Türkiye's AI Action Plan (2026-2030): What It Is and What It Means for Companies

In short: Türkiye's AI Action Plan (2026-2030) is a five-year roadmap, coordinated by the Ministry of Industry and Technology, that sets out 16 actions to build the country's artificial intelligence capacity. It was announced by Presidential Circular 2026/9 in the Official Gazette (Resmi Gazete) of 18 August 2026, and the full text is on the Ministry's website. The actions sit under four pillars, Aware (Fark Et), Benefit (İstifade Et), Construct (Üret) and Direct (Yönet), and cover talent, data, computing power, public procurement, funding and regulation.

Two documents are all you need to read the plan: the 44-page plan itself and the one-page circular that announced it in the Official Gazette. For most companies, the plan means programs that are still to come, since it is written mostly in the future tense and describes programs, councils and targets to be put in place by 2030. Page numbers in this post are the printed page numbers of the Turkish edition, and the English edition follows almost the same layout.

The full text of the plan is on the Ministry of Industry and Technology's website, and the circular that announced it is in the Official Gazette. The documents are:

• Türkiye's Artificial Intelligence Action Plan: the English edition, 44 pages, about 96 MB
• Türkiye Yapay Zeka Eylem Planı (2026-2030): the Turkish original, 44 pages, about 116 MB
• The Ministry's action plan page in English, which leads to both editions
• Presidential Circular 2026/9 (in Turkish): Official Gazette of 18 August 2026, issue 33344, one page

Türkiye's AI Action Plan (2026-2030) is a roadmap that sets the priority actions, targets and responsible agencies for growing Türkiye's AI capacity. In the words of the plan's executive summary, it "outlines priority actions that can be implemented to rapidly expand Türkiye's artificial intelligence capacity, produce measurable results, and be carried out through interagency coordination" (p. 1). The cover names the Ministry's General Directorate of National Technology and Artificial Intelligence.

The plan's vision is "to transform Türkiye into a hub that produces reliable products and services in the field of artificial intelligence, can scale AI solutions in its strategic sectors, and attracts investment and talent to the region" (p. 3).

The plan is built on four pillars with four actions each, 16 actions in all. It ties these actions to a five-layer architecture: energy, computing hardware, infrastructure, models and data, and applications. Talent, funding and regulation are defined as cross-cutting components that support every layer (p. 1).

According to the preface by Mehmet Fatih Kacır, Minister of Industry and Technology, the Ministry held workshops with nearly 500 industry professionals from public institutions, the private sector, universities and civil society. Separately, the Citizen Opinion Survey brought in "over 2,200 substantive action proposals." The same preface says the plan is expected to generate "economic value exceeding 1 trillion TL." The document does not explain how that figure was calculated.

The Official Gazette carried a one-page circular announcing the plan, not the plan itself. Circular 2026/9 of the Presidency, dated 17 August 2026, appeared in the Official Gazette of 18 August 2026, issue 33344. Its subject is the "Türkiye Yapay Zeka Eylem Planı (2026-2030)".

The circular, which is in Turkish, makes four points.

1. It states that the earlier National Artificial Intelligence Strategy (2021-2025) has been implemented with significant gains, and that a new action plan is needed to update the country's vision.
2. It says the plan was prepared under the coordination of the Ministry of Industry and Technology, in cooperation with the public sector, the private sector, universities and civil society. It names the plan's principles as human-centeredness, trustworthiness, ethical responsibility, digital sovereignty and sustainable development.
3. It lists the goals: raising AI awareness, developing qualified human resources, strengthening data and computing infrastructure, spreading AI applications in the public and private sectors, and building an internationally competitive and trustworthy AI ecosystem.
4. It says the plan will be published on the Ministry's website (www.sanayi.gov.tr), and that all public institutions and organizations must carry out the duties and responsibilities the plan assigns them with care.

