Nvidia Tells Suppliers to Tap TSMC Process for Next-Generation Memory Chips

TSMC's 3nm process is now locked in for Nvidia's next generation of HBM chips.

Nvidia has instructed its suppliers for upcoming high-bandwidth memory (HBM) chips to utilize Taiwan Semiconductor Manufacturing Co. (TSMC)'s 3-nanometer (3nm) process technology. This directive indicates Nvidia's strategic decision to leverage TSMC's advanced manufacturing capabilities for its next generation of AI hardware. The report suggests that Nvidia is aiming to secure cutting-edge manufacturing for these critical components, which are essential for the performance of its AI accelerators.

Nvidia's choice of TSMC's 3nm process signifies a deepening reliance on the foundry giant for its most advanced AI silicon. This move highlights the intense competition and complexity in securing leading-edge semiconductor manufacturing, particularly for the specialized HBM required to train and deploy large-scale AI models. By assigning its next-gen HBM production to TSMC, Nvidia signals its commitment to pushing the performance envelope and ensuring a steady supply of critical components for its lucrative AI chip business.

This decision will likely have ripple effects across the semiconductor supply chain. Other AI chip designers may face increased pressure to secure capacity on similar advanced nodes, potentially driving up costs and lead times for their own next-generation products. Furthermore, it reinforces TSMC's dominant position in advanced logic manufacturing and its critical role in enabling the current AI boom, while also underscoring the strategic importance of HBM technology for future AI advancements.