High-Speed Links for AI Infrastructure

The new service is designed to provide dedicated, high-capacity network connections for AI infrastructure, including links between data centers where large-scale AI training and inference take place. By offering speeds of up to 800Gbps, KT positions the offering as a response to the growing bandwidth requirements of AI clusters, which must move vast volumes of data between computing nodes with minimal latency.

Focus on Data Center Interconnection

A central purpose of the AI dedicated line is to strengthen data center-to-data center connections. As AI services scale, operators increasingly need to link distributed facilities into what functions as a single, unified computing resource. Dedicated lines of this kind are intended to give customers predictable performance and secure transmission, in contrast to shared public network paths.

KT, one of Korea's major telecommunications carriers, has been expanding its AI-related networking portfolio as enterprises and public institutions build out AI systems that depend on reliable, high-throughput connectivity between storage, compute and training facilities.

Positioning in the AI Network Market

With the launch, KT joins the broader push among telecom operators to supply the underlying network layer for the AI era. Dedicated high-speed lines are seen as a key enabler for companies running GPU clusters across multiple sites, allowing them to pool resources and transfer datasets without bottlenecking on conventional connectivity.

The company did not announce the service with fanfare about unverified performance figures beyond the headline maximum speed of 800Gbps, and details on pricing, rollout schedule and customer eligibility were not part of the announcement reviewed for this report.

KT's rollout of an AI dedicated line with speeds of up to 800Gbps is True.