Turkey's Artificial Intelligence Action Plan
One of the most comprehensive steps taken by Turkey regarding artificial intelligence has become official. Türkiye Artificial Intelligence Action Plan (2026-2030) entered into force by being published in the Official Gazette on August 18, 2026, with the Presidential Circular. The plan was actually announced to the public for the first time at the Türkiye Artificial Intelligence Summit on June 13; With the circular published in August, it became the official road map for the 2026-2030 period.
The numbers are quite ambitious at first glance: private sector-oriented investment of at least 10 billion dollars, 1 gigawatt data center capacity, AI literacy training for 5 million people, 10 thousand advanced artificial intelligence experts, 100 thousand application professionals, 25 billion TL research and growth funds, "GPU for All" program and at least 2 percent share of AI projects in public investments.
However, it would be incomplete to evaluate this plan solely on investment figures.
The real story is that Turkey is no longer just using artificial intelligence as a software category; It begins to be addressed as a strategic infrastructure issue that simultaneously encompasses computing power, data, human resources, entrepreneurship, energy, robotics and public policies.
What will determine the success of the plan will not be how widely used tools like ChatGPT are in Turkey.
The more fundamental question is:
Will Turkey be one of the countries consuming artificial intelligence, or will it be able to truly take part as a producer in the economic value chain of this technology?
“Recognize, Exploit, Produce and Manage”: The Four-Layered Logic of the Plan
Türkiye Artificial Intelligence Action Plan is built on four basic axes titled "Recognize, Use, Produce and Manage" and includes a total of 16 priority actions. The Ministry of Industry and Technology defines the main goal as Turkey being among the countries that design, produce and use artificial intelligence technologies and can compete on a global scale.
These four words actually answer four different problems of the AI economy.
“Realize” represents the society and business world's understanding of artificial intelligence, “Use” represents the application of existing technology to the economy and public services, “Produce” represents Turkey's ceasing to be a country that only uses foreign models, and “Manage” represents the security, ethics, regulation and institutional coordination side of this whole process.
Therefore, the plan is broader than just the "Turkey should build its own ChatGPT" approach.
Developing models is just one part of it.
For economic independence in the AI era, data, computing capacity, energy, human resources, capital and technology transfer mechanisms are also needed.
New Natural Resource in the Age of Artificial Intelligence: Computing Power
One of the most critical topics of the plan is the calculation infrastructure, which does not attract as much attention from the public as other items.
Because in the modern artificial intelligence economy, algorithm alone is not enough.
GPU required.
High speed network required.
Storage infrastructure required.
Data center required.
And it takes a huge amount of energy to fuel all of this.
Within the scope of the plan, Türkiye aims to increase its data center installed power to at least 1 gigawatt by 2030. In the same period, it is aimed to mobilize at least 10 billion dollars of investment led by the private sector in artificial intelligence, cloud and data center infrastructure. The Ministry of Industry and Technology also associates this investment vision with the data center and artificial intelligence programs within the scope of HIT-30.
It is enough to look at the change in the AI industry to understand why this figure is important.
Once upon a time, one of the technology company's most important assets was its software code.
Today, the competitive advantage of some AI companies is direct:
How many GPUs can you access?
It started to be measured with the question.
Global demand for Nvidia GPUs, huge data center investments and AI infrastructure spending of companies such as Microsoft, Google, Amazon and Meta reaching hundreds of billions of dollars are the result of this.
If Turkey wants to grow its own AI ecosystem, local companies' constant dependence on the pricing and capacity conditions of foreign cloud providers could be a significant disadvantage.
So data center policy is not a technicality.
One of the centers of digital sovereignty policy.
“GPU for All” Perhaps One of the Most Critical Programs of the Plan
For a large technology company, purchasing a GPU may be costly, but it is possible.
The situation is completely different for a university research group or small enterprise.
Today, the first problem faced by a five-person Turkish startup that wants to develop a powerful AI model may often be not the idea, but the computational cost.
