
As NVIDIA gradually strengthens its AI servers based on the B200 and GB200 platforms, paving the way for its next-generation GB300 AI server product line, the US AI giant is reportedly making another cautious move. Masu. According to a report in the Commercial Times, NVIDIA plans to create an ASIC (application-specific integrated circuit) division to develop customized chips.
Additionally, the report notes that NVIDIA has selected Taiwan as the base for its ASIC R&D center and is actively recruiting talent from major local IC design firms, raising concerns about a potential brain drain from Taiwan’s semiconductor sector. It is pointed out that it is increasing.
The Commercial Times, citing sources at Taiwanese IC companies, reported that there will be a wave of talent poaching in mid-2024. With NVIDIA reportedly joining the game, fierce competition for talent could be just around the corner, with major IC design firms such as MediaTek, Alchip Technologies, and TSMC affiliate GUC They are being urged to prepare for the upcoming battle, the source added.
Notably, the tech giant is actively looking for replacements for NVIDIA’s GPUs. As previously reported by CNBC, Apple revealed in July 2024 that AI models supporting Apple Intelligence are pre-trained on Google’s Tensor Processing Units (TPUs).
In late December 2024, Apple executive Benoit Dupin also mentioned the company’s use of Amazon’s Trainium2 chip, indicating that non-NVIDIA training solutions can also be effective, the Commercial Times noted.
So NVIDIA is responding aggressively by creating its own ASIC division and expanding its custom services capabilities, the Commercial Times reports. At the same time, the company plans to hire more than 1,000 experts in areas such as chip design, software development, and AI research and development in Taiwan, the report said.
Notably, Taiwanese ASIC companies play key roles in the development of Microsoft’s Cobalt and Maia, Google’s TPUs, and AWS’s custom chips, according to the Commercial Times, making them attractive targets for NVIDIA’s ASIC business. Worth it.
read more
(Photo courtesy of NVIDIA)
This article quotes the following information: commercial times and CNBC.
Previous article
(News) Semiconductor foundry competition intensifies: Japan and South Korea strengthen
Next article
(News) Samsung accelerates humanoid development to offset business headwinds, challenges tech giant