据IT之家消息，近期AI芯片市场的供需状况出现显著改善，尤其是英伟达的H100 GPU，其供应瓶颈正逐步解除。此前，这款专为人工智能和高性能计算应用设计的旗舰级芯片，交货周期长达8至11个月，给相关行业带来了不小的压力。然而，现在这一情况已大为改观，H100 GPU的交货周期缩短至3至4个月，为用户提供了更为及时的获取途径。
这一变化导致部分先前囤积了H100 80GB处理器的公司开始调整策略，考虑转售其过剩库存。随着亚马逊云服务、谷歌云和微软Azure等大型云服务提供商对AI芯片租赁服务的加强，企业现在更倾向于通过云平台灵活租用芯片，而非大量囤货。这种趋势反映出市场对云计算资源的依赖度提升，以及对资源配置效率的重视。
这一动态变化不仅有助于平衡芯片市场供需，还可能影响相关企业的经营策略，推动AI和HPC领域的创新更加依赖于云端服务。随着英伟达H100 GPU供应的改善，预计未来一段时间内，AI和HPC应用的开发与部署将更为顺畅，进一步推动相关行业的发展。
英语如下：
**News Title:** “NVIDIA H100 GPU Supply Tightness Eases, Fueling Resale Boom”
**Keywords:** AI chip supply, NVIDIA H100, GPU delivery timelines
**News Content:**
Title: AI Chip Supply Crunch Eases, NVIDIA H100 GPU Market Sees Shift
According to IT Home, recent developments indicate a noticeable improvement in the supply-demand situation for AI chips, particularly with NVIDIA’s H100 GPU. The bottleneck in its supply chain is gradually being alleviated. Previously, this flagship chip, designed for artificial intelligence and high-performance computing applications, had an extended delivery timeline of 8 to 11 months, posing significant challenges to the industry. However, the current situation has changed, with the H100 GPU’s delivery周期 now shortened to 3 to 4 months, offering users a more timely access route.
This development has prompted some companies that stockpiled the 80GB H100 processors to reconsider their strategies and consider reselling their surplus inventory. As major cloud service providers like Amazon Web Services, Google Cloud, and Microsoft Azure enhance their AI chip leasing services, enterprises are now more inclined to lease chips flexibly through cloud platforms rather than amassing large inventories. This trend reflects the increasing reliance on cloud resources and the emphasis on resource allocation efficiency.
This dynamic shift not only contributes to balancing the chip market供需 but may also influence corporate strategies, pushing AI and HPC innovations to rely more on cloud services. With the improved supply of NVIDIA H100 GPUs, it is anticipated that AI and HPC application development and deployment will become smoother in the coming period, further propelling the growth of related industries.
【来源】https://www.ithome.com/0/752/328.htm
Views: 2