According to Tom's Hardware and NVIDIA's official blog, NVIDIA and SK Group signed letters of intent on July 24, 2026 in San Francisco for a strategic partnership valued at more than $500 billion. The deal covers a 2-gigawatt AI data center from SK Telecom and next-generation HBM memory co-development with SK hynix, with the first facility set to enter service in 2027.
How large is the NVIDIA–SK strategic partnership, and when and where was it announced?
NVIDIA and SK Group signed letters of intent this week to formalize a new strategic relationship valued at more than $500 billion, according to Tom's Hardware (tomshardware.com), citing the companies' own statement (E1). NVIDIA's official Taiwan blog (blogs.nvidia.com.tw) confirms the same figure, describing an expanded strategic partnership "spanning AI factories and next-generation memory" worth more than $500 billion (E6). Both sources place the announcement on July 24, 2026, in San Francisco, United States (E7).
What is the scale, technology stack, and timeline of SK Telecom's planned AI data center?
Per Tom's Hardware, SK Telecom plans to build a 2-gigawatt AI data center in South Korea, relying on NVIDIA's DSX AI factory platform and deploying Vera Rubin accelerated computing systems equipped with SK hynix High Bandwidth Memory (HBM4) (E2). The first AI data center is set to enter service in 2027 (E3). The report also specifies that NVIDIA's NVL72 VR200 rack system will carry per-rack power consumption ratings of 166 kW to 240 kW (E5).
| Metric | Value | Source |
|---|
| Total partnership value | More than $500 billion | Tom's Hardware (E1); NVIDIA blog (E6) |
| Planned AI data center capacity | 2 gigawatts | Tom's Hardware (E2) |
| First data center service start | 2027 | Tom's Hardware (E3) |
| NVL72 VR200 rack power rating | 166 kW – 240 kW per rack | Tom's Hardware (E5) |
| Announcement date and location | July 24, 2026, San Francisco | Tom's Hardware (E7) |
| SK hynix–NVIDIA memory supply agreement | Long-term, unveiled June 2026 | Tom's Hardware (E4) |
What does the NVIDIA–SK hynix memory agreement cover?
Tom's Hardware reports that NVIDIA and SK hynix already have a long-term memory supply agreement in place, unveiled in June 2026 (E4). NVIDIA's official blog adds that under the new partnership, SK hynix will establish a long-term AI memory partnership with NVIDIA, under which the two companies will jointly develop and optimize next-generation AI memory solutions — including HBM — to meet infrastructure needs ranging from large language model (LLM) training to agentic AI and physical AI (E10).
How do SK Group and NVIDIA leadership describe the significance of this partnership for Korea's AI industry?
According to NVIDIA's official blog, SK Group Chairman Chey Tae-won said: "In the AI era, competitiveness will be determined not just by how effectively we use AI, but by how much intelligence we can create. By integrating SK hynix's AI memory with SK Telecom's AI infrastructure capabilities, SK Group and NVIDIA will build world-class AI factories together, making Korea not just a leading adopter of AI, but a global hub driving AI innovation" (E8). NVIDIA founder and CEO Jensen Huang was quoted in the same post: "Korea has all the elements to become a global AI powerhouse — world-class networks and data centers, leading chip technology, and a massive industrial base. We are building next-generation AI factories together with SK Telecom and SK hynix to power Korea's next wave of growth" (E9).
What this means
The two sources agree on the headline figure — a partnership worth more than $500 billion announced July 24, 2026 in San Francisco (E1, E6, E7) — and together they show the deal resting on two concrete pillars already in motion: SK Telecom's 2-gigawatt AI data center build using NVIDIA's DSX platform and Vera Rubin systems, targeted for 2027 service (E2, E3), and a memory relationship with SK hynix that predates this announcement, with the June 2026 supply agreement (E4) now being extended into joint development of next-generation HBM (E10). Chey Tae-won's and Jensen Huang's remarks frame this infrastructure buildout explicitly around Korea's positioning in AI — from "leading adopter" to "global hub" in Chey's words (E8), and from existing chip and data-center strengths to a "next wave of growth" in Huang's words (E9) — tying the hardware specifics (2GW capacity, 166–240 kW racks, HBM4) to a stated national-scale ambition.