Three years after the “big bang” of generative AI, the tech job market is starting to leave a quantifiable mark. In Spain, job offers for programmers have fallen 31% since the release of ChatGPT (November 2022). The figure is striking, but it doesn’t tell the whole story: the number of employed developers continues to hit record highs and companies are still hiring… although less, more slowly, and with a different profile. The trend emerging in Europe, and more intensely in the United States and parts of Asia, is not one of the “extinction of the programmer,” but rather a change of rules that rewards productivity with AI, system design, data, and platform engineering over the pick and shovel coding of a decade ago.
What has happened in Spain (and why it’s not a paradox)
Eurostat tracks two distinct things: the weight of ICT profiles in posted job offers and how many professionals are employed. Since 2022:
- The ICT sector drops from 13.5% of offers in Q3-2022 to 10.5% in Q1-2025.
- Programmers: from 7.9% to 5.4% (−31%).
- Other ICT profiles (networks, sysadmin, DBA, security, ops…): from 5.6% to 5.1% (−18.7%).
- Developer employment: 554,000 in the latest semi-annual data, +63.5% compared to 2019 and +18.2% since ChatGPT.
How can there be fewer offers and more employment at the same time? Because turnover has decreased. Fewer employees are leaving, and fewer new vacancies are being created; people stay in their jobs (the “Great Tech Retention”). The statistical result is more employed with fewer ads and a less liquid market, where changing jobs is more difficult and salaries are held back outside of big tech.
Data, in tables
1) Spain: demand evolution (ICT offers as a percentage of total posted)
| Indicator | Q3-2022 | Q1-2025 | Change |
|---|---|---|---|
| ICT total | 13.5% | 10.5% | −3.0 pp |
| Programmers | 7.9% | 5.4% | −31% |
| Other ICT profiles | 5.6% | 5.1% | −18.7% |
Source: Eurostat.
2) ICT employment and “exposure” to AI (programmers within ICT employment)
| Country / Area | ICT Employment (% of total) | Programmers (% of total) | “Exposure” (programmers / ICT) |
|---|---|---|---|
| Spain | 4.7% | 2.7% | 52% |
| EU (average) | — | — | 51% |
| Germany | — | — | 48% |
| Ireland | — | — | 76% |
| Bulgaria | — | — | 65% | Poland | — | — | 65% |
How to interpret “exposure”: the higher the share of programmers within ICT employment, the greater the immediate sensitivity to AI-induced changes in coding tasks. Spain (52%) is slightly above the European average (51%) and well below Ireland (76%), where product development carries significant weight.
Source: Eurostat (ICT employment by groups).
3) International thermometer: Europe vs. the US (and signals from Asia)
| Indicator | Europe (EU references) | United States | Asia (qualitative signals) |
|---|---|---|---|
| Offers for programmers | −31% in Spain since 2022 | Layoffs tech 126,000 (2024) | Cooling of traditional outsourcing; shift towards applied AI |
| Programmer employment | Record employed (Spain) | BLS projects −11% 2022-2032 (−147,000 positions) | Growing demand in AI/ML, MLOps, and semiconductors |
| Fired developers | 7.5% report layoffs (OfferZen, EU) | 2023-2024 waves with partial reallocation | Selective reorganizations for efficiency |
Sources: Eurostat; US Bureau of Labor Statistics (BLS); sectoral counts of tech layoffs in 2024; OfferZen (EU).
What is really changing (beyond the noise)
1) AI as a multiplier of boilerplate
Code assistants and generative tools accelerate templates, testing, and documentation. This reduces the need for junior profiles for repetitive tasks and raises the bar for entry: solid CS, systems, security, and data matter more than mastering a framework.
2) From “feature teams” to “platform teams”
The focus shifts towards platform engineering (internal platforms, observability, security, FinOps, MLOps). Fewer scattered repos; more internal products to multiplicative effect for each developer.
3) Measured (and audited) productivity
Companies demand that AI deliver shorter timelines, fewer defects, and lower costs. Those who know how to evaluate and orchestrate AI (prompts, guardrails, evaluations, security) increase their value.
4) Less turnover, more stability
Rising costs of job switching and decreasing bulk offers. This sustains employment numbers but quenches widespread salary increases.
