The AI boom has made new GPUs scarce and expensive, but it has also created a seller's market for used cards. According to Engadget, even well-worn graphics cards can fetch significant sums, with high-VRAM models retaining 40-60% of their original value. Older cards like the NVIDIA GeForce RTX 3090 are especially valuable because they pack more VRAM than some newer models.
Demand comes from both large LLM companies and individuals building local AI setups. The article notes that NVIDIA's RTX 50-series prices have climbed as much as 39%, with AMD cards up to 10% higher. Most AI-focused buyers look for 12-16GB of VRAM, while larger teams seek models with as much as 48GB, keeping used high-capacity cards in demand.
Where you sell matters. Peer-to-peer platforms like eBay and Facebook Marketplace require more effort but can yield prices above original MSRP, as seen with recent listings for the RTX 5060 Ti and EVGA GeForce RTX 3080. Trade-ins through Newegg or local repair shops are more convenient but pay less. For a card in good condition, selling it yourself is worth the hassle; for a worn model, a quick trade-in is a reasonable option.