Nvidia is reportedly considering Samsung as a manufacturing partner for its new GPUs, signaling a potential shift in the AI chip landscape. The move comes as Nvidia’s long-standing relationship with Taiwan Semiconductor Manufacturing Co. (TSMC) shows signs of strain, according to sources familiar with the matter.
The U.S. chip designer is allegedly exploring a deal with Samsung for less complex GPU production, aiming for a 20-30% cost reduction compared to TSMC’s pricing. This strategy could help Nvidia diversify its supply chain and reduce dependence on a single manufacturer.
However, industry observers note that Samsung may not immediately secure orders for Nvidia’s most advanced AI chips. The South Korean tech giant has struggled with yield issues in its 3nm and 4nm processes, limiting its ability to attract major customers for cutting-edge designs.
Nvidia’s decision follows reported delays in its Blackwell chips, manufactured using TSMC’s 4nm process. These setbacks have apparently been resolved, with production expected to ramp up in late 2024. Analyst Ming-Chi Kuo projects Blackwell shipments to reach 500,000 to 550,000 units by Q1 2025, with Microsoft emerging as a key customer.
As competition in the AI chip market intensifies, Nvidia’s potential partnership with Samsung could reshape industry dynamics and pressure TSMC to maintain its technological edge.