The plan divides its 16 actions into four pillars of four actions each. Based on the pillar definitions in the plan, Aware covers awareness, talent, the data ecosystem and governance frameworks. Benefit covers building and using infrastructure and its adoption in the public and private sectors. Construct covers funding, model development and physical AI, and Direct covers global investment, international diplomacy and the security of models (pp. 1 and 5).

In the table, MoIT is the Ministry of Industry and Technology and PCS the Presidency of Cybersecurity.

The plan does not say, action by action, from which date periods such as "the first 6 months" or "the first 12 months" are counted.

One of the Direct targets is about language models. According to the plan, a joint Turkic-languages large language model covering the Oghuz, Kipchak and Karluk language families will be built, starting as a pilot with at least two member states of the Organization of Turkic States, and its first version is due in production by the end of 2027 (p. 25). Work on Turkish large language models will also continue under Action 11 (p. 22). We explain what a large language model is and how it works in What is an LLM.

The plan's highest decision-making body will be the National Artificial Intelligence Council, chaired by the President. According to the plan's governance chapter, the Council will approve annual implementation plans and publish a "National Artificial Intelligence Policy Charter" every year. The Ministry's General Directorate of National Technology and Artificial Intelligence will act as its secretariat, and a Program Office will be set up to track implementation. The Presidency of Cybersecurity's General Directorate of Public Artificial Intelligence will also sit in that office (p. 9). The National AI Ethics Council will provide ethical guidance, and technical security assessment will be run under the coordination of the Presidency of Cybersecurity with the TÜBİTAK Artificial Intelligence Institute (p. 10).

To reduce overlapping mandates, the plan also calls for a responsibility matrix showing, for each action, which agencies are responsible, accountable, consulted and informed (p. 10). As AI use grows inside a company, the same question comes up there: who approves which system, who monitors it and who reports to management? RuneLab's Enterprise AI Governance Framework and Operating Mechanisms course works through that question. It covers building an inventory of in-house models, AI features inside SaaS products and agents, risk-based approval gates and reporting to the board.

Progress will be tracked on the National AI Progress and Transparency Portal. According to the plan, the portal will publish the responsible agency, subtasks, timeline, budget and spending, indicators and a green, yellow or red status for every action (p. 10). The first public progress report will be published on the portal within 12 months of the plan's publication and then every year (p. 31). The documents do not give the portal's address.

The plan runs in two phases (p. 2).

• 2026-2027: setting up the core governance structures, piloting Data Spaces, launching the first AI applications in public institutions and settling the investment process for computing capacity.
• 2028-2030: scaling the pilots, commercializing sector-specific foundation models, deepening international cooperation and meeting the investment targets.

According to the plan, actions that mature early will be scaled without waiting for this timeline. Public budget, private investment, international funds and incentives will be considered together for funding, and each action's resource needs will be set out separately in the annual implementation plans (p. 2). That is why the document gives no budget per action.

For companies, the action plan is less a source of support available today than a map of programs and rules to be set up in the coming years. Most of it is written in the future tense, with phrases such as "will be established" and "will be defined." Based on the text of the plan, these are the mechanisms that matter for different groups of companies.

For startups and SMEs, the most direct tool is the "GPU for Everyone" Program, which will give GPU-hour credits to startups, SMEs and researchers. The plan defines a GPU-hour as "a computing unit used for training and running AI models." For eligible startups and SMEs, the credits will come bundled with software, cloud and hardware support, and the packages will also include technical consulting and mentorship (p. 18).

Vouchers under the SME AI Voucher and Ready-to-Deploy Solutions Package Program will cover software subscriptions, cloud GPU and CPU use, data services and security or compliance testing (pp. 19-20). Sectoral AI Pilot Programs, with healthcare efficiency, grid and energy optimization, smart manufacturing and quality, and geographic information systems as the first priorities, will open with open calls. They will give increasing support through the idea, prototype, field trial and scaling stages (p. 19). The National AI Champions Program will also select companies and consortia in priority sectors through open calls and competitive evaluation (p. 15).

For companies selling AI solutions to the public sector, the plan sets out a path to recognition and procurement. Solutions built by technology startups and private companies and successfully used in the public sector will receive a "digital badge." Arrangements will be made so that badged solutions are listed in the State Supply Office (DMO) Tekno Catalog application and can be procured by public institutions (pp. 18-19).

The plan also targets at least five AI-enabled citizen services on e-Government, piloted and then scaled. The first scenarios to be piloted are personalized service guidance, preliminary application checks and smart document support (p. 19). We explain how AI systems that carry out tasks step by step on a user's behalf work in What is an AI agent.

Companies that hold data will be able to share it in a controlled way, "on a voluntary basis and for mutual benefit," through the Data Spaces and Secure Data Sharing Program (p. 14). Planned incentives for organizations that supply data include tax or R&D deductions, GPU and data credits and "Trusted Data Provider" status (p. 15).

For employers, the key mechanism is the Workforce Transformation Program. It will offer short, applied, sector-specific modules for employees who will use AI tools to become more productive in their current jobs. According to the plan, businesses will be able to use R&D center and technopark incentive mechanisms to direct their employees to the program (p. 13). The target is for at least 100,000 employees to complete these modules within 24 months (p. 14).

Under the sector-specific modules sits a common foundation for everyone: knowing what AI tools can and cannot do. RuneLab's AI Literacy for Organizations course builds that foundation. It covers how large language models work, limits such as hallucination and bias, writing prompts for everyday tasks and using AI in line with company data privacy, with no technical background required and with examples from participants' own departments.

Regulatory sandboxes will be set up in regulated sectors, led by finance, healthcare, energy, mobility and telecommunications. They will have a single digital point of contact and time-limited trials, and a "graduation pathway" will let successful pilots move quickly into permanent legislation or public procurement (p. 26).

International investors and data center companies will get a single-window investor portal, and preliminary eligibility assessments and investment roadmaps will be completed within 30 business days at most (p. 24). The plan names the Data Center and Artificial Intelligence calls under the HIT-30 High-Tech Investment Program as the incentive framework for these investments (pp. 16 and 24).

There is no application to the plan itself. Neither the circular nor the action plan text contains a form, date or rule for companies or individuals to apply to "the action plan." The plan says some programs will run through calls or applications later.

• GPU for Everyone: it will operate through "transparent application and evaluation processes," with the first call due within the first 6 months (p. 18).
• SME vouchers: SMEs will submit their needs through a "digital application platform" and be matched with accredited service providers (p. 20).
• Sectoral AI Pilot Programs and the Champions Program: they will start with open calls (pp. 15 and 19).
• Regulatory sandboxes: a single digital point of contact will be set up (p. 26).
• AI Growth Fund: a dedicated investment window for AI startups will open, and applications under it will be treated as priority investments (p. 21).
• Regional Centers of Excellence: at least six access points will be set up as university and technology park service desks (p. 20).

The documents give no application address, dates, eligibility criteria or support amounts for these programs, and they do not say through which channel the calls will be announced. As of 4 November 2026, the Ministry's announcements list no GPU for Everyone or SME voucher call, the Ministry's plan page has no application link, and the address of the Progress and Transparency Portal has not been published.

The action plan is a roadmap, not a law or regulation. The plan's glossary states that "concepts requiring legal binding force will be defined separately within the framework of relevant legislative work and secondary regulations" (p. 27).

The plan names the areas where it will treat AI applications as high-impact. These are healthcare, education, employment, social assistance, credit assessment, biometric identification, law enforcement and judicial processes, and systems that directly affect children and vulnerable groups (p. 16).

The same list is useful when a company reviews its own AI use. Putting a model that ranks job applicants and a tool that summarizes meeting notes through the same review either wastes resources or leaves dangerous gaps. RuneLab's AI Risk Classification and Tiered Control Mapping course teaches how to draw that line. It works through sorting use cases into risk tiers by purpose, level of autonomy, the people affected and data sensitivity, and attaching the right controls to each tier.

