Nvidia typically relies on Taiwan’s TSMC for fabricating its cutting-edge graphic processing units, but hasn’t ruled out the possibility of looking elsewhere should the need arise, leaving the door open for Samsung to become a possible partner.
At Goldman Sachs Communacopia and Technology Conference in San Francisco this week, Nvidia CEO Jensen Huang said he is willing to consider other companies to produce Nvidia’s chips “if necessary”.
“We use them (TSMC) because they’re great, but if necessary, of course, we can always bring up others,” Huang said.
Huang has not specifically mentioned who the “others” would be, whether it would be Samsung Electronics or its rivals. But Samsung would be a likely candidate, being the foundry market’s No. 2 after TSMC.
Nvidia Blackwell Architecture (Image: Sourced from Nvidia Newsroom)
According to market tracker TrendForce, TSMC took a 61.7 percent share in the global foundry market in the first quarter of this year. Samsung followed with 11 percent, while China-based SMIC came in third with 5.7 percent.
TSMC is currently the sole producer of Nvidia’s most advanced chip products including the H100, H200 and the next-generation Blackwell.
Nvidia is the foundry’s second-largest customer and contributed 11 per cent of the company’s total revenue in 2023.
“We’re fabbing at a TSMC because it’s the world’s best and it’s the world’s best not by a small margin, it’s the world’s best by an incredible margin,” said Huang.
Explaining the conditions under which Nvidia could consider alternatives, Huang said, “Maybe the process technology is not as great, maybe we won’t be able to get the same level of performance or cost, but we will be able to provide the supply.
“In the event anything were to happen, we should be able to pick up and fab it somewhere else.”
Samsung has already adopted the cutting-edge Gate-All-Around technology for its advanced second-generation 3-nanometer process node and 2 nm process.
TSMC intends to adopt the GAA technology for the next 2 nm process.
Nvidia, one of the world’s most valuable companies, has said that it will be designing new AI chips every year.
Until now, Nvidia produced a new architecture roughly once every two years — revealing Ampere in 2020, Hopper (on which the H100 is based) in 2022, and Blackwell in 2024.
In June, Huang announced a Blackwell Ultra chip for 2025 and a next-generation platform in development called Rubin for 2026.
However, with geopolitical tensions soaring between China and the US, Nvidia may keep the door open as to where it plans to undertake its manufacturing in case it is forced to shift production away from Taiwan.
