The graphics card market is facing a shortage amid the growth of AI

The global graphics card market continues to experience shortages and rising prices, due to the redistribution of resources in favor of the artificial intelligence industry. One of the key events was Nvidia's announcement that the company would not release any new graphics cards in 2026, for the first time in three decades. This was reported on March 18 by RBC.

The manufacturer may also reduce the production of individual series, including the GeForce RTX 50, but there is no official confirmation of this information yet. Against the background of limited supply, manufacturers have already begun to raise prices even for mass-produced models, while the shortage persists. In a number of countries, retailers are imposing restrictions on the sale of video cards to curb excessive demand.

Experts attribute the situation to the rapid development of artificial intelligence, which requires significant computing power. Most of the resources of the largest developers are directed to the production of chips for data centers and AI systems. As a result, the traditional segments — gamers, designers, and other users — face hardware shortages.

At the same time, the shortage affects not only video cards, but also other components, including RAM and flash memory, as well as storage devices. The growth of AI-generated data increases the burden on infrastructure and requires additional resources for storing and processing information.

Screwdriver set: users began to assemble memory cards on their own

Is it worth saving on branded components and why is it dangerous?

"In addition, energy and cooling are becoming real constraints: modern accelerators require high power density, which means expensive power supply and heat dissipation systems. The second barrier is data and its delivery: the training and application of the accumulated knowledge by the neural network rest on network bandwidth, speed of access to storage, cost and reliability of data processing pipelines. The third is people and the operational costs of implementation: machine learning operations (MLOps), security, compliance, quality control, and model support often turn out to be no less complex and expensive than buying hardware," the publication says.

Despite the emergence of new players, including from China, the market still remains centered around several large manufacturers. New companies are not yet able to fully compensate for the deficit or significantly affect price dynamics, but they are creating additional competition and can strengthen their positions in certain local markets.

Experts note that the further development of AI is limited not only by the availability of components, but also by other factors, including energy consumption, cooling systems, data transfer rate and transaction costs. Under these conditions, the industry is considering several areas of development, including optimizing the architecture of computing platforms, creating specialized chips and improving the models of artificial intelligence themselves.

It is expected that the combination of these approaches will partially reduce the burden on the market and ensure the further development of technologies in conditions of continuing scarcity.

On December 24, 2025, Galen Hunt, the company's lead engineer, announced Microsoft's plans to translate the code of software products into the Rust language. It was noted that the task will mainly be performed by artificial intelligence (AI) agents.

All important news is on the Izvestia channel in the MAX messenger.