Nvidia-SK Group $500B AI infrastructure deal: 2GW data center and HBM4 tie-up
Nvidia and South Korea’s SK Group signed letters of intent worth over $500 billion to expand AI infrastructure. The Nvidia-SK Group AI infrastructure deal, announced on July 25, 2026, centers on three pillars.
First, SK Telecom will build a 2-gigawatt AI data center using Nvidia’s Vera Rubin platform, with the first phase targeted to start operating in 2027. Second, Nvidia will deepen a long-term collaboration with SK hynix to co-develop HBM4, the next-generation high-bandwidth memory used in AI training. Third, the broader SK Group relationship aims to reduce global memory supply bottlenecks that limit AI scale.
The technology direction builds on an earlier June 7, 2026 Nvidia–SK hynix announcement aligning memory roadmaps with Nvidia’s AI hardware plans. In addition, Nvidia separately agreed in May 2026 with IREN to deploy up to 5 gigawatts of Nvidia DSX infrastructure globally.
Why traders should care: memory supply (HBM) is a core constraint for AI accelerators, and the Nvidia-SK Group AI infrastructure deal signals long-cycle capacity planning. This can strengthen sentiment around crypto AI narratives—especially networks positioning as decentralized alternatives or complements to centralized AI compute—potentially boosting attention for Bittensor (TAO), Render (RNDR), and Fetch.ai (FET). Risks remain: AI token valuations can unwind quickly with broader crypto market sentiment, and these projects are still earlier-stage infrastructure networks.
Bullish
This Nvidia-SK Group AI infrastructure deal is bullish for crypto AI tokens mainly through sentiment and narrative alignment. Large, capital-intensive commitments—especially a 2GW AI data center plus HBM4 co-development—reduce uncertainty around AI compute readiness. When traditional AI infrastructure capacity becomes more credible, traders often rotate toward “AI on-chain” projects that claim to provide decentralized models, compute, or AI-related services.
Historically, major AI/semiconductor supply-chain headlines have tended to lift AI-focused tokens in the short term, similar to how GPU or data-center availability announcements previously sparked momentum in AI narrative baskets. Here, TAO, RNDR, and FET are explicitly cited as likely beneficiaries.
However, the impact may not be purely linear. In the short term, price action depends on broader crypto market risk appetite; AI tokens can dump fast if BTC/ETH sentiment turns. In the long term, sustained visibility into memory (HBM) and data-center build-outs can keep the “AI infrastructure thesis” supported, but traders should watch for valuation overextension and any follow-through (partnership execution timelines, funding, and network traction) beyond the initial announcement.