近期，全球AI芯片供应紧张的状况出现缓解，有消息称，部分公司开始转售囤积的英伟达H100 GPU。据IT之家报道，这款专为人工智能（AI）和高性能计算（HPC）应用设计的英伟达H100 80GB GPU的交货时间已从原先的8至11个月大幅缩短至3至4个月，这一变化显著提升了市场的流动性。
这一供应改善的背后，反映出芯片制造业的产能提升以及市场需求的调整。由于交货周期缩短，一些原本囤积H100 GPU以待价而沽的公司，现在发现从亚马逊云服务、谷歌云和微软Azure等大型云服务提供商那里租用芯片更加灵活和经济，因此开始转售手中的过剩库存。
这一现象不仅有助于平衡市场供需，也使得对AI和HPC需求强烈的公司能够更快地获取所需的硬件资源，进一步推动了相关领域技术的发展和应用。随着供应链的逐步稳定，预计AI和HPC行业的项目实施将更加顺利，有助于全球科技创新的加速。
然而，这一市场变化也提醒业界，随着云服务的日益普及，传统的硬件采购模式可能需要进行调整，以适应更加动态的市场需求和更高效的资源配置。同时，对于芯片制造商而言，如何在满足即时需求的同时，预防未来可能出现的供需波动，将成为一个需要持续关注和解决的问题。
英语如下：
**News Title:** “NVIDIA H100 GPU Supplies Improve, Sparking Resale Surge”
**Keywords:** AI chip relief, NVIDIA H100, GPU resale
**News Content:**
**Title:** AI Chip Market Sees Improved Supply, NVIDIA H100 GPU Delivery Times Significantly Shorten
Recently, the global shortage of AI chips has shown signs of easing, with reports indicating that some companies are now reselling their stockpiled NVIDIA H100 GPUs. According to IT Home, the delivery time for NVIDIA’s H100 80GB GPU, designed for artificial intelligence (AI) and high-performance computing (HPC) applications, has dropped from an initial 8 to 11 months to just 3 to 4 months, significantly boosting market liquidity.
This improvement in supply reflects increased manufacturing capacity in the semiconductor industry and adjustments in market demand. With shorter lead times, companies that had hoarded H100 GPUs for speculative resale are now finding it more flexible and cost-effective to rent chips from major cloud service providers like Amazon Web Services, Google Cloud, and Microsoft Azure, leading them to offload their surplus inventory.
This development not only helps balance the market supply and demand but also enables companies with pressing AI and HPC requirements to access the necessary hardware resources more quickly, thus fueling the advancement and application of related technologies. As supply chains stabilize, it is expected that AI and HPC projects will proceed more smoothly, contributing to a faster pace of global technological innovation.
However, this market shift also underscores the need for the industry to reconsider traditional hardware procurement models in light of the growing popularity of cloud services, to adapt to more dynamic market demands and more efficient resource allocation. For chip manufacturers, the challenge will be to balance meeting immediate needs while mitigating potential future supply and demand fluctuations.
【来源】https://www.ithome.com/0/752/328.htm
Views: 1