: Eurostat (2025a).
Share of enterprises using at least one technology (2025)
Note: AI adoption rates across countries small (10-49 employees), medium (50-249 employees), and large (250 or more employees) enterprises. “Total” refers to enterprises with 10 or more employees. Black lines indicate EU levels of adoption for the respective firm sizes.
Source: Eurostat (2025b).
AI applications in business support functions are frequently delivered through cloud-based platforms. Around 20% of Estonian enterprises integrate AI with cloud computing services, a figure higher than the EU average (Figure 4.7). This cloud-AI integration is particularly important as cloud infrastructure provides the scalable computational power required for AI deployment more cost-effectively than on-premises solutions, particularly for smaller enterprises. Furthermore, the high share of firms currently combining cloud and data analytics without AI indicates a robust digital foundation that may support future AI diffusion.
Share of enterprises with 10 or more employees (2025)
Note: In Panel A, no advanced technologies considers that share of enterprises that do not use AI, data analytics or cloud services.
Source: Eurostat (2025b).
At the technology level, natural language generation applications are the most widely adopted, with Estonian firms reporting usage rates above the EU average (Figure 4.8). By contrast, adoption levels decline substantially for more capital-intensive or technically complex solutions, such as autonomous robotics or AI-enabled vehicles and drones, which remain limited across all firm sizes and in the manufacturing sector. Large enterprises consistently report higher uptake of AI across all categories, while adoption among SMEs generally remains below 3.5%. This pattern suggests that firms are prioritising AI tools that can be integrated into existing digital processes and that deliver relatively rapid gains in decision-making and workflow optimisation, rather than investing in transformative but costly applications that require substantial complementary investments in physical capital, data infrastructure, and workforce skills.
Share of enterprises (2025)
Note: AI adoption rates across countries small (10-49 employees), medium (50-249 employees), and large (250 or more employees) enterprises. “Total” refers to enterprises with 10 or more employees. Black lines indicate EU levels of adoption for the respective firm sizes.
Source: Eurostat (2025b).
The Estonian authorities have taken a proactive approach to the advancement of AI, guided by the national AI strategy commonly referred to as the Kratt Strategy and broader digitalisation strategies (Box 4.1). The 2022-23 AI strategy acknowledged the need to raise awareness of AI’s value for business, provide support to companies ready to test AI-based solutions, and facilitate their access to resources. To support this objective, the AIRE Centre (AI & Robotics Estonia) was established as Estonia’s European Digital Innovation Hub to foster collaboration between industry, academia and government, and to pilot AI solutions in real industrial environments. Early results indicate improvements in the digital maturity of participating manufacturing firms, though systematic evaluation is still needed. Going forward, AIRE will focus its efforts more on manufacturing, expanding outreach to logistics, transport, energy, quality control and other sectors in which AI deployment remains limited, which is welcome.
The first national AI strategy, commonly referred to as the Kratt Strategy, was adopted in 2019. With a budget of EUR 10 million, it focused on piloting AI solutions within the public sector, strengthening the legal and data governance framework, and supporting R&D. A subsequent strategy - with a budget of EUR 20 million - covering the years 2022–23 expanded these efforts, scaling promising public-sector use cases, supporting R&D and education, and overall uptake in the private sector.
The White Paper on Data and AI 2024-30 sets a long-term vision to establish Estonia as a leading data economy and an “AI-powered state,” ensuring that AI-driven solutions are widely embedded across public administration and private enterprises. Key strategic goals are:
Wide AI adoption across both public and private sector by 2030
Human‑centric, secure, transparent use of AI
Protection of Estonian language and culture through AI development
Creation of sovereign AI capacity
A new governance framework is under development, consisting of a national AI centre, AI registry, AI toolbox, mandatory transparency requirements, testing and regulatory sandboxes. For the public sector, the White paper proposes a binding list of 10 processes that should be automated such as writing meeting memos, talking points, translations, small procurement documentation. The authorities announced a commitment of at least EUR 85 million for the 2024-26 period to accelerate deployment and broaden the economic and societal impact of AI technologies.
Furthermore, the National Digital Decade Strategic Roadmap was adopted in 2025. The roadmap stresses that firms will need to update existing solutions and continuously adopt new ones to remain competitive requiring resources, skills, and the capacity to adapt business models, cooperate internationally, and embed data-driven solutions. The Roadmap calls for strengthening digital and ICT skills among the workforce, promoting continuing training and retraining, and increasing the supply of ICT professionals. It also aims to improve infrastructure, including connectivity, data management, computing capacity (e.g., high-performance and scientific computing), which underpins advanced digital services including AI.
Source: (kratid, 2025; kratid, 2019; kratid, 2022; ERR, 2024; Ministry of Justice and Digital Affairs, 2025; Ministry of Economic Affairs and Communications, Ministry of Justice and Ministry of Education and Research, Government Office, 2024).
