TRT World Digital Team Manager Abdulvehab Ejupi wrote about the importance of the recently launched national artificial intelligence platform EVREN for AA Analytics.
We often read the artificial intelligence race as a struggle between the United States (US) and China. The US has the world's leading artificial intelligence companies, advanced chip design capacity and huge capital markets. China, on the other hand, constitutes the other major pole of the race with its large domestic market, production capacity and state-supported technology investments.
However, right behind these two superpowers, another, quieter but increasingly important rivalry is taking place. It is not easy for countries such as Turkey, France, Germany, Spain, Saudi Arabia and Qatar to compete directly with the USA or China in the high-end model race worth hundreds of billions of dollars. But in the age of artificial intelligence, the power is not just in developing the biggest model. Energy, data centers, computing capacity, data, human resources and the application of artificial intelligence to the real economy are increasingly turning into strategic power elements. It is necessary to evaluate Turkey's recently launched national artificial intelligence platform, EVREN, within this broader transformation.
Why is the UNIVERSE important?
It is not correct to define EVREN as "Turkey's ChatGPT". The platform developed within the Presidency of Defense Industries combines different stages of the artificial intelligence development process, from data preparation and labeling to model training, from computer vision applications to running large language models, in a common infrastructure.
Today, the platform provides access to 11 major vulnerability-heavy language models. Users can earn credits by uploading datasets, labeling data, or sharing their trained models, and use these credits for model training and inference on the high-performance Graphics Processing Unit (GPU) infrastructure.
At the center of this infrastructure are 64 Nvidia H200 GPUs, each with 141 GB HBM3e memory. In total, the system offers approximately 9 terabytes of high-bandwidth GPU memory and 47.5 petaflops of FP16 Tensor Core computing capacity. The fact that the export of advanced artificial intelligence chips such as the H200 to some countries is subject to US licensing and export controls shows that this hardware has now become a strategic resource not only technologically but also geopolitically.
The critical concept here is computing power, that is, “compute”. EVREN has reached approximately 15 thousand active users in a short time. In the coming phase, it is planned to add local large language, image and sound models to the system and develop the distributed GPU infrastructure. This is especially important for universities, researchers and technology startups. Because today, one of the biggest obstacles to developing advanced artificial intelligence is no longer just knowledge or talent, but access to sufficient computing power. Therefore, EVREN is not just a software project, but an attempt to create a common artificial intelligence development infrastructure in Turkey.
Digital sovereignty in the age of artificial intelligence
Perhaps the more important aspect of EVREN is data sovereignty. Today, millions of people and companies use foreign artificial intelligence services and send their data to infrastructures in other countries. In EVREN, the aim is to process user requests, responses and usage data on the high-performance GPU infrastructure located in Turkey. This is an important distinction, especially in the defense, public and critical sectors. Because in the coming period, the question "Where is the data kept?" Another question will be added to the question: Where is the artificial intelligence process carried out and who controls this computing infrastructure? Artificial intelligence dominance is increasingly shaped by three basic elements: data, models and computing power.
It would be incomplete to evaluate EVREN independently of Turkey's new Artificial Intelligence Action Plan. Turkey Artificial Intelligence Action Plan (2026–2030), which was announced to the public in June and entered into force on August 18, 2026, reveals Turkey's road map for the next four years under the headings "Recognize, Use, Produce and Manage".
The targets are ambitious: at least 1 GW of artificial intelligence and data center capacity by 2030, at least $10 billion of investment mainly from the private sector, 10 thousand advanced artificial intelligence experts and 100 thousand artificial intelligence application professionals. It is also planned that 5 million people will undergo artificial intelligence literacy programs within two years. The "GPU for All" program for researchers, startups and SMEs, the National Data Library, sectoral artificial intelligence models and the development of physical artificial intelligence and robotics capacity are also among the parts of the plan.
When all these are read together, EVREN's place in the bigger picture emerges: Türkiye is not only using artificial intelligence; It wants to be an economy that can manage its data, computing infrastructure and certain models.
Artificial intelligence difference in the energy race
There is another issue here that is often overlooked. Artificial intelligence is no longer just a software, but also a story of energy and infrastructure. In addition to GPUs, huge data centers require uninterrupted electricity, powerful grids, cooling systems and fiber connections. The success of Turkey's 1 GW data center target will largely depend on its ability to provide these.
Türkiye has been developing its role as an energy corridor between Europe and Asia for years with projects such as TANAP. When electricity transmission projects, renewable energy investments and the nuclear capacity planned to be activated with Akkuyu are added to this, a more interesting question arises: Can Turkey turn into an energy and computing corridor between Europe and Asia, beyond being an energy corridor? This is not yet a reality. However, if the 1 GW data center target can be realized, it may be possible to transfer Turkey's geographical advantage to digital infrastructure.
What does national artificial intelligence mean?
Turkey does not need to emulate the general-purpose model race at the scale of OpenAI, Google DeepMind or Anthropic, which requires hundreds of billions of dollars. A more realistic approach would be for Turkey to specialize in areas where it has a competitive advantage, such as defense and autonomous systems, industrial artificial intelligence and robotics, energy, logistics, agricultural technologies, and artificial intelligence models for Turkish and regional languages.
Artificial intelligence ecosystem of the future; Rather than a structure where each country establishes its own small Silicon Valley, it can turn into a more distributed system in which the Gulf's capital and energy, Turkey's engineering and production capacity, and Europe's research and industrial infrastructure complement each other. One of Turkey's advantages is that it is located at this geographical intersection.
In addition, EVREN's computing infrastructure includes Nvidia H200 GPUs, and a significant portion of the large language models on the platform consist of global open-weighted models. However, technological sovereignty does not mean producing every chip, model and software within the country. The real issue is how much strategic maneuvering space you have in critical technologies. Can you keep your data within the country? Can you run the models on your own infrastructure? Do your researchers have access to computing power? Can you develop models for your own language and sectors? It is more accurate to evaluate the importance of the UNIVERSE through these questions.
The real test starts now
Of course, building a platform is not the same as creating a globally competitive artificial intelligence ecosystem. Türkiye still faces serious challenges such as access to advanced chips, financing, energy and data center infrastructure, advanced research and retaining qualified human resources in the country. For this reason, EVREN's success should be measured after a few years by the results it produces rather than the number of users.
How many domestic models were developed? How many startups have launched products using EVREN's computing power? How many studies have been commercialized? How much of the technologies developed in defense were transferred to the civilian economy? And most importantly, has Türkiye transformed into a country that not only uses artificial intelligence but also exports artificial intelligence technology and products?
Rather than surpassing the USA or China in the artificial intelligence race, the aim is to become an important artificial intelligence production, calculation and application center at the intersection of Europe, the Gulf, the Balkans, the Caucasus and Central Asia. EVREN is not this goal itself, but if it is scaled correctly, used actively by the private sector and universities, and supported by energy, data center and human resources investments, it can be one of the important parts of the infrastructure towards this goal. Because in the age of artificial intelligence, sovereignty will no longer be measured only by controlling land, energy or physical infrastructure. The capacity to control data, computing power and artificial intelligence infrastructure will also determine the economic and strategic power of countries. This is where the real importance of the UNIVERSE lies.
