Nvidia's Tesla V100 Gets $200 Price Premium as AI Boom Revives Demand

With the RTX 5090 gone from Nvidia's marketplace and the RTX 6000 PRO too expensive, the 9-year-old Tesla V100 with 32GB of HBM2 is staging a comeback for local LLM inference. PCIe and SXM2 versions now carry a $200 premium, with full rigs costing around $2,000.

• Tesla V100 with 32GB HBM2 sells in PCIe and SXM2 versions at a $200 premium
• 900GB/s memory bandwidth enables LLM rigs for around $2,000
• RTX 6000 PRO costs over $16,000, while the RTX 5090 vanished from Nvidia's store
• Four V100s on an NVLink board deliver up to 230% gains in Qwen3.8-Flash-Next tests