Türkiye's National AI Action Plan (2026-2030): What Is In It, and Who Will Run It?
Announced in June, the plan became binding through a circular published in the Official Gazette on 18 August. Its targets do not all carry the same weight, and there is one subject the plan stays silent on.
Türkiye's National AI Action Plan (2026-2030) was first presented in public on 13 June 2026, at the Türkiye Artificial Intelligence Summit. It became binding through Presidential Circular 2026/9, signed on 17 August 2026 and published the next day in issue 33344 of the Official Gazette. The plan sets out a road map for the next five years.
The Ministry of Industry and Technology, which prepared the plan, did not keep the process inside government: the private sector, universities and civil society organizations took part as stakeholders. According to the plan's own foreword, close to 500 industry professionals joined the workshops, and a Citizen Opinion Survey collected more than 2,200 submissions.
Why a New Plan?
Türkiye already had an AI strategy. The National Artificial Intelligence Strategy (2021-2025), which entered into force through Circular 2021/18 and was published in the Official Gazette on 20 August 2021, ran for five years and was followed by a 2024-2025 Action Plan. The new plan positions that period as continuity rather than a break: it says the two documents built substantial institutional capacity, and that its own job is to spread and deepen that capacity across society.
The plan was not written from scratch. According to Minister Mehmet Fatih Kacır's foreword, the preparation drew on the 2021-2025 Strategy and the 2024-2025 Action Plan, as well as the report of the Turkish Grand National Assembly's AI Research Committee and the National AI Policy Document of the Presidency's Science, Technology and Innovation Policies Board.
The National AI Strategy had been published with 24 goals and 119 measures under six strategic priorities; the 2024-2025 Action Plan was not that document's first action list but an update aligned with the 12th Development Plan. In the new plan the actions sit in the same document as the strategy: all sixteen are defined the moment it takes effect.
The Vision: A Regional Hub
The plan's vision is to make Türkiye a hub that produces trustworthy AI products and services, can scale its solutions in strategic sectors, and attracts investment and talent in its region. Two ideas sit behind it. The first is industry: in fields where the country already has strong manufacturing capability (automotive, home appliances, defense, textiles and food), AI is expected to deliver not only productivity but new product categories and export channels. The second is public services: the e-Government infrastructure is meant to evolve into an AI-enabled platform, with the state also acting as the first customer for domestic solutions.
The principle underneath the vision is named "digital sovereignty" in the plan: while Türkiye stays open to international cooperation and global capital, it also aims to build secure, resilient and auditable national compute capacity for critical public services and strategic sectors. The instrument that makes this concrete is the "Compute Portfolio," which gathers public reserve capacity, domestic data centers, international cloud partnerships and regional capacity alliances under one heading. According to the plan, Türkiye takes part in the Digital Europe Programme and the EuroHPC Joint Undertaking as a participating country, which gives it access to the high-performance computing infrastructure available under those programs.
The Architecture: Five Layers
The plan builds the AI ecosystem on five layers that feed one another: energy, compute hardware, infrastructure, models and data, and applications. Each layer rests on the one below it, so a bottleneck in energy or hardware constrains everything above it. Three cross-cutting components feed all five layers: talent; financing; regulation and trust. The sixteen actions were designed around this structure, each mapping to one axis and to one or more layers.
In practice the plan rests on four principles.
Speed and clarity. A small number of high-impact priorities, short-term deliverables, measurable targets.
Market focus. Value verified in the field through pilots in the public and private sectors, then rapid scaling for what works. The ministry calls this the "scan-pilot-scale" loop.
Trust and accountability. A proportionate approach to risk; safety backed by transparency and independent assessment.
Phased delivery. The plan's own calendar splits into two windows: in 2026-2027, standing up the core governance bodies, piloting data spaces and launching the first public-sector applications; in 2028-2030, scaling those pilots, commercializing sector-specific foundation models and delivering the investment targets.
Four Axes, Sixteen Actions
The plan is built on four axes: "Recognize," "Benefit," "Produce" and "Govern." Each axis carries four actions.
Recognize targets social and institutional readiness. Three of its four actions concern education and data: a National AI Literacy and Safe Use Program, a research capability program that will train 10,000 advanced specialists, and a data-driven AI push. The fourth is a regulatory environment that protects citizens' rights while giving investors predictability.
Benefit covers building the infrastructure and actually putting it to use: a competitive, low-carbon national compute backbone, the "GPU for Everyone" program, leading the public sector's own AI transformation and shaping the market through public procurement, and sector programs with SME vouchers. That last action aims to create a single support track running from idea to product.
Produce focuses on production capability: AI Growth Zones and Regional Centers of Excellence, a financing ladder running from basic research to entrepreneurship, a secure export track built on sector-specific models, and domestic manufacturing capacity in physical AI and robotics.
Govern concerns global positioning: legal predictability and competitive incentives for AI and data center investment, AI diplomacy, shared infrastructure and model development through regional partnerships, security evaluation capacity, and regulatory sandboxes. A joint Turkic-languages model with members of the Organization of Turkic States also sits on this axis.
