NVIDIA’s CEO Jensen Huang Visits Taiwan As Mass Production Hopper H200 & Blackwell B100 GPUs Nears

NVIDIA's CEO Jensen Huang has been making "crucial" trips in recent days, as he paid a visit to Taiwan, to "streamline" supply for next-gen Hopper H200 and Blackwell B100 GPUs.

It seems like Team Green is gearing up for another progressive year since NVIDIA not only has been making rapid advancements within its AI portfolio but the firm striving towards catering to every region of interest, despite setbacks such as the new US regulations. We reported yesterday about the visit of Jensen Huang to Beijing, where he not only took part in festivities associated with Beijing New Year but reportedly held meetings with several Chinese clients, such as Alibaba and Tencent, to assure them that NVIDIA is here to stay.

It is now being disclosed that apparently after the visit to China, Jensen Huang also decided to go to Taiwan, where he visited the firm's key suppliers such as TSMC and Wistron, to see the progress on mass manufacturing of the company's next-gen H200 and B100 AI GPUs. This year will mark the transition towards a newer generation of AI products, which is why 2024 is seen as "decisive" not only for just NVIDIA, but other firms involved in the AI race as well. It is predicted that global AI shipments are expected to see a tremendous rise this year, with estimated figures reaching around 350,000 chips, involving a hefty share of Team Green.

For a quick recap on what to expect in the upcoming products by NVIDIA, the first is their Hopper H200 GPU, which is equipped with Micron's HBM3e solution with memory capacities of up to 141 GB and up to 4.8 TB/s of bandwidth which is 2.4x more bandwidth and double the capacity versus the NVIDIA A100.

This new memory solution allows NVIDIA to nearly double the AI inference performance versus its H100 GPUs in applications such as Llama 2 (70 Billion parameter LLM). The next big release is their Blackwell B100 AI GPU, and in terms of performance, it will bring decisive uplifts in terms of performance per watt, through the adoption of a chiplet design.

NVIDIA's next-generation Hopper H200 and Blackwell B100 AI GPUs are expected to debut within 2024, with B100 reportedly being pushed back till Q4 as Hopper secures a higher adoption rate in 2024. This year is going to be interesting for the tech industry as a whole, especially with the advancements made in artificial intelligence, and the associated markets with it. Expect NVIDIA to unveil more information about its AI and Data center chips at GTC 2024 in March.