Strengthening support for enterprise adoption should remain a policy priority across firm sizes. Greater emphasis on demonstrating successful use cases in traditional sectors such as computer vision for quality control, predictive maintenance in manufacturing, or AI-enabled optimisation in logistics and energy, could help reduce perceived risks and encourage firms to invest in AI applications beyond administrative functions. Facilitating shared access to cloud and computing resources would further lower costs for SMEs, while targeted financial and advisory support to improve data quality, interoperability and governance within firms could help make internal datasets usable for more advanced AI applications. Adoption support should be complemented by skills development and organisational change measures that link specific AI technologies to concrete business functions, for example, training engineers and managers to integrate generative AI into R&D workflows or machine learning tools into production planning and asset management. Embedding such services within existing intermediaries like AIRE would help ensure efficient delivery and avoid fragmentation of support instruments.
Uncertainty over legal implications, fears of breaching data protection and privacy requirements, as well as difficulties in accessing data of sufficient availability or quality have been often cited as barriers to AI adoption (Eurostat, 2025b). Addressing these barriers will therefore be critical to support a shift from basic, efficiency-oriented AI use towards applications with stronger productivity and innovation impacts.
Compliance with emerging regulatory requirements under the EU AI Act will require technical and legal adjustments by firms. The act establishes a risk-based framework classifying AI systems into tiers (minimal, limited, high, and unacceptable risk), each with corresponding regulatory obligations. High-risk AI systems must meet strict safety, transparency, and quality requirements and undergo conformity assessments before deployment. Compliance with these rules will require companies to adapt their processes and infrastructure. In practice, developing a high-risk AI system means implementing controls such as a quality management system, technical documentation, risk assessments, and human oversight mechanisms to ensure the AI’s safety. While the EU AI Act took effect in 2024, most rules start applying in 2025–26.
Providing clear guidance and dedicated advisory services would reduce uncertainty and prevent compliance burdens from becoming a deterrent to adoption. In this regard, practical tools such as model documentation templates, compliance checklists and regulatory sandboxes could support businesses in deploying AI responsibly. In this regard, the “AI Act Awareness and Toolbox” offered by the AIRE as of 1 January 2026 represents a useful step in this direction and could be further leveraged to support experimentation in higher-impact industrial use cases.
4.3.1. Pioneering AI use in the public sector
Estonia's public sector acts as a vital early adopter and testbed for AI systems. More than 130 AI-related projects have been piloted since 2019, and around 60 public authorities have already implemented AI solutions to enhance efficiency and service accessibility (e-Estonia, 2024) (Kratid, 2025a). The most prominent example is Bürokratt, an interoperable ecosystem of chatbots delivering proactive, user-centric public services. In addition to answering questions, it can help fill in forms, with consent retrieve personal data or initiate benefit application. Other public projects include real-time translation and subtitling, automatic transcription in courts and Parliament, AI-enabled fraud prevention, and remote sensing for natural environment monitoring in public agencies to track deforestation and monitor snow, ice and floods.
Earlier this year, the AI Council, an advisory body to the Prime Minister was set up, consisting of entrepreneurs and experts from the Estonian and international start-up sector, technology companies and research institutions. The role of the Council is to illustrate the potential of AI by focusing on a small number of high-impact projects and guide the government policy on AI issues. The government has also invested in the development and open-sourcing of essential language and AI components, such as speech synthesis, machine translation and topic classifiers, enabling both public and private sector actors to build on shared digital infrastructure and reduce costs.
Training programmes to strengthen digital and AI competencies have been introduced across the public sector. The Digital State Academy, which offers free online courses in Estonian and English, covers a wide range of digital topics and its courses are widely used across the public sector to support continuous learning and professional development. The Estonian Parliament has provided guidance and training to use AI tools to its members and staff. Librarians across the country will receive training in AI, enabling them to support the public in developing basic AI skills. The public sector can also use the ‘AI Support Toolbox’, which includes seminars, brainstorming sessions and ‘deep-dives’ to get started as well as advance the use of AI.
4.3.2. Ensuring compute resources and connectivity infrastructure for AI
Digital infrastructure, including broadband access networks, fibre backbone networks (e.g. submarine cables, Internet Exchange Points, data centres) and compute capacity, is critical to supporting the use and development of AI and its resilience. Estonia currently hosts 12 data centres, comparable to the OECD average of 8 centres per million people. However, not all data centres host the compute capacity needed for AI. While there is no domestically available high-performance computing (HPC) infrastructure, such as supercomputers or Graphics Processing Unit (GPU) clusters, the country participates in the EuroHPC Joint Undertaking and the Finnish CSC-coordinated LUMI pre-exascale supercomputing consortium. Access to the LUMI supercomputer is handled by an Estonia-based cloud service company.
As of 2024, no major cloud service provider hosted a public cloud compute region in Estonia (OECD.AI, 2025a). Local AI infrastructure is highly beneficial for securing processing of sensitive data and to ensure resilience in the face of security threats, as well as attracting talent related to AI. While Estonia has good access to cloud services, a recent report by the Estonian Ministry of Justice and Digital Affairs sets out plans to establish an ‘AI gigafactory’ for secure processing, as a local alternative and to enhance resilience, as well as a need to create conditions for data centre investments in Estonia (Ministry of Justice and Digital Affairs, 2025). The report is based on assessment of AI compute supply and demand conducted in line with the OECD Blueprint (OECD, 2023a).
