Nvidia is preparing to raise prices on AI servers: an increase of more than 15% already in 2027
The AI infrastructure market is entering a new phase of pricing pressure. The largest manufacturer of graphics processors and server solutions for artificial intelligence has notified key partners of an upcoming revision of the cost of its systems. In a number of configurations, price increases will exceed 15% — this is a noticeable signal for the entire ecosystem, from cloud providers to corporate customers.
The adjustment will affect shipments scheduled for early 2027. Particular attention is drawn to systems based on new architectures — Vera Rubin and Grace Blackwell. These platforms represent the next generation of computing clusters focused on training and inference of large language models, and their cost directly impacts the cost of AI products worldwide.
The price increase is not targeted — it is systemic. In conditions where demand for AI computing power continues to grow exponentially and production capacity is limited, the manufacturer is revising its pricing policy. This is a logical step amid a shortage of advanced chips and rising costs for lithography, packaging, and cooling of high-performance systems.
For the market, this means that the cost of renting cloud GPU clusters could increase in the coming quarters. Companies planning large-scale AI initiatives should factor this into their budgets. In the long term, this could spur the development of alternative solutions — from specialized ASIC chips to optimizing existing models for less resource-intensive architectures.
My view: this is not a one-time event, but the beginning of a long-term trend. As long as demand for AI computing remains at its peak and competition in the accelerator segment is limited, the manufacturer will dictate terms. Investors and data center operators should factor in annual equipment cost growth of 10–15% through the end of the decade.