For this reason, the “GPU for All” program included in the plan is strategically important.
The program facilitates access to computing resources for startups, SMEs, university researchers and public projects; It is planned to create separate access windows and allocation quotas for different groups.
If such an infrastructure can be operated in a truly cost-effective, reliable and sufficient capacity, it can reduce one of the important barriers to AI entrepreneurship in Turkey.
Because in the age of AI, the problem of innovation is no longer just:
“Do we have good software developers?”
not.
Also:
“Does our good programmer have a GPU to train his model?”
question.
Perhaps one of the biggest opportunities for Turkey is here.
Not every startup needs to establish its own data center.
It needs to be able to access the common computing infrastructure.
It's Not 10 Billion Dollars Itself, It's Where They Will Spend It That Matters
“$10 billion AI investment” is naturally a strong headline.
But the more important question from an economic perspective is:
At what level of value will this $10 billion create value?
If the majority of the investment is limited to purchasing imported GPUs and hosting the services of foreign technology companies in Turkey, a significant infrastructure gain will be achieved, but the added value of domestic technology may be limited.
In the stronger scenario, this infrastructure is:
domestic AI initiatives,
model development companies,
defense and robotic technologies,
health AI solutions,
industrial automation,
financial technologies,
creative AI products,
SaaS initiatives
It can turn into a nourishing economic layer.
For Turkey, the main goal is not to own a GPU.
Owning intellectual property that runs on the GPU.
The difference between these two is big.
A country may establish a data center but remain a low-value-added infrastructure provider in the digital economy.
The real economic value is created in the model, product, data, software and technology that can be sold to global users.
AI Training for 5 Million People: The Number is Big, But the Level of Education is More Important
One of the most ambitious goals of the plan on the society side is to provide artificial intelligence literacy to 5 million people within 24 months.
In addition to the program planned to be carried out through workshops to be established in 81 provinces, it is aimed to train 10 thousand advanced AI experts and 100 thousand artificial intelligence application professionals. It is also planned that 1 million employees will undergo professional transformation evaluation and 500 international experts will participate in research, training or entrepreneurship activities in Turkey.
Here it is necessary to distinguish three different levels of competence.
The first is AI literacy.
Understanding what artificial intelligence can do, where it makes mistakes, information security, use of prompts and basic ethical problems.
The second is AI implementation.
Integrating existing models into business processes, using APIs, establishing automation, preparing data and developing products.
The third is AI research and engineering.
Training models, performing inference optimization, developing new architectures and directly participating in the international technology race.
It will be especially the quality of the second and third groups that will determine Turkey's real competitive advantage.
Because teaching millions of people to use ChatGPT is beneficial.
But what will change the country's economy is not the number of people using ChatGPT, but the number of engineers and companies that can develop technology at the scale of ChatGPT.
AI Literacy May No Longer Be Optional in the Education System
There is also a longer-term cultural impact of the plan here.
AI literacy can turn into a basic skill similar to computer literacy in a few years.
In the 1990s, being able to use Word was a special skill.
In the 2000s, being able to do research on the internet was an advantage.
In the 2010s, the use of social media and digital tools became normal.
Today, being able to work with AI is increasingly becoming a basic professional skill.
However, the training is only "how to write a prompt?" If it is reduced to this level, a significant part of the opportunity will be missed.
True AI literacy also means:
hallucination,
source verification,
copyright,
personal data security,
bias problems,
model limits
It requires understanding.
So the goal should not only be to produce more AI users, but also to produce more conscious AI users.
National Data Library: Data Problem Must Be Solved Before Model
Most discussions about artificial intelligence revolve around models.
However, the basic raw material of the model is data.
For this reason, the National Data Library is one of the remarkable projects in the Türkiye Artificial Intelligence Action Plan. One of the goals is to make at least 2 thousand public data sets accessible. Creating secure data areas in health, agriculture, defense, e-commerce and similar areas is also included in the data policy of the plan.