Estonia also joined Finland and Latvia in a cooperation agreement for a Nordic-wide consortium developing digital infrastructure across the region. Streamlining of planning, land-use, and construction permitting processes to ensure predictable and competitive timelines for investors is needed. Public support could also come in the form of providing specific guidance on data centre development for foreign firms, following the Norwegian example. A strategic focus on underlying green energy capacity, such as the one developed in Ireland, could further attract data centre development amidst intense global demand. This would need to be addressed in the wider context of Estonia’s climate policies (Chapter 3).
Data centres per million people
Source: Data Centre Map (2025) and OECD calculations.
Fibre-to-the-home connections help “future-proof” networks. Fibre-optic cables, and more broadly high-quality digital connectivity, are important for enabling the development of models and use and adoption of AI. High-quality fixed broadband, such as fibre connections, are ideal for AI workloads due to their low latency and high bandwidth capabilities. In Estonia, the share of fibre in total fixed broadband subscriptions is above the OECD average (Figure 4.10). Fibre is easier to lay in urban settings, in which over a half of the Estonian population lives. Nevertheless, high-quality high-capacity fixed broadband networks are available to 72% of rural households in Estonia, one of the highest coverage rates in the EU (European Commission, 2025).
Estonia ranks well in terms of mobile broadband penetration, with 177 subscriptions per 100 inhabitants, which puts it third in the OECD, after Japan and the United States. Moreover, mobile traffic, in terms of average monthly data consumption per mobile subscription, a measure of the demand placed on networks due to the rise of AI, reached 37 GB per month at the end of 2024, approximately 2.5-fold higher than the OECD average. The share of “gigabit” offers, i.e. fixed broadband subscriptions with advertised speeds equal to or exceeding 1 Gbps, have increased 4.5 times in five years, from 4% in 2019 to 19% in 2024. In terms of broadband affordability, prices in June 2025 for such high usage (i.e. equal to or exceeding 1 Gbps) were similar to the OECD average (64.3 USD PPP and 63.4 USD PPP respectively) but higher than in neighbouring countries such as Latvia and Lithuania, according to the OECD Communication Price Basket for fixed broadband. In part, this has been attributed to regulatory infrastructure requirements and conditions and requirements for infrastructure permits should be reviewed (Foresight Center, 2025a). Estonia has an investment programme in high-quality broadband networks in underserved areas, with a national goal of covering all households with access to 100 Mbps by 2030 (OECD, 2025a). To keep pace with the infrastructure needs for AI users and developers investments to ensure adequate digital infrastructure will be an important determinant of Estonia’s ability to develop a healthy AI ecosystem.
Share of fibre in total fixed broadband subscriptions
Source: OECD (2025), OECD Broadband Statistics Dataset.
4.3.3. Sustaining investments in AI innovation
Total AI investment per capita, which includes R&D, skills, data and equipment, and other intellectual property products, was higher than the OECD average in 2023 (Figure 4.11) (Fonteneau et al., 2025). While remaining below the peak of 2020, total AI investment improved in 2023 across all categories. Estonia has also one of the strongest per capita VC investments in the OECD. Access to venture capital (VC) remained strong in 2025. Since 2020, roughly a third of VC investment in Estonia is focused on AI, on a par with the OECD average. In the first three quarters of 2025, 17 AI-related deals, with a combined value of USD 156 million took place, compared to the 2022 peak of 25 deals worth USD 314 million.
Note: Venture capital data is for the year 2025, while population data is from 2024.
Source: Fonteneau, F. et al. (2025), (OECD.AI, 2025f).
AI venture capital (VC) investments go to a variety of industries. In recent years they have been largely driven by the digital security sector, focused on two fast-growing companies. VC investment is largely dependent on international sources, that accounted for 96% of the flows in 2025. While Estonia scores well on the OECD FDI Regulatory Restrictiveness Index, measures to facilitate foreign direct investment, such as providing concise investment guidelines, as done by the Estonian Business and Innovation Agency (Box 4.2) could be beneficial for AI ventures.
Estonia’s public support programmes for businesses are focused on three aspects: increasing resilience, technological breakthroughs and strengthening competitiveness, mainly in the manufacturing sector.
Funded largely by EU transfers, the Estonian Business and Innovation Agency (EIS) aims to increase labour productivity to 110% of the EU average and increase private sector research to 2% of GDP in 2035, as well as moving from a “strong innovator” status in the European Innovation Scoreboard to an “innovation leader”.
Numerous support programmes are in place and companies can get an overview through a company portal, which groups all public supports in one place. The programmes emphasise digitalisation, automation and innovation, as well as attracting green-tech related foreign investment. Much of the focus remains on the manufacturing sector, that together with the ICT sector form the backbone of Estonian exports. In recent years, food processing has also become a focus area.
The EIS also administers funding of the various programmes, acts as a public investment agency, and operates energy efficiency programmes for housing. In addition, it runs a visa programme for highly skilled workers and an e-residency programme, which encourages setting up a business in the country without a physical presence. The agency’s budget reached EUR 258 million (0.6% GDP) in 2026, out of which over 60% will be directed towards enterprise support programmes and the rest towards research, development and innovation programmes.
Source: Estonian Business and Innovation Agency.
