Advanced Micro Devices (AMD), a U.S.-based chipmaker and rival to companies like Intel and Nvidia, will consider diversifying its production and partner foundries, as they are currently dependent on Taiwan Semiconductor Manufacturing Co., a reliance that could cause supply disruptions in case of conflict between the U.S. and China.

Lisa Su, chair and CEO of AMD, told Nikkei Asia while visiting Tokyo on Friday that the company will "consider other manufacturing capabilities" besides TSMC to make AMD-designed chips to "ensure that we have the most resilient supply chain."

Su said that "for advanced [chip] development, we don't have anything [planned] currently," admitting that searching for a suitable candidate will not be easy, as TSMC has been dominant in the chipmaking industry and holds cutting-edge technology.

TSMC rivals include South Korea's Samsung, Taiwan's United Microelectronics Corp. and U.S.-listed GlobalFoundries. Su did not name any candidates.

She is open to utilizing TSMC sites located outside Taiwan, including its plant in the state of Arizona where the U.S. company is planning to use some of the facility's capacities. "The fact that there is more manufacturing being developed around the world, including in the U.S. and Japan, I think that's a good thing," Su said, adding, "We would like to use manufacturing [sites] across different geographies to give us some flexibility."

Semiconductor makers are trying to prevent supply disruptions in the event of a conflict between the U.S. and China over Taiwan, as a confrontation in the Taiwan Strait will seriously jeopardize the flow of goods, including semiconductors and production, by TSMC.

The semiconductor industry is enjoying a major boost as tech companies race to develop generative AI such as ChatGPT. Developing AI requires a huge capacity in data centers.

The company recently unveiled its new graphics processors (GPUs), which are good for fast calculations and in developing AI. U.S.-based Nvidia currently holds a major share in the space, while AMD has been prominent in other types of semiconductor chips called central processing units, which have broader applications.

"There is a large opportunity in AI with GPUs, so we have increased our resources significantly," said Su, adding, "AI is the highest priority for the company."

AMD is projecting that the semiconductor market for accelerating the capabilities of data centers will grow around 50% over the next three or four years to $150 billion, driven by bullish AI demand.

Asked about AMD's strategy to compete with Nvidia, Su said that "whenever you have an inflection point [in technology] like AI is today, you can have a significant opportunity to bring different technology capabilities to the marketplace."

The CEO emphasized that in some areas of AI development, AMD's technology and knowledge would be more competitive than Nvidia's.

"AI is going to have many winners," said Su. "There's not just one solution. As workloads [for developing AI] are different, whether you're talking about large language training or inference or some predictive AI, you will have different solutions."

She explained that in areas such as inference workloads -- a process in which AI evaluates and understands new information based on previously processed data -- AMD believes it has "the strongest solution."

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved