Nvidia To Redesign Feynman AI Chip Due To TSMC Capacity Shortage
Nvidia is reworking the design of its next-generation Feynman AI chip platform. This change comes after it failed to secure enough production capacity on TSMC’s most advanced manufacturing process. As a result, the company must adopt a more flexible approach to production.
Under the revised plan, only the most critical components will use TSMC’s A16 1.6-nanometer process. Meanwhile, less essential parts will shift to the older N3P 3-nanometer node. Consequently, this hybrid strategy may introduce design trade-offs and higher production costs. In addition, supply constraints could emerge for a platform central to Nvidia’s AI roadmap.
Growing Pressure on Chip Manufacturing
This situation reflects a broader capacity bottleneck that has developed over several years. Currently, demand for advanced nodes below 2 nanometers continues to surge. Therefore, production capacity remains fully booked through 2028 and possibly beyond.
At the same time, TSMC is expanding its manufacturing footprint. The company has filed plans for a new fabrication facility in Tainan Science Park. Construction is expected to begin this year, with completion targeted for 2028. As capacity tightens, pricing pressure is also increasing across the semiconductor industry.
Earlier this year, CEO Jensen Huang highlighted the scale of demand facing the company. He noted that overall manufacturing capacity could more than double over the next decade. Nevertheless, near-term shortages continue to challenge production timelines.
Feynman’s Role in Future AI Hardware
Nvidia first introduced the Feynman architecture at its GTC 2026 conference in San Jose on March 15. The platform is positioned as the successor to the Vera Rubin chip family. Initially, it was designed to rely fully on TSMC’s A16 process. However, the revised approach reflects current manufacturing limits.
Looking ahead, the company plans to launch Feynman in 2028. Customer deliveries may extend into 2029 or even 2030. Meanwhile, analysts suggest Nvidia could diversify its supply chain. For example, Intel may produce less complex components such as the I/O die.
Overall, this development highlights the growing tension between rising AI demand and limited chip manufacturing capacity. As demand accelerates, companies must adapt their strategies to secure production and maintain innovation.
