SUN, JULY 26, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

Nvidia and SK Group Sign $500B+ AI Deal: 2GW Vera Rubin Factory and SK Hynix Locked as HBM Supplier

Nvidia and SK Group signed letters of intent July 25, 2026 for a $500 billion-plus AI infrastructure partnership — the largest single AI infrastructure commitment announced to date. SK Telecom builds a 2-gigawatt Vera Rubin DSX AI factory (online 2027). SK Hynix locked as Nvidia's long-term HBM supplier and co-developer of next-gen HBM4. Nvidia invests $1B in Naver. Jensen Huang announced the deal during a meeting with South Korean President Lee Jae Myung.

By AIToolsRecap July 26, 2026 6 min read 71 views
Home Articles News Nvidia and SK Group Sign $500B+ AI Deal: 2GW Ve...

NVIDIA × SK GROUP — KEY FACTS (JULY 25, 2026)

Deal value: $500 billion+ (letters of intent signed)
AI Factory: 2-gigawatt facility by SK Telecom using Nvidia Vera Rubin DSX — online 2027
Memory: SK Hynix locked as Nvidia's long-term HBM supplier, co-developing next-gen HBM
Naver investment: Nvidia commits $1B to Naver for Korean data center expansion
SK Hynix market position: ~60% of global HBM market, Nasdaq-listed (raised $26.5B in June 2026)
Announced: Jensen Huang + South Korean President Lee Jae Myung meeting, San Francisco, July 25
Chips: Nvidia Vera Rubin + SK Hynix HBM4 — next generation beyond Blackwell

What Was Signed and What It Means

SK Group and NVIDIA expanded their strategic collaboration with a $500 billion-plus initiative spanning AI factories and next-generation memory. SK Telecom will build a 2-gigawatt NVIDIA Vera Rubin DSX AI Factory to serve global compute demand. NVIDIA and SK Hynix established a long-term partnership to secure and co-develop next-generation AI memory, including HBM. These are letters of intent — a formal agreement structure, not a completed commercial contract. But at $500 billion-plus, even a letter of intent at this scale signals infrastructure commitments that take years and billions to deliver. The 2027 factory online date makes this a planning announcement, not a near-term capacity event.

SK Hynix, which currently controls roughly 60 percent of the global HBM market, will enter a long-term supply arrangement with NVIDIA and co-develop next-generation high-bandwidth memory covering applications from large language model training to physical AI. SK Hynix's record Nasdaq debut last month raised $28 billion and formally put the South Korean chipmaker into the global AI infrastructure race. The memory supply deal is the more significant of the two components for the AI industry. HBM (high-bandwidth memory) is the critical bottleneck for AI chip performance — Nvidia's H100 and H200 GPUs use SK Hynix HBM3E, and the next generation Vera Rubin will use HBM4. Locking SK Hynix into a long-term supply and co-development arrangement secures Nvidia's access to the memory substrate that determines how fast its chips can actually run AI workloads.

The 2GW Vera Rubin Factory — What 2 Gigawatts Actually Means

2 gigawatts of AI data center power is a meaningful reference point. Microsoft's entire global data center portfolio consumes approximately 10 gigawatts. Anthropic's Colossus 1 deal covers 300 megawatts. A 2-gigawatt facility is roughly 6.7× Colossus 1 — a very large single AI factory by any measure. SK Group Chairman Chey Tae-won framed the deal as a competition to produce intelligence rather than merely consume it. This reframes Korea's AI strategy: not buying AI services from US providers, but owning the infrastructure that produces AI compute — and selling that compute globally through SK Telecom's cloud business. The Vera Rubin DSX platform is Nvidia's next-generation full-stack AI factory architecture, succeeding the H100/H200 DGX systems. Vera Rubin chips are expected to ship in 2026-2027.

Why This Deal Matters for AI Infrastructure Competition

HBM supply security: Global investments in data centers are projected to be close to $1.6 trillion by 2030, while tech giants worldwide are likely to spend $600 billion on AI infrastructure in 2026 alone. The limiting factor is not chips — it is memory. SK Hynix controls 60% of HBM. Locking them into a long-term co-development agreement secures Nvidia's supply chain for the next 3-5 years of AI infrastructure buildout.

Korea as AI infrastructure hub: The Nvidia-Naver-Brookfield data center expansion plus the SK Telecom 2GW factory means South Korea is building significant sovereign AI infrastructure rather than depending on US or Chinese cloud providers. The $26.5B SK Hynix Nasdaq raise was explicitly to finance this infrastructure push.

Context: who else is building at this scale: Google ($44.9B CapEx in Q2 2026 alone), Microsoft ($13.75B/quarter), Anthropic ($1.25B/month at Colossus 1), Meta ($14B+ CapEx in Q2). The Nvidia-SK deal adds a non-US sovereign actor to the list of entities building at genuine frontier AI infrastructure scale.

Sources: SK Hynix Newsroom · CNBC · Reuters · GlobeNewswire · EasternHerald · Cryptopolitan · Related: Google $44.9B CapEx Q2 2026 → · Anthropic $1.25B/month Colossus deal →

Tags
NvidiaAI NewsGenerative AI2026

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →