GPU technology and AI inference: CoreWeave rebuttal to GPU depreciation claims

CoreWeave CEO Michael Intrator says “GPU technology” remains strategically valuable beyond crypto, arguing that the market’s “GPU depreciation” narrative is exaggerated. He links CoreWeave’s shift from crypto-related workloads to CGI rendering, batch computing, and medical research to show GPU technology’s adaptability to new demand. Intrator frames AI economics around AI inference: monetization comes when compute powers large-scale inference running daily. He also claims CoreWeave is key to deploying Nvidia’s latest architecture at commercial scale, supporting faster deployment of next-gen AI infrastructure. On GPU lifespan, Intrator calls claims that GPUs become commercially obsolete after 16–24 months “farcical.” He says customers typically buy compute for 5–6 years (his view is tied to contract structure rather than short-term depreciation arguments). He also suggests the “GPU depreciation” debate is pushed by traders with short positions. Finally, he highlights that competition is increasing as more firms try to supply AI infrastructure, and that specialized financing structures (a cash-flow “box” model) are used to manage compute contract cash flow. For traders, the message is about AI infrastructure durability and financing visibility rather than immediate crypto fundamentals—potentially supportive for AI/compute sentiment but not a direct catalyst for major crypto price moves.