
NVIDIA has recently refuted claims suggesting that the sales of its H20 GPUs have negatively affected the availability of the H200 and H100 AI graphics processing units (GPUs).
NVIDIA Assures Adequate Supply of H100 and H200 AI GPUs Amid H20 Sales
Unfounded reports have circulated alleging that NVIDIA is prioritizing the sales of its H20 GPUs, thereby limiting the supply of the H100 and H200 AI models. In response, NVIDIA wasted no time in addressing these claims, issuing an official statement to clarify the situation.
NVIDIA labeled the circulating rumors as the result of “misleading information.”The company emphasized that their cloud partners are fully capable of managing all existing H100 and H200 units available online, and they can also fulfill new orders promptly.
We’ve seen erroneous chatter in the media claiming that NVIDIA is supply constrained and “sold out”of H100/H200. As we noted at earnings, our cloud partners can rent every H100/H200 they have online — but that doesn’t mean we’re unable to fulfill new orders. We have more than…
— NVIDIA Newsroom (@nvidianewsroom) September 2, 2025
The H20 chip from NVIDIA was specifically developed to comply with U. S.export regulations, allowing it to be exported to China amidst ongoing trade restrictions. Despite being a simplified version, the H20 still surpasses the capabilities of certain Chinese alternatives, such as Huawei’s Ascend 920 series.
Moreover, last month, there were allegations concerning potential security risks, including backdoors and spyware within NVIDIA’s chips intended for the Chinese market. These accusations were also firmly dismissed by NVIDIA through a blog post.
Given NVIDIA’s soaring presence in the data center and AI sectors, particularly with its innovative Hopper and Blackwell architectures, it appears that the rumors attempting to undermine its position stem from competitive tensions. The recent financial results highlight NVIDIA’s strength, reporting a remarkable $41.096 billion in revenue from its AI segment.

NVIDIA is confident enough in its supply chain and partnerships to encourage potential buyers to approach them. The company is now adapting to an annual production cycle aimed at delivering cutting-edge AI GPUs: with the release of Rubin set for 2026, followed by Rubin Ultra in 2027 and Feynman in 2028.
Leave a Reply