Nvidia Pushes TSMC to Boost 3nm Chip Production by 50% for Next-Gen AI Processors
The race for dominance in artificial intelligence (AI) has reached a crucial stage, with Nvidia leading the charge through its Blackwell Ultra accelerators.
Nvidia Secures Larger Share of TSMC’s 3nm Capacity
Nvidia CEO Jensen Huang recently visited Taiwan on a key mission — to secure a significant share of TSMC’s 3-nanometer (3nm) chip production capacity. The main goal is to ensure an adequate supply for the company’s next-generation AI chips, known as the Vera Rubin series.
According to reports, TSMC is planning a major expansion of its production capabilities. The company will increase capacity at its Southern Taiwan Science Park facility from 100,000 wafers per month to around 160,000 wafers per month, representing a growth of roughly 50%. Notably, a large portion of this additional capacity will be allocated exclusively to Nvidia. Nvidia firmly believes that demand for its Rubin chips will be massive, prompting it to secure supply well in advance.
Vera Rubin
The Vera Rubin line is expected to mark a critical milestone for Nvidia, promising a significant leap in computing performance. These chips will not only use TSMC’s advanced N3P process but also feature HBM4 high-bandwidth memory.
Anticipation is already high — Nvidia has secured several major clients for the Rubin chips even though mass production has yet to begin, with roughly six months still to go. For TSMC, this growing demand from high-performance computing (HPC) clients — led by Nvidia — will make its 3nm process a key revenue driver for upcoming quarters.
