NVIDIA, in partnership with SK Group, announced a sprawling "AI factories" initiative in South Korea, projected to be worth over $500 billion over multiple years, following South Korean President Lee Jae Myung's visit to San Francisco. Actual financial terms and regulatory filings for the massive sum remain undisclosed.

SAN FRANCISCO — Graphics processing unit giant NVIDIA and South Korea’s SK Group announced a strategic partnership to build "AI factories" in South Korea and co-develop next-generation memory, following a meeting between NVIDIA CEO Jensen Huang and South Korean President Lee Jae Myung in San Francisco. The initiative, revealed on Friday, July 24, 2026, is projected by NVIDIA to be "worth $500 billion or more over a number of years."

This ambitious projection, while announced publicly, currently lacks the specifics of a regulatory filing or a detailed public financial breakdown. Searched SEC Form D filings for both "SK Group NVIDIA" and "NVIDIA AI factories" as of Tuesday showed no corresponding offerings, and no chatter on Hacker News has emerged beyond the initial reports. The announced partnership stems from "letters of intent," indicating a forward-looking agreement rather than a closed financing round.

The core of the initiative involves SK Hynix affiliate SK Telecom building a cloud business utilizing NVIDIA’s Vera Rubin systems. This "massive buildout," according to NVIDIA, aims for a capacity requiring 2 gigawatts of power, with the first AI factory anticipated to begin operations in 2027. Raj Mirpuri, NVIDIA enterprise vice president, confirmed aspects of the deal on a call with reporters, stating, "The expansion will include a co-develop opportunity for us on the next-generation SK Hynix AI memory, and this will help us secure a stable supply of HBM memory." Concurrently, SK Hynix, South Korea's second most valuable company, recently listed on Nasdaq, an action CNBC reported was "part of an effort to finance infrastructure developments." Furthermore, NVIDIA separately committed $1 billion to Naver, a Korean cloud company, for data center development.

The declared $500 billion figure is significant, even in the context of recent large-scale tech investments, and signals NVIDIA's aggressive strategy to secure high-bandwidth memory supply amidst a global shortage—SK Hynix being a leader in HBM production. However, without public financial terms, a specific investment vehicle, or regulatory filings detailing the commitment, the announced figure remains largely an aspirational projection. The absence of U.S. government incentives or lobbying disclosures also distinguishes this deal from other major semiconductor industry plays.

What remains unconfirmed are the precise financial mechanisms underpinning the "over a number of years" projection, the legal binding nature of the letters of intent, and a detailed timeline for the multi-phase AI factory development. The industry will be watching for concrete public filings and a clearer articulation of the financial roadmap that substantiates this half-trillion-dollar aspiration.