NVIDIA 2026-07-26
Industry Signal Impact: Major Conf: 90%

NVIDIA and SK Group Lock HBM4 Supply and Launch Sovereign AI Factory Model with $500B+ Deal

Summary

NVIDIA and SK Group announced a $500B+ AI partnership including a 2GW AI factory using Vera Rubin and HBM4, long-term HBM4 supply lock, and a $1B NVIDIA investment in Naver (with $9B from Brookfield). Samsung and Broadcom signed a $200B deal. This signals a new era of sovereign AI infrastructure and supply chain deep-locking.

Key Takeaways

NVIDIA and SK Group announced a $500B+ comprehensive AI partnership covering AI factories, AI storage, and AI chips. SK Telecom will build a 2GW AI data center using NVIDIA Vera Rubin and DSX AI Factory platform with SK Hynix HBM4, operational by 2027 with 60,000 GPUs. SK Hynix and NVIDIA will co-develop next-gen HBM; SK Hynix holds 58% HBM market share (Q1 2026 Counterpoint) and its full-year capacity is booked. NVIDIA invested $1B in Naver with $9B from Brookfield, tripling Naver's data center scale. Samsung and Broadcom signed a $200B deal for chips, 2nm foundry, and advanced packaging. This pushes AI Capex to national scale, locks HBM4 supply for NVIDIA through 2027-2028, and forms a sovereign AI model with bidirectional capital binding, but raises concerns about circular transactions.

Why It Matters

NVIDIA's move is essentially a strategy to starve competitors (AMD, Intel) of HBM4 supply, securing exclusive advantage for Vera Rubin. By investing in Naver and locking them into DSX AI Factory, NVIDIA creates a closed ecosystem that risks platform lock-in for enterprises. The 2GW AI factory's power and cooling challenges are downplayed, while large-scale GPU clusters often face network congestion and tail latency issues, masked by NVIDIA's full-stack approach. The circular transaction model (NVIDIA investing in customers to generate demand) may distort market signals and attract antitrust scrutiny. The HBM4 supply lock until 2028 also creates technology lock-in risk.

PRO Decision

【Vendors】Competitors (AMD, Intel) should partner with Samsung and Micron for HBM4 supply, promote open platforms (OCP, ROCm, OneAPI) to break NVIDIA's DSX lock, and offer open reference designs for sovereign AI factories. They should leverage antitrust scrutiny on NVIDIA's supply chain lock. 【Enterprises】CIOs should audit single-vendor dependency on NVIDIA, demand multi-architecture support (AMD MI400, Intel Gaudi 3), and assess DSX AI Factory's API openness and data portability. Avoid long-term lock-in contracts. 【Investors】Beware of NVIDIA's circular transaction model creating artificial demand and systemic risk. Monitor antitrust reviews of NVIDIA's HBM4 supply lock. Diversify into open ecosystem players (AMD, Intel) and HBM suppliers (Samsung, Micron).

Source: 36氪
View Original →

Get 3-5 key AI infrastructure signals weekly →

💬 Comments (0)