Quality may be more important than quantity here.
Publishing 2 thousand data sets alone may not create great value.
Your data:
machine readable,
current,
clean,
standardize,
well documented,
Accessible via API
It should be.
Well-prepared public data can turn into a real innovation infrastructure for startups.
For example, transportation data can create new mobility initiatives.
Agricultural data can support precision farming practices.
Energy data can be transformed into optimization products.
Geographic data can form the basis of logistics and urbanization systems.
If the data is opened correctly, the state will not only publish information.
It provides the infrastructure on which new companies can build products.
What Can Artificial Intelligence Fast Patent Pipeline Change?
The plan also envisages the commissioning of the Artificial Intelligence Fast Line application by TÜRKPATENT.
The aim is to review industrial property applications related to artificial intelligence technologies more quickly and to establish transparent declaration mechanisms regarding the use of AI.
At first glance, this may seem like one of the smaller items of the plan.
However, the patent process is critical for technology startups.
Especially in deep-tech companies seeking investment:
intellectual property,
patent portfolio,
licensing,
technology ownership
It can directly affect company valuation
Reducing legal uncertainty regarding the technology developed by an AI startup means reducing risk for the investor.
But in the age of AI, the job of the patent system is not easy.
The status of the inventions created by the model, the training data used, human contribution, and the limits of intellectual property of the model outputs are still debated around the world.
For this reason, the clarity of the legal framework will be as important as rapid examination.
25 Billion TL Fund: Can Turkey Grow Its AI Initiatives?
There are two separate mechanisms in the financing leg of the action plan.
It is planned to allocate 10 billion TL for the National Artificial Intelligence Research Fund and 15 billion TL for the Artificial Intelligence Growth Fund.
While it is aimed for at least 40 percent of the projects supported in the research fund to progress to the pilot or prototype stage, the growth fund is aimed to contribute to the Series A/B or scaling investments of at least 20 startups.
This distinction touches on a very valid problem.
One of the common problems of technology entrepreneurship in Turkey is not only establishing a company but also growing it.
A startup can develop a prototype.
He can find his first customer.
He can even sell his product in Turkey.
But when you want to grow on a global scale, millions of dollars of capital are needed.
Especially in AI companies, GPU, data, engineer and global sales costs are high.
Therefore, scale-up financing is as critical as the seed stage.
If the 15 billion TL fund can carry successful startups to the international market instead of just keeping them alive, the economic return of the program could be much higher.
But are 20 Successful Companies Enough?
Here, the target should not be evaluated only by the number of attempts.
In the AI economy, even a few successful companies can create enormous value.
OpenAI.
Anthropic.
DeepSeek.
Mistral.
Scale AI.
Perplexity.
Each of these has produced billions of dollars of value in their own ecosystems.
Turkey does not necessarily need hundreds of basic model companies.
Perhaps the more realistic opportunity is in vertical AI, that is, artificial intelligence products specific to certain sectors.
Health.
Defense.
Production.
Tourism.
Logistics.
Finance.
Retail.
Agriculture.
Education.
Global competitive advantage may be more attainable when AI is integrated into sectors where Turkey is already strong.
For example, developing the world's largest general-purpose LLM is prohibitively expensive.
But developing one of the world's best AI systems for quality control in textile production is a much different problem.
The strategic opportunity may be here.
At Least 2 Percent for AI in Government: Government Could Be the First Big Customer
Within the scope of the plan, it is aimed to allocate at least 2 percent of the public investment program budgets to artificial intelligence projects.
The effect of this article may be greater than the number.
Because the role of the state in technology ecosystems is not just to provide funds.
The state can also be a customer.
For example, let's imagine that a Turkish AI startup is developing a healthcare system.
Maybe his biggest problem is not finding investments, but finding his first big customer.
If public health institutions become customers of successful pilots, the company will obtain real data, real users and real references.