Targets for 2030
The plan sets concrete, dated targets for every action. The headline ones:
- AI literacy training for 5 million citizens within two years, across all 81 provinces (1.5 million in the first 12 months)
- Training 10,000 advanced AI specialists
- Bringing 100,000 AI application professionals into the workforce
- Opening more than 2,000 public data sets in machine-readable form through the National Data Library
- Raising installed AI and data center capacity to at least 1 GW by 2030, and reaching at least 10 TWh of annual low-carbon electricity purchase agreements
- Raising annual GPU-hour allocation under "GPU for Everyone" to at least 20 million by the end of 2028 (at least 2 million a year in the first phase)
- Mobilizing at least $10 billion in mainly private-sector investment for AI infrastructure
- Allocating at least 2 percent of the total public investment program budget to AI projects
- Establishing a two-step financing ladder through the National AI Research Fund (at least 10 billion TL) and the AI Growth Fund (15 billion TL), with scale-up support for at least 20 startups
- Launching regulatory sandboxes in at least five priority sectors
- Creating more than 1 trillion TL in economic value from public and private sources over the life of the plan
These targets do not carry the same weight. The 2 percent share of the public investment program budget is a budget decision and can be tracked directly. The 1 trillion TL in economic value is an aggregate whose calculation method is not stated. Progress against them is expected to be published on the National AI Progress and Transparency Portal that the plan proposes.
Who Will Run It?
Of the bodies described below, the Council, the Program Office, the Ethics Board and the Portal do not exist yet; the plan proposes them as its own governance design.
Delivery is to be led by a National AI Council chaired by the President. The council would set strategy, priorities and resource allocation, approve annual implementation plans, resolve jurisdictional disputes between institutions, and publish a National AI Policy Letter each year to signal direction to the public and to investors. Its secretariat would sit with the Directorate General of National Technology and Artificial Intelligence within the Ministry of Industry and Technology, with delivery tracked through a Program Office established in the same directorate. The Program Office would also include the Cybersecurity Directorate's Directorate General of Public AI, which is to plan and coordinate AI applications across the public sector.
Ethical guidance is assigned to a separate body: a National AI Ethics Board drawn from government, the private sector, academia and civil society would publish guidance around fairness, data privacy, explainability and human oversight, and give opinions on high-impact systems. The board's remit is drawn explicitly in the plan: its role is guidance and ethical assessment, while binding supervision and enforcement stay with sector regulators. Technical security evaluation would be coordinated by the Cybersecurity Directorate with support from the TÜBİTAK AI Institute.
On monitoring, the Transparency Portal would publish each action's alignment with its target, budget and timeline in public, using green, amber and red status colors. The first report is expected within twelve months of the plan's publication, and would carry an independent assessment component prepared with academic or international input.
Four Things Delivery Will Be Judged On
The plan sets targets as a trio of short-term deliverable, accountable institution and measurable indicator. A national plan is tested in delivery; four points will decide this one.
How success is measured. The plan judges the literacy program on learning outcomes rather than headcount: assessment before and after the program, five competency areas, a minimum pass threshold for micro-credentials. It also states that indicators will be tracked at the level of outcomes rather than outputs. What is left open is disaggregation: the plan does not say whether this data will be published broken down by province, age and education level.
Regional equity. A target of reaching all 81 provinces guarantees geographic coverage, which is not the same as equality of opportunity between regions. The scale on the production side shows the difference: the plan foresees at least three Regional Centers of Excellence and at least six access points. Workshops in 81 provinces, production capacity at six locations. That is exactly why the question of disaggregated data in the previous point matters; without it, there is no way to see whether the first indicator has quietly taken the place of the second.
Coordination. The plan puts dozens of institutions on the same timetable, and for each action it foresees a responsibility matrix naming the institutions that are responsible, accountable, consulted and informed. What matters is not that the matrix exists, but whether it is made public. If jurisdictional disputes are not visible on the Portal, neither is the place where delays accumulate.
The copyright gap. Intellectual property appears in the plan in several places: a TÜRKPATENT "AI Fast Track" is defined, users are promised protection for their rights over model, data and code assets, and rules are set for processing personal data in model training. There is no action, however, on the copyright regime covering the content that models are trained on. The plan aims to develop vertical models for domestic cultural output, digital art, gaming, cinema and advertising, without proposing any rule on the rights of the people who own that content.
For the first three points, the Transparency Portal and the independent assessment component are counterweights; what matters is not that they exist, but whether they arrive on the timetable the plan sets. For the fourth, no counterweight is defined.
As an association, we will follow the sector pilot programs, the "GPU for Everyone" and voucher mechanisms aimed at SMEs, and the data access agenda, and we will share our committees' findings with our members. We will put ourselves forward for civil society representation in multi-stakeholder bodies such as the National AI Ethics Board.
