From cloud to edge: the value of distributed AI in South Korea

South Korea has one of the most advanced digital economies in the world. It was first to roll out 5G, back in 2019, and its ICT sector makes up 13% of GDP, the highest share in the OECD. And yet beneath that digital strength sit three structural problems no infrastructure alone can fix: by 2050, 40% of the population will be over 65; 61% of local manufacturing SMEs already can’t find enough workers; and with trade equal to 91% of GDP, growth is heavily exposed to labour scarcity and geopolitical shocks.

Our new report, From Cloud to Edge: The Value of Distributed AI in South Korea, argues Korea has a lever capable of acting on all three simultaneously: distributed AI.

Why cloud alone won’t get Korea there

Korea has built a genuinely complete AI stack, from sovereign foundation models to domestic chip manufacturing to a 6G roadmap targeting 2028. But as AI moves from answering questions to controlling physical systems, cloud-based inference starts to show its limits. Healthcare monitoring needs to work instantly without sending biometric data offsite. Industrial robots working alongside people need response times that round-trip latency can’t guarantee. Manufacturers don’t want proprietary data leaving the building at all.

Demand for AI inference compute in Korea could grow 30 times over by 2030, and even with major data centre investment, Korea could still face an 86% shortfall in inferencing capacity. Cloud remains essential, but it can’t do this job alone.

Where the value is

Distributed AI, processing on-device, at the edge, or on-premises rather than in a remote data centre, closes that gap. Across three priority use cases, we estimate this could generate more than USD 50 billion in annual economic value for Korea by 2035:

Autonomous industrial operations (over USD 23 billion), offsetting labour shortages on factory floors and construction sites.
Export competitiveness via distributed AI (over USD 21 billion), as on-device intelligence becomes a point of differentiation for Korean manufacturers abroad.
Companion and assistive robotics (nearly USD 6 billion), supporting older Koreans through mobility assistance, fall detection and health monitoring.

What it will take

Realising this isn’t automatic. We estimate it will require around USD 18 billion of incremental investment between 2026 and 2035, about a quarter of Korea’s total 2026 AI budget, for returns we project at 25 times that annual outlay.

Getting there depends on the policy choices made now: building global interoperability through standards partnerships, promoting vendor diversity, front-loading pre-6G infrastructure investment, and establishing clear regulatory benchmarks for distributed AI before the AI Basic Act’s grace period ends in January 2027.

Korea has already assembled nearly every piece of the AI stack other countries are still trying to build. The opportunity now is to connect them, from cloud to edge, and turn that advantage into leadership.

Read the full report, From Cloud to Edge: The Value of Distributed AI in South Korea, commissioned by Qualcomm and prepared by Access Partnership.