Similar scenarios:
energy optimization,
smart cities,
geographic information systems,
public call centers,
document processing,
cyber security,
transportation
It is also possible in the fields.
Therefore, it may be important to link the public AI budget with the domestic technology purchasing and piloting mechanism.
Why Are Health, Energy and Industry Priority?
In the plan, sectors such as efficiency in health services, energy and network optimization, smart production, quality control and geographical information systems are particularly highlighted.
The logic of this is understandable.
AI's biggest economic impact may not just come from chatbot subscriptions.
The real big value can come from the productivity of existing sectors.
Reducing a factory's error rate by 5 percent.
Optimizing the consumption of an energy grid.
To speed up a hospital's imaging processes.
Improving the routes of a logistics company.
Each of these are very economically measurable gains.
Therefore, for Türkiye, the AI strategy is only about "how many technology startups have been established?" It needs to be measured not only in terms of AI, but also in terms of how much AI the current industry uses.
Why Do Physical AI and Robotics Hold a Special Place?
One of the striking aspects of the plan is that it sees robotics and physical AI technologies as a separate strategic area.
It is aimed to transfer the autonomous systems and robotic competencies developed in the defense industry to the civilian sectors. Launching three priority pilot applications, two university-industry robotics centers, three test sites and three defense-to-civilian technology transfer projects by the end of 2027; Export or joint development agreements are planned to be made until 2030.
This is a very logical area for Turkey.
Because the defense industry has gained in recent years:
computer vision,
autonomous navigation,
sensor fusion,
drone technologies,
embedded systems,
robotics
Their competencies are transferable to the civilian economy.
Warehouse robots.
Agricultural robots.
Industrial inspection systems.
Autonomous logistics.
Smart machines.
AI-supported production lines.
These are areas where Turkey can develop products not only in software but also in combination with hardware + AI.
Why Are Test Sites Necessary for Autonomous Vehicles?
Fully autonomous driving is not possible just by making a good car.
Digital maps.
Road infrastructure.
5G/6G communication.
Insurance.
Legislation.
Liability rules.
Test environments.
Sensor standardization.
These all need to work together.
For this reason, it is important that the plan allocates special space for autonomous driving infrastructure and test sites.
Turkey is a strong country in automotive production.
Classical automotive production capability if AI transformation is managed correctly:
software-defined vehicle + autonomous driving + AI
It can be transformed into an ecosystem.
Otherwise, Turkey may remain a production center that produces automobiles but purchases the most valuable software and AI layers of the vehicle from abroad.
The economic difference is huge.
What Does the 10 TWh Low-Carbon Electricity Target Have to Do with AI?
One of the plan's less noticeable but perhaps most critical elements in technology news is energy.
10 terawatt-hours of low-carbon electricity purchase agreement volume is targeted by 2030.
The reason for this is simple.
AI consumes energy.
Too many.
One of the questions as important as the GPU price for a data center investor:
“Where will I buy electricity and at what cost?”
If Turkey wants to become a data center hub, it has to provide cheap, reliable and low-carbon energy as possible.
Therefore, AI policy cannot be considered separately from energy policy.
In the coming period, technology competition will unexpectedly:
GPU + electricity + water + land
It is likely to turn into the equation.
One of Turkey's advantages may be its ability to plan its renewable energy capacity simultaneously with data center growth.
AI Growth Zones: Are Tech Free Zones Coming?
The plan also aims to create AI growth zones to accelerate large artificial intelligence and data center investments.
The approach here may be slightly different from classic technology parks.
For a large data center investment, the company does not only need an office.
High capacity electrical connection.
Fiber infrastructure.
Cooling.
Land.
Energy security.
Accelerated permit processes.
These need to be ready.
If growth regions can provide this infrastructure in advance, Turkey's attractiveness to international data center investors may increase.
But the same critical question still applies:
Will these regions be campuses hosting only the servers of foreign hyperscalers?
Or will a real Turkish AI ecosystem form around it?
The measure of success should be the second option.
