Nvidia denies H100 and H200 GPU shortage, says supply remains strong

Nvidia recently dismissed reports of H100 and H200 GPU shortages created to prioritize H20 GPU orders, calling them "erroneous chatter". Nvidia claims that cloud partners can now run all H100 and H200 units online, and it can still fulfil new orders without delay. This clarification was made in response to recent market speculation about a constrained supply.

Earlier commentary made by the company during their latest earnings call fueled "sold out" speculation for the H100 and H200 GPUs, which some took as confirmation of scarcity. This directly contradicts Nvidia's current statement of sufficient supply, with the contradiction most probably caused by lack of context.

The H20, built with U.S. export controls in mind, saw strong demand when Washington recently gave the green light for the company to start selling the GPU again. Nvidia, however, claims that increased H20 demand does not impact H100 or H200 production.

Recent allegations of backdoors and "kill switch" from Beijing were denied by the company and remain unproven. There are rumors of a new China-specific GPU in development, code-named the B30A, but specific details remain under wraps for now.

Nvidia's AI segment reported a record $41.096 billion in revenue, showcasing high demand. Even so, the company claims that it can still meet buyer demand and even invites direct outreach from customers. Nvidia plans to release annual updates for its future high-end AI GPUs, with Rubin reportedly set to arrive in 2026, Rubin Ultra in 2027, and Feynman in 2028. Plans are underway to produce the new Rubin chips in Taiwan, although details remain unconfirmed.