Does this mean “programming is over”? No: it’s about what is programmed and how
In the short term, AI does not eliminate development work; it reconfigures it:
- System design and architecture focused on security and resilience.
- Data and applied AI: data engineering, MLOps, feature stores, model observability.
- Platforms: cloud-native, Kubernetes, networks, service mesh, policy as code.
- Software quality and evaluation (and AI-assisted code): testing, fuzzing, supply chain security, SAST/DAST, licenses.
Developers who embrace these layers increase their value; those limited to assembling boilerplate without understanding systems see how the tool reaches them.
The US case: a mirror of what might come
The BLS projects a −11% decrease in employment for “programmers” by 2032 (−147,000 jobs). It does not cover the entire software sector (which grows in other areas), but the classic programmer segment. The clear political message: if AI automates part of basic programming, the market absorbs fewer new entrants and reassigns roles towards higher value.
Europe, with a smaller and less volatile tech industry, does not show this employment decline, but fewer vacancies and more permanence. If demand does not rebound, the bargaining power will gradually diminish — as already happening in parts of the US market.
And Asia? From outsourcing to industrial AI
Without a harmonized panel like Eurostat/BLS, the dominant signals in 2024-2025 have been:
- Less growth in traditional coding outsourcing.
- More investment in applied AI and semiconductors (Taiwan/Korea).
- Search for hybrid profiles (software + data + MLOps), with higher qualification requirements.
The common conclusion: less space for “commodity coding,” more for system and data engineers who know how to operate with AI.
What this means for Spain (professionals and companies)
For professionals
- Learn to “program with AI”: it’s not “copy and paste,” but design, evaluate, test, and secure. Know the limits (licenses, privacy, prompt injection).
- Elevate to the second layer: cloud, Kubernetes/Docker, Proxmox, observability, security applied, data, and MLOps.
- Impactful Portfolio: measurable projects (savings, revenue, reliability), not just demos.
For companies
- AI Governance: guardrails, evaluations, security, data, and privacy. Real productivity above hype.
- Re-skilling: training in platforms, automation, and data; reassigning ICs to product ops and quality.
- Retention Offerings: flexibility, learning pathways, projects with purpose. In a cold market, culture makes a difference.
Is this the beginning of a disaster for programmers? No… if we play our cards right
Spain maintains a structural margin: the weight of programmer vacancies (5.4%) remains well above its share in employment (2.7%). That “disparity”—the seventh-highest in the EU—indicates room to grow and demand better conditions… if demand stops shrinking and professionals strengthen their competencies where value currently lies.
The other risk: that employment rises only due to less turnover, while demand continues to decline. Then salaries would tend to stagnate, and the sector would lose momentum. Precisely what should be avoided through productivity, training, and impactful projects.
Methodology and limits
- The figures for offers and employment come from Eurostat (ICT groups and “software and applications developers” subgroup).
- Reference for US: combines BLS forecast (2022-2032 −11% for “programmers”) and sectoral counts of 126,000 tech layoffs in 2024.
- For Asia: no harmonized series, but market indications (less commodity outsourcing, more AI/semiconductors).
- The country percentage figures in the “exposure” table are based on Eurostat comparative panels.
Frequently Asked Questions
1) If programmer offers decrease, should I leave software?
No. The work doesn’t disappear; it changes. Those who combine coding with systems, data, security, MLOps, and AI-driven productivity gain importance.
2) Will AI replace the programmer?
It replaces repetitive tasks, not the full function. The value shifts toward system design, quality assurance, orchestration of tools, and impact measurement.
3) Why does Spain keep creating tech jobs despite fewer vacancies?
Because less firing and less turnover sustain employment and reduce advertising volume. The challenge is to ensure that stability doesn’t lead to flat wages and less innovation.
4) What skills will be most in demand in 2025-2027?
Platform engineering, observability, security (shift-left), data/AI, automation (IaC, CI/CD), and AI governance (evaluations, guardrails, privacy).
In summary
AI has not “killed” programming work; it has raised it to a higher level. Spain sees a slowdown in offers but retains a margin if companies and professionals act strategically: less boilerplate, more systems with AI that deliver measurable value. That is the employment that not only endures but also grows.