Samsung, SK Hynix to supply memory chips to OpenAI’s Stargate project

Proposed order for 900,000 advanced DRAM wafers by 2029 could exceed $70b

Stargate
Google search engine
  • Joint ventures will launch two flagship AI data centres with an inaugural capacity of 20MW.

OpenAI’s quest to cement its place at the forefront of artificial intelligence just took a dramatic leap forward—this time, with powerful allies from South Korea.

Samsung Electronics and SK Hynix, two titans wielding dominance in the global memory chip arena, have both signed pivotal letters of intent to supply high-performance memory chips to OpenAI’s rapidly expanding data centre empire.

The collaboration marks a defining turn in the race to deliver on the ambitions of the Stargate project, OpenAI’s $500 billion brainchild set to transform the world’s AI infrastructure landscape.

South Korea as Asia’s new AI hub

The energy in central Seoul was palpable as OpenAI CEO Sam Altman joined President Lee Jae Myung and the leaders of Samsung and SK Hynix at the presidential office. Far more than a photo-op, the event laid the foundation for two advanced data centers on Korean soil, part of a “Korean-style Stargate.”

Seoul’s determination to evolve into Asia’s AI nucleus is backed by its massive base of ChatGPT subscribers—outnumbered only by those in the United States. Altman captured the moment, stating, “We’re very excited to get to build Stargate Korea and data centres with our wonderful partners to support the sovereign AI needs of Korea.”

The Stargate ecosystem

South Korea’s chipmaking power is foundational to this vision. Combined, Samsung and SK Hynix control nearly 70 per cent of the global DRAM (Dynamic Random Access Memory) market and an even more commanding share (close to 80 per cent) in high-bandwidth memory (HBM)—the must-have component for turbo-charged AI applications.

Analysts estimate that OpenAI’s proposed order for 900,000 advanced DRAM wafers by 2029 could exceed 100 trillion won, or around $70 billion, though that figure may shift with memory cycles.

United States leadership remains central. It was US President Donald Trump who announced Stargate’s ambitious grounds in January, charging partners like Oracle, SoftBank, and now the Korean chipmakers, with ensuring US primacy in AI. Nvidia, meanwhile, has pledged up to $100 billion in data centre technology for OpenAI, further amplifying the global stakes.

Building the future

As OpenAI cements itself in Korea—having opened its first Seoul office this year—it also signals a time of expansion and high expectations. The joint ventures with Samsung and SK Hynix will launch two flagship AI data centres with an inaugural capacity of 20 megawatts.

South Korea’s administration has even kept the door open to potential public financing. Presidential adviser Kim Yong-beom highlighted that “the significant part of the Stargate project would be impossible without memory chips from the two companies,” describing the partnership as a leap forward for Korea’s domestic chip industry.

An AI gold rush is now well underway: alongside the chip supply deals, Samsung SDS, Samsung Heavy Industries, and Samsung C&T are all joining the Stargate ecosystem.

From developing floating offshore data centres to pioneering energy-efficient cooling, their projects aim to redefine both infrastructure and sustainability in AI.

Opportunity and caution

Of course, not everything is smooth sailing. Industry watchers worry about the prospect of overbuilding and the risk of an AI investment bubble. Stargate itself has experienced delays tied to complex negotiations and site selection.

Yet, the prevailing sentiment is bullish: the collective expertise of Korea’s industrial giants, global AI leaders, and US tech visionaries is building a new backbone for the planet’s digital future.


Discover more from TechChannel News

Subscribe to get the latest posts sent to your email.

https://www.techchannel.news/wp-content/uploads/2024/06/arrow.jpg