Debates about the economics of AI often collapse into a technology race: who has the better foundational model, more GPUs or larger data centers. For a country the size of Hungary, the more relevant question is likely different: how much of the economic value created by AI will originate in and remain in Hungary.
The central issues are whether Hungarian companies will build their own technology, products and intellectual property; whether workers now doing transactional or execution-focused tasks can be moved into higher-value roles; whether businesses can adopt AI without compliance and governance costs eroding productivity gains; and whether Hungary will be part of the international research, standardisation and knowledge networks shaping the next generation of AI.
What the numbers and current situation indicate
Over recent decades Hungary’s growth model relied heavily on foreign direct investment, manufacturing and business service centres. According to the National Investment Agency (HIPA) 2025 survey, there were 245 business service centers operating in Hungary with 118,543 employees; the sector has been bringing increasingly complex, higher-value activities to the country.
AI changes where human labour is needed: many standardisable, well-documented and execution-intensive tasks can increasingly be automated. That does not necessarily mean an immediate disappearance of jobs on a massive scale, but it does imply that producing the same economic output in a few years will require different competencies, a different division of labour and often different staffing levels.
Domestic ownership and venture capital
One of AI’s opportunities is to reduce the human capacity needed to scale certain corporate functions. A small team can today build a product or service that previously required a much larger organisation, meaning a tech company with a few dozen employees and one or a few founders can generate significant revenue, profit and enterprise value quickly. That, however, requires capital: startups, especially deep-tech firms, often need prolonged financing before achieving stable positive cash flow, and venture capital plays a key role.
From a competitiveness perspective, policy debates such as the public consultation on the draft wealth tax matter. The issue is not only whether high wealth should bear higher tax burdens; for a fast-growing startup the company’s paper value and the founder’s realised liquidity can be very different. If an annual personal tax tied to company valuations creates a liquidity need before real income exists, it can deter founders from keeping capital in the company. Entrepreneurial associations have already raised the special treatment of operating entrepreneurial wealth in this context.
The competition for skills and reskilling
Capital alone is not enough. A critical question is what happens to workers whose current jobs are significantly transformed. The World Economic Forum’s 2025 Future of Jobs survey found 63% of surveyed employers listed the skills gap among the biggest barriers to corporate transformation, and respondents estimate that by 2030 some 59 out of every 100 workers will need upskilling or reskilling. Rapidly growing skills include AI and big data, general digital literacy, cybersecurity, analytical thinking, adaptability and creative thinking.
For Hungary this has high stakes: with nearly 120,000 people employed in business service centres, labour market strategy cannot assume today’s job titles will remain unchanged. The question is how quickly people can be shifted from execution-focused tasks to judgment- and decision-focused roles — from transaction processing to exception handling and decision support, from report production to analysis, from routine knowledge work to AI-augmented, higher-value roles. This requires more than generic AI training or prompt-engineering courses: strategic workforce planning is needed to identify which tasks can be automated, which will evolve, where new demand arises, what competencies will be required, and which current workers can realistically transition. Retraining must start before roles disappear; waiting until the headcount effects are visible often wastes years.
Proportionate governance and compliance
Governance is often presented as a binary choice — be innovation-friendly or regulate strictly — but that’s misleading. AI poses real risks: not only data protection and discrimination concerns, but also business, operational, reputational and agentic control and authorization risks. Managing these is not optional; the question is how to do so proportionately.
AI risk does not scale linearly with company size. A 50-person firm might operate a recruiting system that processes personal data or any other automated workflow accessing sensitive data. The risk is not automatically a small fraction of that faced by a large company, but smaller firms generally lack the 20-person legal, data protection, cybersecurity and AI-governance teams big companies have. If the costs of governance, documentation and compliance exceed the productivity gains from an AI solution, adoption is economically unjustifiable.
It is therefore welcome that Hungarian experts can increasingly participate in the standardisation processes that will shape implementation of the European Union’s AI regulation (AI Act). The author participates in the MSZT CEN-CENELEC JTC 21 follow-up Hungarian mirror committee. The European Commission has asked the joint CEN and CENELEC JTC 21 committee for harmonised standards in ten areas, including risk management; data governance and quality; logging; transparency; human oversight; accuracy; robustness; cybersecurity; quality management; and conformity assessment. If a harmonised standard is referenced in the Official Journal of the European Union, its application can provide a presumption of conformity with covered AI Act requirements.
