Case Study: SK Telecom achieves 5× LLM throughput and 3× cost savings with FriendliAI

SK Telecom achieves 5× LLM throughput and 3× cost savings with FriendliAI

SK Telecom (SKT), South Korea's leading telecom operator, faced the challenge of scaling its enterprise AI agents to meet strict SLAs and handle heavy, unpredictable traffic from millions of users, all while contending with high operational costs and deployment complexity. To address this, they selected the Dedicated Endpoints service from FriendliAI as their production inference platform.

By implementing FriendliAI's Dedicated Endpoints, SKT achieved a 5x increase in LLM throughput and a 3x reduction in operational costs. The vendor's solution provided the isolated, high-performance infrastructure needed to ensure exceptional reliability and efficient handling of concurrent traffic. SKT reached full production readiness within hours, allowing its engineering teams to focus on AI innovation rather than infrastructure management.

SK Telecom