Nvidia 称消费级 GPU 短缺非其所能控制
Nvidia Says Consumer GPU Shortages Are Beyond Its Control < Semiconductor < 기사본문 - The Elec Inc.
Nvidia 开发者关系副总裁 Ida Keita 表示，消费级 GPU 短缺源于显存供应不足，涉及 DRAM 等更广泛的半导体供应链问题，并非 Nvidia 能控制。旗舰消费级 GPU RTX 5090 售价高达 1500 万韩元，约为含税建议零售价的五倍。同日发布的 RTX Spark 笔记本起售价约 470 万韩元，可本地运行超 1000 亿参数 AI 模型，无需云端服务因而没有 token 使用成本。
正文
当前显示原文，中文全文译文尚未就绪。
가
The body content of the article will be changed to this letter size.
스크롤 이동 상태바
Ida Keita, vice president of developer relations at Nvidia. (Photo: THE ELEC)
An Nvidia executive said shortages of consumer graphics processing units (GPUs) stem from insufficient memory supplies.
Ida Keita, Nvidia’s vice president of developer relations, told THE ELEC after an RTX Spark launch event at COEX in Seoul’s Samseong-dong district on Oct. 16 that supply difficulties facing consumers, including gamers, reflect broader semiconductor supply chain problems involving not only GPUs but also DRAM.
“This is not something Nvidia can control,” Keita said in response to a question about consumer GPU shortages.
Popular consumer graphics cards are available in limited quantities and are often sold through lotteries or on a first-come, first-served basis. Nvidia’s flagship consumer GPU, the RTX 5090, is trading for as much as 15 million won or more, roughly five times its tax-inclusive suggested retail price.
Citing the RTX Spark launched that day, Keita said, “Considering the workloads RTX Spark can handle, it can save users a tremendous amount of money over the long term compared with buying other computers.”
Keita added that the device can run artificial intelligence (AI) models with more than 100 billion parameters locally. “Because it doesn’t require external cloud services, there are no token usage costs,” Keita said, emphasizing that the product offers value despite its high price.
RTX Spark-based laptops start at approximately 4.7 million won, with prices reaching as much as 10 million won depending on specifications. The company attributed the high prices to rising memory costs.
“When we were developing the product, 128GB was reasonably affordable, and we considered it the optimal configuration,” an Nvidia official said at the event. “We didn’t anticipate memory prices rising this much.”
JEONG IL JOO 1week@thelec.kr
My comment