Trend Cloud providers, chipmakers and generative AI firms race to build AI data centres
Summary
KISDI describes AI data centres as specialised infrastructure built to support high-performance computing, machine learning, deep learning and generative AI workloads. They are equipped with GPUs, TPUs, AI accelerators and optimised cooling and energy management systems, and both hyperscale centres for large-scale training and edge computing for real-time processing are expanding, along with investment to address high power use. The market is being led not only by traditional cloud providers but also by semiconductor companies and generative AI firms. Google, Amazon and Microsoft are investing heavily to meet demand for AI-as-a-service, while Nvidia, Samsung Electronics and SK hynix are expected to lead the hardware base of GPUs, HBM and DRAM. OpenAI is investing in large AI data centres worldwide through the Stargate project and, in Korea, has announced plans with Samsung SDS and SK Telecom to build AI data centres in Pohang and South Jeolla.
Classification
Evidence 1
- 정보통신정책연구원(KISDI) ICT 산업 중장기 전망(2026~2030) 및 대응 전략 정보통신정책연구원(KISDI) no link — bibliographic entry p. 143 2025-12-31 accessed 2026-09-30
Observed signals 4
- SignalAI data centre market projected to grow 28.3 percent a year globally and 36.5 percent in Korea to 2030
- SignalKorea and US tech giants commit to a $950 billion AI and chip alliance
- SignalNscale and Figure Sign Strategic Partnership to Power the Next Generation of Physical AI
- SignalReflection AI begins $6.3B compute lease at SpaceX Colossus 2 (Memphis)
Part of issues 3
- IssueAI infrastructure shortfalls in power, sites, data and talent could slow Korea's ICT growth1 trends · 0 signals
- IssueKorea's plan for a public-private National AI Computing Center with 50,000-plus GPUs1 trends · 0 signals
- IssuePolicy task 4: fast-track AI infrastructure and expand the AI workforce1 trends · 0 signals
Relation types: constitutes · supports
Public id: fm-c5867f2f35a4