4.4.1. AI-related skills are fairly widespread but career transitioning is less common
A primary barrier to adopting AI is a lack of relevant expertise, a challenge that is especially pronounced among larger firms that may require more specialist skills to integrate AI in functions such as production processes. This constraint is compounded by a strong competition for talent from the vibrant ICT sector in Estonia. Digital skills and those related to ICT provide the foundation for the development and use of AI (Calvino and Fontanelli, 2023). The Estonian workforce shows a promising basis of foundational skills including both digital and ICT ones. The working-age population scores above the OECD average in literacy, numeracy and adaptive problem-solving (OECD, 2023c) while 15-year-old pupils in Estonia perform above the OECD average in mathematics, reading and science. Furthermore, a large majority (85%) achieve at least level 2 proficiency in mathematics (OECD average of 69%), and the share of top performers in mathematics is also above the OECD average (OECD, 2023b).
In 2025, 62.5% of the Estonian population between 16-74 years old had basic digital skills, two percentage points higher than the EU average, it is therefore well placed to build AI skills (Eurostat, 2026). Given that AI must often be integrated into existing legacy systems and workflows managerial skills are also important for AI adoption, requiring significant organisational change in addition to specialised expertise (Calvino et al., 2022). The authorities should focus on providing upskilling opportunities to the existing workforce, both in terms of digital and AI skills and mainstreaming such programmes into lifelong education.
Demand for ICT and AI skills is expected to continue to grow. According to the 2024 Estonian labour market monitoring study (OSKA), the demand of ICT skills is expected to grow by 16.7% between 2024 and 2027, rising to 40,200 ICT specialists in 2027 (Nõmm, 2025b). Close to 25% of the additional ICT specialists are expected to come from workforce reskilling and 50% from new graduates of ICT-related higher-education programmes. Skills demand will also be affected by automation. Under 10% of the Estonian workforce was at high risk of automation in 2022, where the risk of automation refers to the share of the workforce holding occupations where a quarter of skills used are automatable, somewhat lower than the OECD average (OECD, 2024b). In 2022, around a third of the Estonian workforce was exposed to generative AI, for which at least 20% of the occupations’ tasks could be performed twice as fast with the use of generative AI, above the OECD average of 26%. This share varies across regions and is more accentuated in cities (36.2%) compared to rural areas (24.5%) and towns and suburbs (25%) (OECD, 2024a).
Estonia’s AI talent concentration, as measured by the share of LinkedIn members with AI skills, at 1.17% in 2024 was among the highest of the OECD countries studied, with only Israel (2%) and Luxembourg (1.4%) displaying higher shares (OECD.AI, 2025b). The highest concentration was in the technology, information and media sector (5.36%), followed by the financial services and education sectors (3.96% and 3.62% respectively). A gender gap, although improving and slightly above the OECD average of 30.5%, remains, with women accounting for approximately 32% of Estonia’s AI-skilled workforce. Nevertheless, Estonia remains the country with the highest share of female ICT specialists in the EU (European Commission, 2025).
This growing and high rate of AI skill concentration is reflected in the career transition data. Between 2019 and 2024, 56.6% of employees moving into AI occupations in Estonia did not previously work in AI roles (OECD.AI, 2025c). Expanding lifelong learning programmes and modular professional qualifications could help drive wider diffusion of AI skills into traditional sectors. Such initiatives should include digital literacy but also prioritise function-specific and role-specific training, including the change-management capabilities required to integrate AI into core business processes. Such expansion would benefit, for example, from strong collaboration between the workforce, the government and social partners, and from embedding training within existing intermediaries such as AIRE. Priority should be given to adults with low levels of data-driven skills and to workers in sectors where AI adoption remains limited (OECD, 2025c). Estonia would benefit from strengthening its support to AI talent attraction and retention (Figure 4.12).
Note: Panel B shows the concentration of LinkedIn members with at least two AI engineering skills or who perform an AI occupation per country. Please see the methodological note for more information.
Source: OECD.AI (2025d), data from LinkedIn Economic Graph.
While AI talent is difficult to define and track, LinkedIn data offer insights on both AI talent concentration and international flows. AI skills are self-reported and countries in this dataset are included only if at least 40% of labour force is on LinkedIn. In 2024, the net migration flow, measured by the total arrivals minus departures of LinkedIn members with AI skills, was 0.1 per 10 000 LinkedIn members, demonstrating that more LinkedIn members with AI skills were moving to Estonia compared to those who had left the country (Figure 4.13). This number is, however, below the OECD average of 0.8 and has dropped significantly since the peak of 5 in 2021.
Estonia offers multiple visas and programmes to attract talent and tech companies. Doing business in English, and startup-friendly immigration policy helps. In 2020, Estonia launched the Digital Nomad Visa, which allows remote workers to live in the country for up to a year. Other initiatives such as the e-Residency, available since 2014, provide individuals the possibility to register a company in Estonia while remaining abroad. The Startup Visa, in place since 2017, targets foreign entrepreneurs who wish to establish a technology-based, innovative and scalable company in Estonia. It provides visa for one year, extendable to 5 years, a streamlined application process and can include family members. Since its launch, over 5 000 visas have been issued for company founders and almost 4 500 to employees of these start-ups. In 2023, another programme aimed at companies already present and growing was launched (“Scale-up Visas”), which allows companies meeting criteria to bring foreign workers.
