SK Telecom: 15 GW of AIDC capacity for Korean sovereign AI
SK Telecom's official 5 July 2026 announcement sets a 15 GW AIDC capacity target to support Korea's ambition to enter the global "AI G3". The first step aims for 5 GW activated progressively from 2029, with Ulsan as the initial site.
1. What changes in the announcement
SK Telecom is not presenting only another data center. The company describes a national infrastructure program that connects power, land, GPUs, cloud operations, and demand for AI services. The announced scale, 15 GW, moves the topic into industrial policy rather than ordinary IT expansion.
The timeline matters just as much: the operator targets an initial 5 GW of capacity from 2029. That creates a long investment trajectory aligned with the construction, power, and cooling cycles required by AI data centers.
2. Why this is a sovereign AI signal
Korea explicitly links compute capacity, national strategy, and economic autonomy. Sovereign AI becomes tangible here through local capacity able to serve models, enterprises, public agencies, and industrial ecosystems without relying only on foreign cloud regions.
Ulsan is a meaningful choice: the site combines available land, an industrial base, and energy access. For Europe, the reminder is useful: AI sovereignty is not only data residency. It also depends on contractable power, operating skills, and market demand large enough to fill the capacity.
3. Practical reading for decision-makers
For Belgian and French organizations, the announcement argues for a layered approach. Sensitive use cases should be mapped according to real needs for residency, dedicated compute, supervision, and reversibility. Not everything belongs in maximum-sovereignty infrastructure, but critical processes need a credible option.
In Odoo Enterprise projects, this logic helps distinguish generic automations from sensitive flows: contracts, HR data, financial decisions, supply chain, or public-sector interactions. These flows need governed AI architecture before they are connected to external agents or models.
Translate this into a practical roadmap: data criticality, AI execution zones, cloud/sovereign choices, and operating governance for Odoo workflows.
Structure the roadmap