AI adoption is reshaping IT job ads in Romania and Hungary: fewer routine junior-role listings and more hybrid, higher-skill vacancies, suggesting early-stage demand shifts away from entry-level programming work.

Main Finding

AI (especially generative AI) is actively reshaping the IT labour market in Romania and Hungary between 2023–2026. Romania’s market is larger and software-centric with more entry/junior opportunities and remote work; Hungary’s market is smaller, more centralized (Budapest), industry-oriented (automotive, robotics) and biased toward mid‑to‑senior roles. Demand is shifting from traditional IT roles toward AI‑specific profiles (LLM/GenAI engineers, MLOps/LLMOps, prompt engineers, AI evaluators), raising urgent needs for reskilling, updated education programs, and new recruitment strategies.

Key Points

• Scope and trend

  • AI adoption across EU firms (2024–25) is heterogeneous: reported enterprise AI adoption ranges ~5.2%–42.0% by country; Romania ~5.2%, Hungary ~10.4%.
  • McKinsey-style macro estimates: GenAI could automate ≈30% of work hours by 2030 (medium adoption), adding $2.6–4.4 trillion globally — relevant given RO/HU exposure to software and automotive sectors.
  • 75% of knowledge workers report using AI at work (LinkedIn/Microsoft indicators), often BYOAI — governance and security gaps follow.

• Comparative market structure (Feb 2026 snapshots)

  • Romania: ~7,000+ active IT jobs on LinkedIn (8,000+ on local boards); ~430 active IT companies; >4,000 AI/ML/Data jobs; multi‑center geography (Bucharest, Cluj, Ilfov); balanced seniority mix (intern → senior); tech stack: Python, TensorFlow/PyTorch, Docker/K8s, cloud (Azure/AWS/GCP), LLM tools (LangChain, RAG).
  • Hungary: ~4,000+ active IT jobs on LinkedIn (5,000+ local); ~124 IT companies; ~3,000 AI jobs (2,600+ in Budapest); centralized industrial hub (robotics, automotive); predominance of mid/senior positions; tech stack: C++, computer vision, CUDA, Azure AI, robotics/embedded toolchain.
• Roles and skill composition
  • Emerging AI role taxonomy beyond ML/DS: LLM engineer, prompt engineer, RAG/LLMOps, AI product manager, AI evaluator, data curator.
  • Technical intensity: Python, PyTorch/TensorFlow, MLOps/LLMOps, LLM orchestration, RAG, LangChain, cloud, containerization, DevSecOps (GenAI security).
  • Transversal skills: critical thinking, creativity, project and product management, business communication — exposure to AI increases demand for social/business skills.
• Work models and geography
  • Romania offers more remote roles and has a distributed ecosystem; Hungary shows stronger onsite/centered employment tied to industrial R&D.
• Seniority & labour mobility
  • Romania appears more accessible to juniors/entry-level candidates; Hungary seeks more experience, which affects entry pathways and mobility.
• Wage and recruitment implications
  • Authors raise the question of an “AI wage premium” but don’t provide conclusive wage causal evidence; large firms adopt AI faster (55% vs 30% medium vs 17% small enterprises), implying potential within‑country firm‑level wage/skill divides.
• Education and training
  • Evidence of mismatch between current curricula and emerging AI skill needs; authors call for vocational upskilling, higher‑education redesign, and continuous learning programs.

Data & Methods

• Primary data source: manual collection/filtered extraction of LinkedIn job postings (standardized filters for location, industry, job title, work model, required skills).
• Supplementary sources: Jooble.org, Glassdoor, BestJobs.ro, eJobs.ro, TechBehemoths.com, Profession.hu for triangulation.
• Coverage & counts (Feb 2026):
  • Romania: ~7,000+ LinkedIn IT jobs; >4,000 AI/ML/Data roles.
  • Hungary: ~4,000+ LinkedIn IT jobs; ~3,000 AI roles (2,600+ in Budapest).
• Normalization: Romanian raw counts adjusted by RO:HU population ratio (1.98) to compare market intensity rather than absolute size.
• Temporal comparison: two observation points (2023 vs 2026) to detect directional changes; no long time series claimed.
• Role & skill coding: jobs categorized into occupation groups and tagged for technical/transversal skills (examples listed in the chapter).
• Limitations explicitly noted by authors:
  • Reliance on LinkedIn and other job boards (selection bias toward roles advertised there).
  • Only two temporal checkpoints — directional (not causal/time‑series) inference.
  • Manual extraction (API not used) and focus on seven main job types — potential undercoverage of niche roles.
  • Job ads reflect employer signaling and requirements, not realized hires or wages.

Implications for AI Economics

• Labour supply/demand and wage structure
  • Sectoral and geographic specialization (Romania: distributed software; Hungary: centralized industrial AI) suggests divergent local labour demand elasticities and potential regional wage premia for AI skills concentrated in hubs.
  • Faster AI adoption among large firms implies within‑country inequality: large firms can internalize fixed AI costs and pay premiums, widening wage/benefits gaps relative to SMEs.
  • Seniority skew in Hungary may constrain upward mobility for juniors and raise wage pressure for mid/senior AI specialists.
• Productivity and job composition
  • Generative AI augments many cognitive tasks (software engineering, customer ops, marketing, R&D), potentially raising firm productivity in RO/HU sectors exposed to these functions; displacement risk concentrated in routine tasks.
  • Recomposition of job tasks will increase demand for complementary human skills (project mgmt, creativity), shifting human capital investments.
• Human capital policy and education economics
  • Urgent need to realign tertiary curricula and vocational training to include LLM orchestration, MLOps, cloud-native development, and AI governance — otherwise frictions in matching and underemployment for graduates may persist.
  • Public policy should prioritize scalable reskilling (short courses, industry‑university partnerships) and incentives for SMEs to adopt/absorb AI benefits.
• Labour market institutions and regulation
  • BYOAI prevalence raises externalities (security, data governance). Regulators and firms must coordinate on standards, liability, and training to mitigate negative spillovers.
  • Migration and mobility policies: Romania’s openness to junior hires could attract mobility from neighboring markets; Hungary’s demand for experienced talent may raise international recruitment and brain‑import dynamics.
• Research and measurement implications
  • Job‑ads based indicators are useful leading signals of skill demand but need complementing with wage, hiring, and longitudinal employment microdata to assess causality (wage premia, displacement, updating of tasks).
  • Cross‑country heterogeneity in firm size, sectoral mix, and platform use cautions against one‑size‑fits‑all policy prescriptions.
• Macro distributional concerns
  • If AI adoption proceeds faster in certain sectors/large firms and cities, regional and firm‑level inequalities could widen. Policy instruments (training subsidies, regional development, SME support) will matter for equitable gains.

Suggestions for further analysis (concise) - Link job‑ad demand signals to realized hires and wages using administrative payroll or firm HR datasets. - Track skill trajectories with panel data (workers’ upskilling, occupational transitions). - Quantify wage premia and displacement risk by skill group and sector, controlling for firm size and geographic concentration. - Evaluate effectiveness of targeted reskilling programs and university curricula updates in reducing matching frictions.

If you want, I can convert this into a one‑page policy brief highlighting recommended actions for Romanian and Hungarian policymakers and universities.