Other EU countries introduced simplified processes to attract foreign tech talent. For example, the French Tech Visa, introduced in 2017, offers a simplified procedure for employees and investors in the technology ecosystem to obtain a renewable 4-year visa.
AI skills migration between countries based on LinkedIn members (2024)
Note: This chart displays the net migration flows of LinkedIn members with AI skills in 2024. Please see the methodological note for more information.
Source: OECD.AI (2025d), data from LinkedIn Economic Graph.
Source: Visit Estonia (2025); Republic of Estonia e-Residency (2025); Startup Estonia (2025); La mission French Tech (2023).
4.4.2. Estonia is one of the first OECD countries to test generative AI at scale in education
In secondary education, public initiatives such as the AI Leap and EdTech Estonia (where EdTech refers to “Education Technology”) have been launched to support the use of AI tools. The AI Leap programme aims to provide students and teachers the appropriate resources needed for AI literacy and to integrate AI tools into school curricula (Box 4.4) (Petrone, 2025). EdTech Estonia focuses on the development and application of technological solutions including tools such as software, hardware and processes intended for educational purposes (EdTech Estonia, 2025).
The AI Leap 2025 programme, launched in September 2025, aims to provide students and teachers at the level of secondary education with the resources needed to integrate AI into education. Developed in co-operation with the private sector it supports the uptake of AI in the education system.
It builds on Estonia’s Tiger Leap programme of the mid-1990s, which focused on providing computers and high- speed internet to schools, teacher training in ICT basics and Estonian-language electronic courseware, which is often associated with the success of Estonia’s digital sector. The AI Leap programme aims at providing students with similar opportunities to build and access skills, tools and the mindset to use AI technologies and support teachers in providing the tools to develop programmes via web-based trainings, seminars and virtual communities.
A pilot project granted access to AI-based educational tools to 20 000 high school students from 11th and 12th grades and 4 700 teachers underwent training to integrate these tools in their curriculum. An Estonian version of ChatGPT Edu is being trained not to give direct answers, but to act more like a teacher, helping students to plan their learning approach, providing guidance to find answers and formulate their own conclusions.
A curriculum has been put into place by working groups of the Ministry of Education and Research hosting stakeholders including academia, teachers, students, and businesses. Both short-term and long-term monitoring are in place. The long-term impact on students is measured by educational psychologists and researchers from the University of Tartu. The programme is planned for deployment in vocational learning schools and younger cohorts, including 10th graders. By 2027, the programme is expected to have impacted close to 58 000 students and 5 000 teachers, about one third of the 159 400 expected number of students in the general education system in 2025 (from 1st to 12th grade).
Source: Foundation TI-Hüpe (2025), Petrone (2025), Ministry of Education and Research (2024)
Targeted initiatives have also been introduced to strengthen digital and AI skills across the workforce. The Education and Youth Board (Harno), a government agency of the Ministry of Education and Research, has launched an education programme to support data-driven management in companies. The initiative provides training to develop data analysis and evidence-based decision-making skills including the use of tools such as SQL and Python (Nõmm, 2025b). The programme is aimed at enhancing data literacy and digital capability and seeks to increase gender diversity in the ICT sector. Further learning opportunities for the workforce include a partnership between AIRE and the academic sector (University of Tartu, Tallinn University of Technology, the Estonian University of Life Sciences and the IMECC Development Centre) to provide learning opportunities for companies and working-age population. These offer various types of courses designed to support in-depth understanding of AI and robotics, their applications and related practical skills. Moreover, private-led initiatives include the //kood, a free of charge study programme, offering courses to gain full-stack developing skills (//kood, 2025).
4.4.3. Students are increasingly interested in ICT and AI education
Recent evidence points to both high expectations of AI-driven change and sustained interest in digital skills among students in Estonia. According to a study commissioned by the Education and Youth Board (Harno), 63% of respondents indicated they believe AI will be widely used and may reduce the need for programmers in the future. Yet, one quarter of Estonians report having considered studying programming or having already done so (Nõmm, 2025b). Such trends are also reflected in formal education, with a growing share of students enrolled in ICT-related fields. The share of students studying ICT-related topics doubled in recent years, increasing from 5.8% in 2016 to 10.2% in 2023 in Masters degrees, well above the OECD average of 4.36% (Figure 4.14). Nevertheless, informal learning remains a key pathway for acquiring AI related skills. Close to 82% of the Estonia-based respondents of the 2024 Stack Overflow Annual Developer Survey reported learning coding skills via online resources and 49% through online courses and certifications (OECD.AI, 2025e).
Note: The term STEM refers to the aggregation of the broad fields of natural sciences, mathematics and statistics; information and communication technologies; and engineering, manufacturing and construction.
Source: OECD (2025d).
4.4.4. Capitalising on a burgeoning AI research community
A rising AI research hub in a small economy
Estonia’s research output on AI-related topics has been expanding in recent years. R&D lies at the heart of efforts to advance AI as a general-purpose technology, alongside ongoing innovation in its applications and integration. As countries across the OECD strive to position themselves at the forefront of AI development, the ability to foster robust research ecosystems and innovation networks has emerged as a key determinant of technological competitiveness. While AI research output and investment is generally dominated by large economies, smaller and emerging economies can play a role in the global AI landscape. In 2015, Estonians were credited a total of 18 AI-related research publications, and by 2024, that number has expanded to 292 (Figure 4.15). Research output, as a share of total population, is currently above the OECD average with 115 publications per million people (Figure 4.15). Estonia’s research is having an impact, with about 1 277 citations per million people, just above the OECD average.
Fractional publications per million people (2024)
Note: Publications are expressed here as fractional counts, where equal weights are assigned to each publication’s co-author to avoid double-counting of publications.
Source: OECD.AI (2025), data from Elsevier.
AI research focuses primarily on adapting existing models to local contexts
Research in Estonia on AI appears to be largely focused on computer science applications and electronic engineering. Nevertheless, language and linguistics remain an under-researched area. Unlocking research on AI is conditional on adequate supply of training data. As of 2025, Estonian was the 35th most common language appearing in training data, a relatively high position given the small population of 1.3 million. Public sector efforts have directly contributed to this good training data availability. An open public data portal has improved access, while the establishment of ‘high-value’ datasets supports the development of AI. According to the OUR Index, an OECD benchmark that assesses how effectively governments design and implement open government data policies, Estonia scores well on the availability and accessibility of the ‘high-value’ dataset. A public campaign “Donate a speech” in 2022 enhanced language resources by crowdsourcing spoken Estonian. Citizens can contribute voice recordings across different accents and dialects, which are anonymised and made openly available for training speech-recognition and other language technologies (OECD, 2025c; e-Estonia, 2022).
Despite the good training data availability, there are currently no large language models (LLMs) credited to Estonian developers (Epoch AI, 2025; OECD.AI, 2025a). A large amount of textual data has been made available to Meta to train their models on the Estonian language and according to Hugging Face - a website that maintains open-source tools and models - over 1 300 models are trained on the Estonian language, which corresponds to about 0.5% of all models available with language tags. Only 99 of these models are specifically tailored for the Estonian language, and are retrained from other languages, signalling that Estonian researchers remain focused on applying existing models and infrastructure to local use cases, rather than building new LLMs. While strategic partnerships allow for effective AI model applications, they do not contribute directly to AI development in the domestic language context. However, it remains to be seen how necessary it is for small economies to develop AI models in their own languages, especially for Estonia which has a high percentage of English and other language speakers.
Collaboration between private and academic researchers is limited
The focus on adaptation rather than development is mirrored in patenting activity, with only a single AI-related patent application during 2022-23 (OECD, 2025a). This lack of patenting reflects, in part, limited collaboration between Estonia’s comparatively strong academic AI research base and its public and private sectors. To address this gap, Estonia could strengthen institutional linkages between academia, government, and industry to foster a more development-oriented AI ecosystem. Public digital innovation hubs such as AI & Robotics Estonia (AIRE) should be promoted and expanded, when possible, to support applied research, and facilitate experimentation. To attract international expertise AIRE can also capitalise on existing international academic networks, given that 73% of all AI-related publications in Estonia involve international collaboration.
Strengthening Estonia’s research and development in AI for language and multilingual applications could generate both productivity gains in the public sector and exportable capabilities in small-language natural language processing (NLP). Research incentives, grants, and upskilling focused on the use of local language datasets and new model development and applications may help to foster a domestic AI ecosystem. The Netherlands’ GPT-NL, an open language model developed through a partnership between the government and non-profit organisations, offers a useful prototype of funding and governance arrangements for such efforts (TNO, 2023).
References
//kood (2025), //kood, https://kood.tech/en/ (accessed on 27 November 2025).
Calvino, F. et al. (2022), “Closing the Italian digital gap: The role of skills, intangibles and policies”, OECD Science, Technology and Industry Policy Papers, No. 126, OECD Publishing, Paris, https://doi.org/10.1787/e33c281e-en.
Calvino, F. and L. Fontanelli (2023), “A portrait of AI adopters across countries: Firm characteristics, assets’ complementarities and productivity”, OECD Science, Technology and Industry Working Papers, No. 2023/02, OECD Publishing, Paris, https://doi.org/10.1787/0fb79bb9-en.
European Commission (2025), Digital Decade 2025: Broadband coverage in Europe 2024.
Data Center Map (2025), Data centers, https://www.datacentermap.com/datacenters/.
EdTech Estonia (2025), EdTech Estonia, https://www.edtechestonia.org/home (accessed on 1 December 2025).
e-Estonia (2024), AI strategy - factsheet, https://e-estonia.com/wp-content/uploads/factsheet-ai-strategy.pdf (accessed on 15 November 2025).
e-Estonia (2022), Estonian launched donate your speech campaign to make digital government services more inclusive and intelligent, https://e-estonia.com/estonians-launched-donate-your-speech-campaign-to-make-digital-government-services-more-inclusive-and-intelligent/.
Epoch AI (2025), Data on AI models, https://epoch.ai/data/ai-models.
ERR (2024), Estonia to invest €85M in boosting AI uptake across public, private sectors, https://news.err.ee/1609250613/estonia-to-invest-85m-in-boosting-ai-uptake-across-public-private-sectors (accessed on 14 November 2025).
European Commission (2025), Estonia 2025 Digital Decade Country Report, https://digital-strategy.ec.europa.eu/en/factpages/estonia-2025-digital-decade-country-report.
European Commission (2024), Estonia 2024 Digital Decade Country Report, https://digital-strategy.ec.europa.eu/en/factpages/estonia-2024-digital-decade-country-report (accessed on 26 November 2025).
EUROSTAT (2025), Artificial intelligence by size class of enterprise (dataset).
Eurostat (2026), “Skills for the digital age”, Statistics Explained, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Skills_for_the_digital_age (accessed on 1 May 2026).
Eurostat (2025a), Artificial intelligence by NACE Rev. 2 activity, https://ec.europa.eu/eurostat/databrowser/view/isoc_eb_ain2__custom_19134595/default/table (accessed on 19 December 2025).
Eurostat (2025b), Statistics Explained: Use of Artificial Intelligence in Enterprises, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Use_of_artificial_intelligence_in_enterprises (accessed on 18 December 2025).
Filippucci, F., P. Gal and M. Schief (2024), Miracle or Myth? Assessing the macroeconomic productivity gains from Artificial Intelligence, OECD Publishing, https://www.oecd.org/content/dam/oecd/en/publications/reports/2024/11/miracle-or-myth-assessing-the-macroeconomic-productivity-gains-from-artificial-intelligence_fde2a597/b524a072-en.pdf.
Fonteneau, F. et al. (2025), “Advancing the measurement of investments in artificial intelligence”, OECD Artificial Intelligence Papers, No. 47, OECD Publishing, Paris, https://doi.org/10.1787/13e0da2f-en.
Foresight Center (2025a), Mobile internet is cheap in Estonia, but prices of fixed line connections are high compared to neighbouring countries, https://arenguseire.ee/en/news/report-mobile-internet-is-cheap-in-estonia-but-prices-of-fixed-line-connections-are-high-compared-to-neighbouring-countries/ (accessed on 14 February 2026).
Foresight Center (2025b), The Situation and outlook for the Estonian economy in 2025 – Report of the Competitiveness Expert Panel to the Riigikogu.
Foresight Center (2025c), The situation and Perspectives of the Estonian Economy.
Foundation TI-Hüpe (2025), AI Leap, https://en.tihupe.ee/ (accessed on 1 December 2025).
Kergroach, S. and J. Héritier (2025), “Emerging divides in the transition to artificial intelligence”, OECD Regional Development Papers, No. 147, OECD Publishing, Paris, https://doi.org/10.1787/7376c776-en.
Kratid (2025a), AI use cases, https://www.kratid.ee/ai-use-cases.
kratid (2025b), What is AI or kratt as we call it in Estonian?, https://www.kratid.ee/en/tehisintellekt (accessed on 12 November 2025).
kratid (2022), Estonia’s National Artificial Intelligence Strategy or Kratt Strategy for 2022–2023, https://e964029b-7c32-42a7-817c-37c391a3b976.filesusr.com/ugd/980182_4434a890f1e64c66b1190b0bd2665dc2.pdf.
kratid (2019), Estonia’s national artificial intelligence strategy 2019-2021, Government of the Republic of Estonia, https://e964029b-7c32-42a7-817c-37c391a3b976.filesusr.com/ugd/980182_8d0df96fd41145739dff2595e0ab3e8d.pdf.
La mission French Tech (2023), French Tech Visa, https://lafrenchtech.gouv.fr/en/come-work-in-france/french-tech-visa/ (accessed on 19 December 2025).
Ministry of Economic Affairs and Communications, Ministry of Justice and Ministry of Education and Research, Government Office (2024), Andmete ja tehisintellekti valge raamat 2024-2030 [Data and Artificial Intelligence White Paper 2024-2030], https://www.mkm.ee/sites/default/files/documents/2024-02/Tehisintellekti%20ja%20andmete%20valge%20raamat%202024-2030.pdf.
Ministry of Education and Research (2024), Statistics and analysis, https://www.hm.ee/en/ministry/statistics-and-analysis#general-education (accessed on 1 December 2025).
Ministry of Justice and Digital Affairs (2025), Europe’s Digital Decade Strategic Roadmap: Estonia, https://www.justdigi.ee/sites/default/files/documents/2025-03/Estonian%20National%20Digital%20Decade%20Strategic%20Roadmap%202025.pdf (accessed on 10 November 2025).
Nõmm, T. (2025a), Education and Youth Board supports the development of data-driven management in Estonian companies.
Nõmm, T. (2025b), Free retraining opportunity for newcomers to the IT sector, https://harno.ee/en/news/free-retraining-opportunity-newcomers-it-sector (accessed on 25 November 2025).
Nõmm, T. (2025c), Free retraining opportunity for newcomers to the IT sector, https://harno.ee/en/news/free-retraining-opportunity-newcomers-it-sector (accessed on 27 November 2025).
OECD (2026), The OECD.AI Index, OECD Publishing, https://doi.org/10.1787/32c01014-en.
