The demand for artificial intelligence infrastructure is outpacing the availability of GPU servers. According to Gigant Computer Systems, companies are transitioning from pilot projects to industrial AI deployment, which may lead to a shortage of accelerators, price hikes, and longer delivery times in 2027.
The average monthly cost of dedicated GPU-accelerated infrastructure for AI and machine learning tasks increased by 64% year-over-year during the first five months of 2026. Costs reached 2.3 ₽ million, compared to approximately 1.4 ₽ million in the previous year.
AI and ML tasks currently account for 33% of all GPU server orders. The remaining demand is driven by rendering, software development, and data processing. Businesses are increasingly requiring specific hardware configurations featuring high-capacity video memory and high-bandwidth interconnects.
The shift toward industrial AI applications in sectors such as banking, retail, telecommunications, and manufacturing has increased the need for stable, high-performance computing. Companies are moving away from cloud-based pilot environments toward dedicated infrastructure to handle sensitive data and complex engineering tasks.
CEO Sergey Semikin warns that project delays are risky due to rapidly changing prices and supply chain constraints. He advises companies to prioritize early infrastructure planning, including power, cooling, and scalability requirements, to avoid bottlenecks in their AI projects.
