SK Group and Nvidia announced plans for a "$500-billion-plus comprehensive partnership" spanning "AI factory construction" and "AI memory supply"1. Nvidia is also investing $1 billion directly in Naver to finance an AI data center in South Korea4,2. ANALYSIS Together, these commitments convert South Korea from a component supplier into a vertically integrated node of Nvidia's global AI infrastructure — locking down constrained high-bandwidth memory supply while seeding sovereign-AI compute capacity on the Korean peninsula.
Why it matters
Jensen Huang told Bloomberg that Nvidia is "constrained in HBM memories, LPDDR memories" and "constrained in just about every part of the supply chain," including "land and power and construction workers to set up the data centers". SK Hynix is widely regarded as the leader in HBM production3. ◆ By tying SK Hynix into a multi-year framework and simultaneously financing Korean cloud and data-center capacity through Naver, Nvidia is attempting to secure both the memory inputs and the downstream compute outlets for its next-generation systems.
The big picture
SK Group and Nvidia signed letters of intent for what they described as a "$500-billion-plus comprehensive partnership" spanning "AI factory construction" and "AI memory supply". The deal was announced at an AI summit in San Francisco attended by South Korean President Lee Jae Myung. Under the agreement, SK Hynix affiliate SK Telecom will build a cloud business using Nvidia's Vera Rubin systems, and Nvidia said it is targeting enough capacity to require 2 gigawatts of power, with large-scale data centers expected to come online in 2027. Nvidia's Raj Mirpuri said the expansion will include a "co-develop opportunity on the next-generation SK Hynix AI memory" and help "secure a stable supply of HBM memory".
Separately, Nvidia is investing $1 billion directly in Naver to finance an AI data center in South Korea. Naver's Gak Sejong data centre in Sejong will serve as the first physical base for Nvidia's DSX sovereign-AI platform partners. The Naver project would provide potential customers in South Korea and globally the ability to secure AI computing capacity before 200 megawatts of capacity is completed.
The Korea sweep also extends to Doosan and LG. Huang said Nvidia was working with LG Group on electronics, mechanical systems, humanoid robot AI and more. ANALYSIS Taken together, these deals pull at least five major Korean conglomerates — SK, Naver, Doosan, LG, and implicitly Samsung's competitive orbit — into Nvidia's AI factory stack.
Between the lines
Huang's own language frames the strategic logic. He called this "the golden ages for Korea," noting that "their semiconductor business is booming" and "their industrial business is booming" and that Korea "has the ability to help the world build out the AI infrastructure". But his constraint commentary reveals the urgency beneath the diplomacy: "We don't have enough bits," he said, adding that the industry has "the ability as an industry to double each year but we're going to have a hard time growing much faster than that".
ANALYSIS The $500 billion SK framework and the $1 billion Naver equity stake represent two distinct instruments aimed at the same problem. The SK deal secures upstream memory supply — the HBM chips without which Nvidia's GPUs cannot ship — while the Naver investment seeds downstream demand by financing sovereign-AI data centers that will consume those GPUs. The 2-gigawatt power target disclosed for the SK Telecom cloud buildout signals a deployment measured in hundreds of thousands of GPUs.
The competitive backdrop sharpens the picture. Samsung Electronics signed a memorandum of understanding with Broadcom to expand collaboration across memory and foundry technologies in a deal estimated at $200 billion. Nvidia's move to bind SK Hynix into a co-development relationship for next-generation AI memory effectively deepens the split in Korea's semiconductor sector: SK Hynix aligns with Nvidia, while Samsung gravitates toward Broadcom.
SK Hynix listed on the Nasdaq earlier this month as part of an effort to finance infrastructure developments. ANALYSIS That listing now looks like a precondition for the scale of capital the SK-Nvidia partnership envisions — a Korean memory champion accessing U.S. capital markets to fund a buildout denominated in hundreds of billions.
What's next
The letters of intent signed by SK and Nvidia are not yet binding contracts. Data centers under the agreement are expected to come online in 2027. Naver's Gak Sejong facility will be an early test of whether Nvidia's DSX sovereign-AI model can generate paying customers beyond the hyperscalers. ◆ The pace at which these frameworks convert into purchase orders will determine whether Korea becomes a durable pillar of Nvidia's supply chain or another headline partnership that outran its execution.