Why Are Ethics Committee and Artificial Intelligence Governance Necessary?
Institutional structuring also has an important place in the "Manage" section of the AI plan.
In the governance model announced after the plan, the National Artificial Intelligence Board, Program Office and ethics/security mechanisms stand out. The National Board is planned to be designed as a high-level coordination body regarding strategic priorities, resource allocation, risks and ethical principles.
This structure is necessary not only for bureaucratic coordination, but also for the trust problem.
Because as AI enters public services, it will approach much more sensitive decision areas.
Risk assessment of a patient.
A credit evaluation.
A social assistance system.
A public personnel evaluation.
A security application.
In all of these, the "algorithm said so" explanation will not be enough.
Transparency, auditability, discrimination risks and how to establish objection mechanisms are important.
So successful AI policy must build trust, not just accelerate innovation.
What is Turkey's Advantage?
The USA is strong in giant model companies.
China is strong in hardware manufacturing, robotics and scale.
Europe is assertive in terms of regulation and industrial technology.
It may not be realistic for Turkey to win the same race with exactly the same method.
However, Turkey has some unique advantages.
Large young population.
Advanced defense industry.
Strong production infrastructure.
Geographical location between Europe, the Middle East and Asia.
Automotive.
Textile.
Tourism.
Logistics.
Finance.
E-commerce.
Large scale of public services.
Therefore, Turkey's strong AI strategy should be considered broader than the goal of “let the next OpenAI come from us”.
Perhaps the real advantage:
Being one of the countries that integrates AI into the real economy fastest.
Should Turkey Develop Its Own Basic Model?
This question is inevitable.
It is known that Türkiye is working on local large language models, and models such as BİLGE have begun to be opened to developers. The Ministry of Industry and Technology also positions these studies as part of the broader artificial intelligence ecosystem.
For Turkey, models with high Turkish language skills and understanding of local law, culture and public needs can be strategic.
But trying to produce the world's largest frontier model from scratch is very expensive.
Therefore, the more rational approach may be three-fold:
Taking advantage of global open models.
Developing models for Turkish and regional needs.
Producing some of the world's best custom models in specific industries.
This approach both reduces costs and brings economic value directly to the area of use.
Where are Advertising, Design and Creative Sectors in this Plan?
Although at first glance the plan seems to focus on data centers, robotics and industry, the creative economy will also be directly affected by the transformation.
Advertising agencies even today with AI:
text,
visual,
video,
sound,
3D,
research,
coding,
analysis
It can produce.
If cost-effective computing infrastructure and local models develop in Turkey, new companies may emerge, especially in Turkish content production.
Specific to Turkey:
AI advertising platforms,
Turkish voice models,
video models,
e-commerce creative automations,
social media tools,
native customer service AIs
Products such as this may pave the way.
This is also important for creative agencies.
Because in the future, agencies' competition will not only be based on using Photoshop well.
Who better combines AI infrastructure with creative strategy?
The question will come to the fore.
Perhaps the Biggest Change for SMEs is Beginning
A significant part of the Turkish economy consists of SMEs.
Therefore, it is not enough for the AI transformation to occur only in large conglomerates.
This is why it is important to include SME coupons, calculation access and sectoral programs in the plan.
It may not be realistic for an SME to establish its own AI department.
But with ready-made services:
demand forecast,
stock optimization,
customer service,
quality control,
sales analysis,
content production,
price optimization
It is possible.
If Turkey is to truly achieve great economic efficiency from AI, millions of small businesses need to be able to use this technology.
The Biggest Risk to the Plan: Failure to Turn Goals into Implementation
It is relatively easy to set strong goals in strategy documents.
The real hard part is the application.
In order to consider the plan successful when we come to 2030, we only need to:
“program started”,
“center established”,
“fund created”
Saying shouldn't be enough.
Tougher questions are needed.
How many AI companies generate global revenue?
How much has Turkey's GPU capacity increased?
What are the foreign sales of AI startups?
