SK Telecom AI data center revenue jumps 89%, signaling telco-led infrastructure push
The AMW Read
Incremental update on a known telco-turned-AI-infra player; significance is segment-level because it shows a non-hyperscaler revenue trajectory that could reshape compute supply dynamics.
SK Telecom AI data center revenue jumps 89%, signaling telco-led infrastructure push
South Korea's SK Telecom reported an 89% year-over-year increase in AI data center revenue, underscoring the telecom giant's deepening pivot toward AI infrastructure. The company has been expanding its GPU-as-a-service offerings and building out data center capacity to serve both domestic and global enterprise AI workloads.
The revenue surge reflects a broader structural force in the AI industry: incumbent telecom operators are leveraging their existing real estate, power agreements, and network connectivity to become competitive players in the AI infrastructure segment. This pattern — telcos morphing into AI compute providers — adds a new dimension to the compute supply chain, where traditionally only hyperscalers and pure-play data center REITs dominated. SK Telecom's performance validates the thesis that telco-owned data centers can capture meaningful share of the AI inference and training market, especially in regions with high energy costs or grid constraints.
From an expert lens, this is an early signal of the "telco-as-infrastructure-layer" pattern gaining traction. SK Telecom's 89% growth rate, while off a smaller base, outpaces most hyperscaler data center revenue growth rates reported this quarter. If the company sustains this trajectory, it could force a reassessment of how AI compute assets are valued — particularly whether telcos deserve a re-rating similar to what Equinix and Digital Realty have enjoyed. The key watchpoint is whether this growth is driven by inference demand (higher margin, recurring) or training burst rentals (lumpy, lower retention).