Action 4's targets are to publish the proportionate risk guidance and sector annexes within the first 12 months, establish the National AI Ethics Council and fully operationalize TÜRKPATENT's "AI Fast Track" (p. 16). The documents do not say when, or through which instrument, these will take effect.

The new action plan follows the National Artificial Intelligence Strategy 2021-2025 and builds on it. The earlier strategy was put into effect by Presidential Circular 2021/18 of 19 August 2021, published in the Official Gazette of 20 August 2021, issue 31574. Circular 2026/9 states that the strategy was implemented with significant gains and that a new action plan is needed.

According to the preface of the plan, the new plan draws on four documents: the 2021-2025 National Artificial Intelligence Strategy, the 2024-2025 Action Plan, the report of the Grand National Assembly of Türkiye's Artificial Intelligence Research Commission and the National Artificial Intelligence Policy Document of the Presidential Council for Science, Technology, and Innovation Policies. The plan says the institutional knowledge built through the 2021-2025 strategy and the 2024-2025 action plan will be spread across society under the Aware pillar (p. 7).

The full text was published on the website of the Ministry of Industry and Technology. The English PDF has 44 pages and is about 96 MB, and the Turkish original is about 116 MB on the same site. The Official Gazette carries only the one-page circular that announced the plan.

Presidential Circular 2026/9 of 17 August 2026, which announced the plan, was published in the Official Gazette of 18 August 2026, issue 33344. The full text of the plan is not in the Official Gazette but on the Ministry's website, as the circular states.

No, there is no application to the plan itself. The plan says programs such as GPU for Everyone, SME vouchers, sectoral pilot programs and regulatory sandboxes will run through calls or applications later. The documents give no dates, addresses or criteria for these programs.

GPU for Everyone is Action 6 of the plan, a program that will give GPU-hour credits to startups, SMEs and researchers. The plan aims to open the first call within the first 6 months and to allocate at least 2 million GPU-hours of credits a year in the first phase. Annual allocations are to reach at least 20 million GPU-hours by the end of 2028.

The plan's vision is to make Türkiye a hub that produces reliable AI products and services and attracts investment and talent to the region. Its headline targets are training 10,000 advanced AI experts and 100,000 AI application professionals, reaching 1 GW of installed data center capacity, mobilizing $10 billion in private investment and running regulatory sandboxes in at least five priority sectors.

The Ministry of Industry and Technology coordinates the plan. The highest decision-making body will be the National Artificial Intelligence Council, chaired by the President, with the General Directorate of National Technology and Artificial Intelligence as secretariat and a Program Office tracking implementation. The Presidency of Cybersecurity leads some actions.

The plan has 16 actions across four pillars, four in each. The pillars are Aware, Benefit, Construct and Direct. The actions range from literacy and talent to computing power, funding and AI security.

• Presidency of the Republic of Türkiye, Circular 2026/9 on the Türkiye Yapay Zeka Eylem Planı (2026-2030) (in Turkish): Official Gazette, 18 August 2026, issue 33344. The rationale, principles and goals, and where the plan is published
• Ministry of Industry and Technology, Türkiye Yapay Zeka Eylem Planı (2026-2030), Turkish full text: the Turkish original and its printed page numbers
• Ministry of Industry and Technology, Türkiye's Artificial Intelligence Action Plan page: both editions of the plan and whether an application link exists
• Ministry of Industry and Technology, Türkiye's Artificial Intelligence Action Plan: the English edition, used for the official English names, the preface, executive summary, vision, pillars, 16 actions, governance, glossary and measurement cycle
• Presidency of the Republic of Türkiye, Circular 2021/18 on the National AI Strategy (2021-2025) (in Turkish): Official Gazette, 20 August 2021, issue 31574. The circular that put the earlier strategy into effect

• What is sovereign AI?
• What is an AI agent?
• What is an LLM? Large language models explained
• How to set up offline AI with a local LLM
• Course: Enterprise AI Governance Framework and Operating Mechanisms