How much money have public AI projects saved?
What is the productivity increase in SMEs using AI?
How many of the trained experts can actually do advanced AI engineering?
How much of 1 GW data center capacity supports native AI products?
How many of the funded companies scaled?
If these are not measured, large numbers can easily turn into good-looking presentations.
The Other Big Risk: Brain Drain
It is ambitious to train 10 thousand advanced AI experts.
However, it may be more difficult for these people to stay in Turkey.
AI engineers are among the most in-demand workers in the world today.
The USA, Europe, Gulf countries and China offer high salaries and large research budgets.
Therefore, human resource strategy:
“training experts”
It shouldn't end with
.
Research budget.
International salary competition.
Stock options.
Startup ecosystem.
Academic freedom.
GPU access.
Global projects.
All of these affect the retention of talent in the country.
Training the best AI researchers and then sending them to another country's technology company will only slightly increase Turkey's competitiveness.
$10 Billion Is a Big Number, But the Global AI Race is Much Bigger
Turkey's 10 billion dollar investment target needs to be evaluated in the right perspective.
This is an important size for Turkey.
But it's not huge compared to single-year AI infrastructure spending by global tech companies.
Therefore, Turkey's competitive strategy cannot be to try to catch up with the USA in the race for capital.
He has to specialize smarter.
Turkey's opportunity is not to be the greatest in every field.
Being very good in certain areas.
Defense AI.
Industrial AI.
Turkish NLP.
Robotics.
Fintech.
Retail.
Tourism technologies.
Agriculture.
Logistics.
A more realistic strategy may be to produce products that can be sold to the world market in these sectors.
The Real Issue is Artificial Intelligence Sovereignty
It is also incomplete to evaluate the Turkey Artificial Intelligence Action Plan only as an economic development document.
In the framework of the plan published in the Official Gazette, the concepts of technological independence, data-driven development, reliable artificial intelligence and sustainable digital transformation stand out.
The reason for this is becoming increasingly clear.
A country:
Stores its data in a foreign cloud,
It gets its AI model from a foreign company,
Imports its GPU,
It uses the operating system from a foreign company,
If the AI API cannot provide service when interrupted
There is serious dependency in the digital economy.
Full independence may not be realistic in today's technological world.
But it is possible to create alternative capacity in critical layers.
This is what the “digital sovereignty” debate is about.
How Will We Understand Success in 2030?
The success story of Turkey's Artificial Intelligence Action Plan should not be the sentence "5 million people were trained" in 2030.
Real success should look more tangible.
If global AI companies emerge from Turkey,
If domestic companies export AI,
If the productivity of SMEs has increased,
If Turkish AI models have created a strong ecosystem,
If researchers want to stay in Turkey,
If public services have really improved,
If new export areas have emerged in robotic and autonomous systems,
If data centers feed domestic technology production rather than just hosting foreign services,
It means that the plan has reached its goal.
Otherwise, it may turn into another technology program with huge numbers but limited economic output.
The Real Race for Turkey Begins Now
2026-2030 Artificial Intelligence Action Plan is important for Turkey because, for the first time, it introduces a comprehensive framework that treats artificial intelligence as a technology infrastructure on a national scale, rather than a single software trend.
The education target of 5 million people is important.
10 billion dollars is important.
25 billion TL fund is important.
The 1 GW data center target is important.
“GPU for Everyone” is important.
But none of these are successes on their own.
These are tools.
Success lies in what Turkey will produce with these tools.
Because in the age of artificial intelligence, economic value is shifting towards those who build products on the model, export the product, transform data into value and have the critical infrastructure of technology, rather than those who use the model.
The choice before Turkey is actually quite clear.
We can be a large market that uses the artificial intelligence tools developed by the world very well.
Or we can turn into a technology producer that creates its own products, companies, models and intellectual property in addition to using them.
The real importance of the new action plan is not in the 10 billion dollar investment target.
The second option has now become an official state target.
Nur Oğuz