The EU regulatory framework increasingly recognises the needs of small and medium-sized enterprises, startups and organisations deploying AI: member states are expected to provide targeted training, tailored guidance and contact points, access to regulatory testing environments, and support for participation in standardisation. The EU’s 2026 simplification steps further strengthened this direction by aiming at more proportionate compliance obligations and broader access to regulatory testbeds for smaller firms.
Good AI governance is therefore proportionate to actual risk and must ensure compliance costs do not consume the economic benefit of AI. For Hungary this will require not only well-designed rules but practical implementation support, common methodologies, regulatory test environments and accessible verification and conformity-assessment infrastructure for smaller businesses.
Don’t reinvent everything domestically, but integrate internationally
It is unrealistic for a small country to aim to build critical mass independently in every advanced AI research field. What matters more is that Hungarian universities, researchers, firms and civil society participate in the international networks where next-decade technological and governance solutions are developed.
Hungary is not starting from zero: for example, through HunAIFA Hungary joined the European AI Factory Antenna network, which aims to broaden access to AI infrastructure, compute capacity and innovation ecosystems for startups and SMEs. But participation alone is insufficient: the question is whether these involvements translate into lasting knowledge transfer, collaborative research, researcher and talent mobility, new ventures, corporate partnerships and a sustained informal professional network.
A relevant benchmark is ELLIS (European Laboratory for Learning and Intelligent Systems), which now connects 47 European sites across 19 countries with more than 2,300 members, 16 international research programmes, and over 500 PhD and postdoctoral researchers. The public site list does not currently include a Hungarian ELLIS site. That does not mean Hungarian researchers are absent from European AI research, but it shows room for deeper institutional integration.
Civil society and social science participation in international research and standardisation networks is equally important, because AI’s social impacts — employment, discrimination, privacy, accessibility, labour rights and human oversight — are not purely technical matters.
AI must reach SMEs and mid-sized firms
If AI adoption remains concentrated among a few multinationals, technology startups and university labs, its benefits will not diffuse broadly. Eurostat reports that in 2025 some 20% of EU enterprises with at least ten employees used at least one AI technology; in Hungary the share was about 10.5%. By comparison, Denmark was at 42%, Finland 37.8% and Sweden 35%. Company size matters too: across the EU 55% of large firms used AI in 2025, while only 18.9% of SMEs did.
A large portion of AI’s economic impact will come from everyday firms: a 500-person Hungarian manufacturer planning production more quickly, a 50-person service firm automating administration, a logistics company producing more accurate forecasts, or a mid-sized firm using AI to support its salesforce and decision-makers. If these tools are mostly accessible to large firms, AI may widen productivity gaps within the Hungarian economy.
Therefore support for SME AI adoption should not be mere technological promotionism. The goal is to deploy AI where it creates measurable economic value, ensuring access to knowledge, infrastructure and capital, and offering proportionate governance and compliance solutions that do not make AI adoption uneconomic for smaller firms.
Conclusion: a strategy for economic transition, not only an AI strategy
The key lesson is that AI competitiveness is easy to treat as a technology question — more datacentres, more AI specialists, more AI projects. Those elements matter, but alone they are insufficient. For Hungary, AI is fundamentally an economic adaptation challenge: it reduces demand for some tasks that were previously located here, raises the value of higher-level knowledge and decision-making work, enables smaller teams to build internationally scalable firms, and increases the importance of moving capital, knowledge and labour rapidly toward new opportunities.
If policy and market actors align incentives, training and regulation well, AI can help Hungary move up the value chain: more domestically owned, exportable technology firms; higher-value business services; more productive SMEs; and a workforce that competes on knowledge rather than cost. Achieving that requires consistent incentives, careful handling of tax and liquidity effects, genuine pathways for reskilling, proportionate governance that keeps compliance costs reasonable for smaller companies, and deep engagement in international research and standardisation networks. Hungary does not need to beat the United States or China at foundational model development to succeed; it needs to adapt quickly to the reallocation of value between capital, labour and knowledge — and in that competition it can indeed win.
This article appears in Portfolio’s "on the other hand" opinion section. The author’s views do not necessarily reflect the editorial stance of Portfolio.
