Nvidia is committed to launching new AI chips every year

Nvidia just made a profit of $14 billion in just one quarter thanks to AI chips, and this growth momentum will continue. According to CEO Jensen Huang, Nvidia will design new chips every year instead of every two years like before.

During the Q1 2025 earnings call , Mr. Huang stated: “I can announce that after Blackwell, there will be another chip. We are moving to a one-year release cycle.”

Previously, Nvidia usually launched new chip architectures about every two years, for example Ampere in 2020, Hopper in 2022 and Blackwell in 2024. (The famous AI chip H100 belongs to Hopper architecture, and B200 belongs to Blackwell, these architectures are also used for gaming and design GPUs).

However, analyst Ming-Chi Kuo earlier this month reported that the next “Rubin” architecture will launch in 2025, bringing the R100 AI GPU next year at the earliest. Mr. Huang’s comments suggest this report may be accurate.

Huang said that Nvidia will also accelerate production of other chips to match this pace. “We will develop all types of chips at a very fast pace.”

Talking about upcoming products, this leader said: “There will be many new chips hitting the market in the near future, including new CPUs, new GPUs, new network cards and new switches.”

Earlier in the call, when an analyst asked him to explain the sale of new Blackwell GPUs while Hopper GPUs were still selling well, Huang explained that Nvidia’s new generations of AI GPUs are backward compatible. electrical and mechanical, and run with software. Customers can “easily migrate from H100 to H200 to B100” in existing data centers.

Huang also shared a few reasons for the incredible demand for Nvidia’s AI GPUs: “We predict demand will exceed supply for a while, when we move to H200, when we move to Blackwell. Everyone is looking forward to it.” want to get their infrastructure online as soon as possible. The reason for that is because they’re saving money and making money and they want to do it as soon as possible.”

He also gave an interesting reason that plays on the “fear of missing out” (FOMO) mentality: “The next company to achieve a major breakthrough will be notified of a breakthrough AI, and the second company after that.” will announce something that’s 0.3% better. Do you want to be the company that delivers breakthrough AI, or the company that delivers something that’s 0.3% better?”

Interestingly, Nvidia’s CFO said the automotive segment will be “the largest enterprise vertical in the data center this year,” pointing to Tesla’s purchase of 35,000 H100 GPUs to train its “autonomous driving” system. completely”, while “consumer internet companies” like Meta will remain a “strong growth vertical”.

Some customers have even purchased or plan to purchase more than 100,000 Nvidia H100 GPUs – Meta plans to operate more than 350,000 of these GPUs by the end of the year.