OECD (2025a), Closing Broadband Connectivity Divides for All: From Evidence to Practice, https://doi.org/10.1787/d5ea99b2-en.
OECD (2025b), Distribution of enrolled students, new entrants and graduates by field of education, https://data-explorer.oecd.org/s/2p0 (accessed on 1 August 2025).
OECD (2025c), Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions, OECD Publishing, Paris, https://doi.org/10.1787/795de142-en.
OECD (2025d), Identifying emerging AI technologies using patent data: A semi-automated approach, https://doi.org/10.1787/d17e9a1a-en.
OECD (2025e), OECD Skills Outlook 2025: Building the Skills of the 21st Century for All, OECD Publishing, Paris, https://doi.org/10.1787/26163cd3-en.
OECD (2024a), Job Creation and Local Economic Development 2024 - Country Notes: Estonia, https://www.oecd.org/en/publications/job-creation-and-local-economic-development-2024-country-notes_ad2806c1-en/estonia_726d830d-en.html (accessed on 26 November 2025).
OECD (2024b), Miracle or Myth? Assessing the macroeconomic productivity gains from artificial intelligence, OECD Publishing, https://www.oecd.org/content/dam/oecd/en/publications/reports/2024/11/miracle-or-myth-assessing-the-macroeconomic-productivity-gains-from-artificial-intelligence_fde2a597/b524a072-en.pdf.
OECD (2024c), OECD Economic Surveys: Estonia 2024, OECD Publishing, Paris, https://doi.org/10.1787/33e6beee-en.
OECD (2024d), OECD Economic Surveys: Estonia 2024, https://doi.org/10.1787/33e6beee-en.
OECD (2023a), “A blueprint for building national compute capacity for artificial intelligence”, OECD Digital Economy Papers, No. 350, OECD Publishing, Paris, https://doi.org/10.1787/876367e3-en.
OECD (2023b), PISA 2022 Results, Estonia, OECD, https://www.oecd.org/content/dam/oecd/en/publications/reports/2023/11/pisa-2022-results-volume-i-and-ii-country-notes_2fca04b9/estonia_2d5cc238/dafed886-en.pdf.
OECD (2023c), Survey of Adults Skills 2023: Estonia, OECD Publishing, https://www.oecd.org/en/publications/survey-of-adults-skills-2023-country-notes_ab4f6b8c-en/estonia_60a7b919-en.html#:~:text=In%20Estonia%2C%20adults%20aged%2016,average)%20(Figure%201). (accessed on 7 January 2026).
OECD.AI (2025a), AI Compute and the Environment; Cloud compute regions, https://oecd.ai/en/site/compute-climate/data.
OECD.AI (2025b), Live data, AI jobs and skills, https://oecd.ai/en/data?selectedArea=ai-jobs-and-skills&selectedVisualization=ai-talent-concentration-by-country (accessed on 24 November 2025).
OECD.AI (2025c), Live data, AI jobs and skills, https://oecd.ai/en/data?selectedArea=ai-jobs-and-skills&selectedVisualization=career-transitions-into-ai-occupations-all-countries-2023 (accessed on 24 November 2025).
OECD.AI (2025d), Live data, AI jobs and skills, https://oecd.ai/en/data?selectedArea=ai-jobs-and-skills&selectedVisualization=between-country-ai-skills-migration (accessed on 25 November 2025).
OECD.AI (2025e), Live data, AIKoD database, https://oecd.ai/en/data?selectedArea=ai-models-and-datasets.
OECD.AI (2025f), Live data, Preqin, https://oecd.ai/en/data?selectedArea=investments-in-ai-and-data.
OECD.AI (2025g), Stack overflow survey, https://oecd.ai/en/stackoverflow-survey-data (accessed on 28 August 2025).
Petrone, J. (2025), AI Leap 2025: Estonia sets the global standard for AI in education, https://e-estonia.com/ai-leap-2025-estonia-sets-ai-standard-in-education/ (accessed on 28 November 2025).
Republic of Estonia e-Residency (2025), become an e-resident, https://www.e-resident.gov.ee/ (accessed on 19 December 2025).
Startup Estonia (2025), Startup visa, https://startupestonia.ee/startup-visa/ (accessed on 19 December 2025).
Statistics Estonia (2025), Statistics Estonia’s ICT survey: half of Estonian enterprises use data analytics, Statistics Estonia's ICT survey: half of Estonian enterprises use data analytics | Statistikaamet.
The Estonian Ministry of Justice and Digital Affairs (2025), TULEVIKUKINDLA ARVUTUSVÕIMEKUSE TAGAMINE, https://www.justdigi.ee/sites/default/files/documents/2025-12/Tulevikukindla%20arvutusvo%CC%83imekuse%20tagamine%20v1.1_0.pdf.
TNO (2023), The Netherlands starts realisation GPT-NL, its own open AI-language model, https://www.tno.nl/en/newsroom/2023/11/netherlands-starts-realisation-gpt-nl/ (accessed on 22 December 2025).
Visit Estonia (2025), Digital nomad visa, https://visitestonia.com/en/what-to-do/digital-nomad-visa (accessed on 19 December 2025).