SEOUL — SK Telecom (SKT) and Nvidia have announced a landmark partnership to co-build gigawatt (GW)-scale Artificial Intelligence Data Centers (AIDCs), commonly referred to as "AI Factories." The two tech giants aim to launch a comprehensive "Full-Stack AI Cloud" business, spanning everything from semiconductor supply to advanced data center operations. The ambitious initiative will debut in South Korea next year before expanding aggressively across the Asian continent.
The agreement materialized following a high-level meeting in Taiwan between Chey Tae-won, Chairman of SK Group, and Jensen Huang, CEO of Nvidia. According to SKT on June 8, the two leaders reviewed their shared AI infrastructure roadmap and agreed to elevate their partnership to a conglomerate-wide strategic alliance. Leveraging Nvidia’s DSX platform, the joint venture will integrate chips, systems, infrastructure software, facilities, and partner technologies into a seamless AI cloud ecosystem.
A New Era of Data Infrastructure: The 'AI Fab'
Unlike traditional data centers that focus primarily on general-purpose computing and storage, the newly envisioned AI Factories are designed specifically for intensive AI training and inference. Huang described these facilities as factories that consume power and data to manufacture "tokens"—the fundamental units of AI computation.
"AI Factories will become indispensable infrastructure for Korea’s citizens, educational institutions, universities, research labs, startups, and industries," Huang stated during a briefing at the SK Seorin Building in Seoul. "Just as we once needed traditional fabs for semiconductors, we now need AI fabs for artificial intelligence. AI infrastructure will become as essential as electricity, water, and the internet, and SKT will lead the charge in establishing it here."
The first AI Factory is scheduled to operationalize in South Korea by 2027. Moving forward, the infrastructure will scale up to gigawatt capacity, establishing a robust AI computing network across major Asian markets.
Synergizing Blackwell, Rubin, and HBM Technology
Through this partnership, SKT officially joins the "Nvidia Cloud Partner" program, cementing its position as a premier AI cloud provider. The South Korean telecom giant plans to roll out AI training and inference services utilizing Nvidia’s cutting-edge Blackwell GPUs, with plans to sequentially adopt the next-generation "Vera Rubin" platform, which is slated for release later this year.
The collaboration represents a highly complementary alliance. SKT secures world-class computing and software infrastructure from Nvidia, while Nvidia capitalizes on SK Hynix’s industry-leading High Bandwidth Memory (HBM) technology alongside SKT’s operational expertise in building large-scale data infrastructures.
Driven by this synergy, SKT plans to rapidly scale its AI Cloud business. Unlike conventional cloud services, AI Cloud specializes in heavy workloads such as machine learning and data processing. It has recently emerged as a critical bottleneck in the global supply chain, with global Big Tech firms actively competing to lease large-scale AI infrastructure.
Expanding Beyond Chips into Physical AI and Robotics
The scope of cooperation between SK Group and Nvidia extends far beyond hardware supply. The two companies are forming a joint consultative body to conduct research on next-generation AI factory architectures. This shifts their historical cooperation—which focused heavily on HBM memory supply—into a co-design paradigm where GPU and memory performance are optimized simultaneously from the foundational architectural stage.
Furthermore, collaboration in Physical AI and robotics is already yielding tangible results. At the recent 'GTC Taipei' keynote, Nvidia highlighted a large-scale digital twin technology built by SKT using the Nvidia Omniverse platform. SKT is currently deploying this technology across SK Hynix’s semiconductor manufacturing lines to optimize production. In the robotics sector, the companies are also co-developing simulation and training platforms utilizing Nvidia Cosmos and Isaac GR00T.
"Through this close partnership with Nvidia, we have secured full-stack AI infrastructure competitiveness that spans from chips to data center operations," said SK Group Chairman Chey Tae-won. "Going beyond simple service provision, our two companies will jointly address GPU, memory, and energy challenges, establishing ourselves as Asia's leading AI cloud provider driving the regional AI ecosystem."
Nvidia's Huang echoed this vision, emphasizing the evolving role of telecommunications. "Telecom networks are transitioning into national AI infrastructure," Huang noted. "Through the Nvidia DSX platform, SKT will build a massive AI cloud, enabling them to deliver Agent AI, Enterprise AI, and Physical AI to leading companies and industries both in Korea and globally."
[Copyright (c) Global Economic Times. All Rights Reserved.]
