Samsung Electronics said its contract manufacturing trade plans to offer a one-stop shop for clients to get their artificial intelligence (AI) chips made quicker — joining its worldwide No. 1 memory chip, foundry, and chip packaging administrations to tackle the AI boom.
Samsung said on Wednesday (12-06-2024) that clients working with a single communication channel that coordinates Samsung’s memory chip, foundry, and chip packaging groups at once have cut the time it takes to create AI chips, ordinarily weeks, by around 20%.
“We are really living in the age of AI. The rise of generative Artificial Intelligence is totally changing the innovation landscape,” said Siyoung Choi, president and general manager of the foundry business, at a Samsung event in San Jose, California.
Choi said Samsung anticipates the worldwide chip industry's income to reach $778 billion by 2028, boosted by AI chips. Samsung unveils blueprint for AI Chips that promises advantages of AI technology.
At a briefing with correspondents ahead of the event, Executive Vice President of Foundry Sales and Marketing Marco Chisari said the company accepts OpenAI CEO Sam Altman’s loose projections on soaring requests for AI chips are realistic.
Reuters has already reported that Altman has told officials at contract chipmaker TSMC that he needs to construct roughly three dozen new chip factories.
Samsung is one of the few companies that offers memory chips, foundry services, and plans chips under the same roof. That combination has frequently worked against it in the past, as a few clients were anxious that doing trade with its foundry might advantage Samsung as a competitor in another field.
However, with the skyrocketing demand for AI chips and the need for all the chip parts to be exceedingly coordinated to prepare or gather tremendous amounts of information quickly by utilizing less power, Samsung accepts that its turnkey approach will be a quality going forward.
The South Korean tech giant also touted its cutting-edge chip design known as gate-all-around (GAA), a sort of transistor engineering that helps progress chip execution and diminishes power consumption.
GAA is seen as imperative to keep making more capable chips for AI as chips become better and push the boundaries of physics.
Although competitors such as worldwide foundry No.1 TSMC are also working on chips utilizing GAA, Samsung began applying GAA before and said it plans to mass produce its second-generation 3-nanometer chips utilizing GAA in the second half of this year.