Nvidia and SK Group commit over $500 billion to artificial intelligence infrastructure
The chipmaker and the South Korean conglomerate are committing more than $500 billion to build massive data centers and secure next-generation memory, signaling a massive acceleration in global artificial intelligence infrastructure spending.
A massive artificial intelligence infrastructure pact valued at over $500 billion has been announced by Nvidia and South Korea’s SK Group. The agreement covers the development of large-scale data centers alongside next-generation memory systems, securing a critical supply chain for the chip designer while locking in significant regional deployment.
Under the terms of the agreement, SK Hynix will enter a long-term partnership to guarantee memory supplies for the American chipmaker. The two firms will also collaborate on developing high-bandwidth memory tailored for artificial intelligence training, autonomous agents, and physical computing applications.
To operationalize this compute power, SK Telecom intends to construct a massive two-gigawatt artificial intelligence data center. The new facility will run on Nvidia’s Vera Rubin processors and incorporate HBM4 high-bandwidth memory from SK Hynix, with the first phase scheduled to begin operations in 2027.
The broader capital commitment also features a separate expansion of existing digital infrastructure in the region. Nvidia is teaming up with Brookfield and the local search operator Naver to scale up the latter's current artificial intelligence data center footprint across South Korea.
For investors, the sheer scale of the capital commitment underscores the transition of artificial intelligence from a software trend to a foundational physical asset class. The massive financial figure highlights the intense competition among technology giants to secure compute capacity and memory supply chains well into the next decade.
The integration of Vera Rubin chips and HBM4 memory also provides a clear roadmap for the next-generation product cycle of the American chipmaker. By tying its future silicon directly to the infrastructure rollout of the South Korean conglomerate, the company is effectively guaranteeing a baseline of demand for its most advanced hardware.
Furthermore, the agreement solidifies the position of South Korea as a central node in the global technology supply chain. The role of SK Hynix in co-developing high-bandwidth memory ensures that the domestic semiconductor sector remains deeply embedded in the most advanced tiers of global computing infrastructure.
Market participants will now watch closely to see how quickly the telecommunications arm can mobilize the capital required for the two-gigawatt facility. The 2027 launch date for the first site sets a definitive timeline for when this next wave of compute capacity will actually reach the market.
