Nvidia가 SK Hynix와 $500억 규모의 AI 얼라이언스를 체결하며, 사상 최대 규모의 HBM 메모리 공급 파트너십을 확보했다.

Nvidia has entered into a strategic partnership with SK Group valued at more than $500 billion (approximately KRW 720 trillion), announced in San Francisco on the 26th (local time) according to CNBC. The collaboration, highlighted by SK Group Chairman Chey Tae-won and Nvidia CEO Jensen Huang, encompasses the establishment of AI data centers alongside joint development of next-generation memory. This initiative represents a proactive effort to secure a stable core supply chain amid surging demand for AI semiconductors that has triggered a global shortage of high-bandwidth memory (HBM).

The partnership’s scope extends beyond chip supply to include large-scale AI infrastructure development led by SK Telecom. The telecom giant plans to launch an AI cloud business with a total capacity of 2 gigawatts (GW) powered by Nvidia’s next-generation “Vera Rubin” system, with the first facility scheduled to begin operations in 2027. A 2-GW capacity is sufficient to run hundreds of thousands of graphics processing units (GPUs), signaling a broader shift in AI infrastructure investment from a handful of major U.S. technology firms to foreign governments and large conglomerates.

Under the agreement, SK hynix and Nvidia will jointly develop next-generation AI memory, including HBM, and establish a framework for long-term supply. For Nvidia, this secures a reliable pipeline of HBM, a critical component for AI accelerator performance. For SK hynix, it offers early involvement in the memory design phase of Nvidia’s upcoming platforms. As GPU demand accelerates following the AI boom, HBM has emerged as a primary bottleneck in the global semiconductor supply chain. SK hynix continues to dominate the HBM market and remains a key supplier to Nvidia.

On the same day, Samsung Electronics signed a memorandum of understanding with American semiconductor company Broadcom to deepen cooperation in memory and foundry sectors. The two firms plan to co-develop and jointly supply the memory and contract manufacturing technologies required for next-generation AI infrastructure. Samsung expects the partnership to exceed $200 billion in value over the next five years through 2030. By integrating both memory and foundry operations, Samsung aims to broaden its footprint across the AI semiconductor supply chain.

Separately, Nvidia has committed $1 billion in investment to Naver. The funding will support participation in a domestic AI data center project co-promoted by Naver and global asset management firm Brookfield. Naver intends to initially deploy AI infrastructure with a capacity of 55 megawatts (MW), with plans to scale up to the gigawatt level in subsequent phases.