Tech Giants Partner on AI Data Centers

Why it matters

This partnership directly addresses the core bottlenecks in scaling AI: power, cooling, and speed. AI-enabled racks can consume up to six times more power than traditional ones, and constructing a conventional data center can take 18 to 24 months, a timeline completely misaligned with the rapid six-month evolution cycle of AI technology. The prefabricated modular approach cuts deployment time by up to 60%, potentially bringing a data center online in just 6 to 9 months. By manufacturing and testing components in a factory, this model also reduces initial capital expenditure by up to 40% and allows for a "pay-as-you-grow" scaling model, which is critical for managing the massive upfront investment AI infrastructure requires. Each partner brings a distinct piece of the puzzle. SK Telecom provides its expertise in operating large-scale data centers as part of its new "AI Native" strategy, which aims to establish South Korea as Asia's largest AI data center hub with a 1-gigawatt infrastructure plan. Schneider Electric, a leader in energy management, will supply the critical mechanical, electrical, and plumbing (MEP) systems, including the advanced liquid cooling solutions needed for high-density AI clusters. Supermicro will deliver the core computing hardware with its high-performance GPU servers optimized for AI workloads. The company specializes in building blocks-style, liquid-cooled server racks designed to handle the latest high-demand processors from NVIDIA and AMD, positioning itself as a key provider of full-stack AI infrastructure, not just components. This collaboration is a key element of SK Telecom CEO Jung Jai-hun's strategic overhaul, in which he stated the company faces a "crisis" and will "perish" if it fails to adapt to the AI era. The company is investing billions to rebuild its core systems around AI, from its mobile network to its customer service operations.

Key numbers

• AI-enabled racks can consume up to six times more power than traditional ones, and constructing a conventional data center can take 18 to 24 months, a timeline completely misaligned with the rapid six-month evolution cycle of AI technology.
• The prefabricated modular approach cuts deployment time by up to 60%, potentially bringing a data center online in just 6 to 9 months.
• By manufacturing and testing components in a factory, this model also reduces initial capital expenditure by up to 40% and allows for a "pay-as-you-grow" scaling model, which is critical for managing the massive upfront investment AI infrastructure requires.
• SK Telecom provides its expertise in operating large-scale data centers as part of its new "AI Native" strategy, which aims to establish South Korea as Asia's largest AI data center hub with a 1-gigawatt infrastructure plan.

What happens next

• SK Telecom provides its expertise in operating large-scale data centers as part of its new "AI Native" strategy, which aims to establish South Korea as Asia's largest AI data center hub with a 1-gigawatt infrastructure plan.
• Schneider Electric, a leader in energy management, will supply the critical mechanical, electrical, and plumbing (MEP) systems, including the advanced liquid cooling solutions needed for high-density AI clusters.
• Supermicro will deliver the core computing hardware with its high-performance GPU servers optimized for AI workloads.

Sources

• have signed
• This partnership directly
• The prefabricated modular
• SK Telecom provides its
• Schneider Electric, a
• Supermicro will deliver
• The company specializes
• This collaboration is
• The company is investing

Quick answers