Microsoft has not slowed down procuring Nvidia's H100 compute GPUs despite rumors, according to the Chairman of Quanta, a leading maker of servers and a close partner of Nvidia, reports CTEE.com.tw. The cloud and software giant keeps purchasing and installing machines powered by Nvidia's H100 processors for artificial intelligence (AI) and high-performance computing (HPC) workloads.
Lin Baili, Quanta's Chairman, addressed rumors about Microsoft reducing its orders for the Nvidia H100 compute GPUs at the AI Technology and Application Outlook Forum. He refuted these rumors and expressed confidence in the future of AI, indicating that his company has consistently secured many large orders and remains optimistic about the subsequent development of the AI servers market.
Quanta's Chairman highlighted Taiwan's essential role in AI, driven by its robust tech supply chain. Beyond TSMC, Taiwan's strength lies in its comprehensive tech supply chain, spanning from chip design to semiconductor foundries and from processors to PCs and servers. Meanwhile, the latter are primarily assembled in China, which is worrisome to Quanta.
Market research company Omdia recently said that Nvidia shipped over 300 thousand H100s in the second quarter, and it expects Nvidia to sell roughly the same amount of GPUs in the coming quarters. As a result, the company will ship over 1.2 million GPUs this year alone, which translates into tens of billions of dollars in revenue.
Market rumors indicate that Nvidia has sold out its compute GPUs till the second quarter of 2024, so even if Microsoft decided to slow down purchases of compute GPUs in the second half of 2024, this will have no impact on the market or Nvidia for months to come.
