The GeForce RTX 5090, originally marketed as a gaming graphics card, is facing severe shortages as AI companies purchase the units in bulk. While the recommended retail price was $2,000, current market prices have surged to $5,000 in the United States and over €5,200 in Europe.
The trend of using consumer-grade GPUs for AI infrastructure is not new. As early as April 2025, reports emerged of a Vietnamese PC builder integrating seven RTX 5090 units into proprietary AI server systems. These setups often utilize standard models rather than the restricted RTX 5090 D version.
The standard GeForce RTX 5090 offers an AI performance of 3,352 TOPS, significantly higher than the 2,375 TOPS provided by the RTX 5090 D. Nvidia has not implemented strict controls on bulk sales through its partners and distributors, despite the ability to track large shipments via serial numbers.
Industry observers note that Nvidia continues to prioritize the promotion of its hardware for AI applications. Consequently, the GeForce RTX 5090 is increasingly viewed as an enterprise-grade tool for AI tasks rather than a dedicated gaming component.
