For most asset types, ESG due diligence and power supply are separate workstreams. For a Korean data center they are the same workstream. A facility that cannot secure grid capacity has no operations to assess — and by the most recent published count, most capital-region applicants cannot.
On most Korean assets, ESG due diligence and power supply sit in different sections of the report. The building has an energy rating; the operator has a utility contract; the two are reviewed separately, as we set out in our carbon and energy due diligence guide and in our broader survey of environmental risk by asset type. On a data center the split does not hold. Power is not an input to the operation — it is close to the entire operation, and whether the site can get enough of it, on what timeline, and from what source, determines whether there is a business to run an ESG review on at all.
That reordering changes what an early-stage screen should look like. A data center project with a signed land contract, an architect, and a hyperscale tenant letter of intent can still be, in the terms that matter to a lender, a project with no confirmed power — and no amount of building-level ESG work substitutes for that gap. We have carried out ESG and environmental & social due diligence on data center assets in the Seoul capital region and in Busan, and the pattern holds across both: the power question has to be answered before the rest of the scope is worth pricing.
Korea's distributed-energy legislation introduced a power grid impact assessment (전력계통영향평가) aimed at large electricity users — data centers squarely among them — to prevent heavy loads from concentrating in parts of the grid that are already saturated. It replaces the older advance notice of intended power receipt, and functions as a substantive review rather than a formality: an applicant intending to draw large-scale electricity files an assessment, generally through an accredited preparer, ahead of its permit application, and the reviewing authority can require corrective measures — or, in effect, decline to confirm supply — rather than simply acknowledging the filing. Local government consent is also required to proceed on projects that trigger the assessment.
For diligence, the operative fact is that this assessment sits upstream of the building permit, the environmental permits, and the ESG story entirely. A target that has "applied for power" has done exactly that — applied. What diligence needs is the assessment's actual stage: filed, under first-stage technical review, advanced to full review, or resolved, and on what capacity.
The scale of the bottleneck is now measurable. Cumulative capital-region applications for the assessment's first-stage technical review reached 522 cases, totaling 33,592MW of requested capacity, by March 2026. Of those, 279 cases — 53.4% of applications and 53.7% of the capacity requested — were found unable to be supplied at first-stage review. Only 24 cases advanced to full review, and of the full set, just 10 cases, totaling 1,010MW, received a final supply approval: 1.9% of applications and 3% of requested capacity (CBRE Korea, cited by industry press, 14 July 2026).
The diligence implication is specific. A pipeline of applications, or even a favorable first-stage result, is not evidence of secured power — most of the funnel does not clear. What should be verified is where the target's specific application sits in this sequence, what capacity was requested versus what (if anything) has been confirmed, and whether the substation or grid segment in question has already absorbed nearby approvals that would compete for the same headroom.
The assessment has reportedly operated for roughly two years without a finalized implementing notice (고시) codifying its procedures — administered, in effect, on a provisional basis while the formal rule remains outstanding. That matters for how diligence should treat any finding about grid status. A result obtained under a settled rule is a fact that decays slowly; a result obtained under a system still being formalized can shift with the next procedural update, a change in reviewing-office practice, or a revision to how capacity is allocated across competing applicants in the same area. Findings on grid status should therefore carry an explicit as-of date and be re-confirmed close to signing, rather than carried forward from an earlier phase of the deal as though settled.
A dedicated law for the sector, the Act on the Promotion of the AI Data Center Industry, was enacted in June 2026 and is scheduled to take effect in March 2027. Its relief is deliberately geographic: non-capital-region AI data center projects gain exemptions from parts of the power grid impact assessment below a specified scale, access to integrated and streamlined permitting, a basis for direct renewable energy supply arrangements, and eligibility for designated AI data center special zones with supporting infrastructure. Projects designated as national strategic facilities can also route through an integrated approval process under national-level deliberation rather than the standard sequence.
For a target under diligence, this converts location from a cost and latency tradeoff into a regulatory one. A capital-region site inherits the bottleneck described above in full; a comparable non-capital-region project may sit outside much of it once the law takes effect. Deal timing relative to the March 2027 effective date, and whether a project is structured to qualify for the exemptions and special-zone provisions, are now diligence questions in their own right — not just siting preferences.
Once power access is established, the ESG review for a data center concentrates on the sourcing of that power and on the facility's own emissions profile, more than on the building itself:
A Korean data center diligence scope should treat the power grid impact assessment as the first gate to clear, not a line item confirmed by a checkbox. The specific questions worth asking before pricing the rest of the review: what stage has the target's assessment reached, and as of what date; what capacity was requested against what has actually been confirmed; whether the site falls inside the capital-region bottleneck described above or could qualify for the AI data center special act's relief once it takes effect; and what renewable supply is contractually signed rather than corporately promised. Building certifications, cooling design, and generator permitting are real workstreams — but on this asset class, they are worth scoping only after the power question has an answer.
This article is general information, not legal or investment advice, and expresses no view on any company's valuation. Legal provisions and administrative procedures are summarised functionally as reported at publication; article and notice numbers are deliberately not cited, the implementing notice for the power grid impact assessment remained unfinalized as reported, and Korean-language texts and subordinate legislation under the AI Data Center Industry Promotion Act — not yet in force at the time of writing — will control the specifics once issued. The application and approval figures for the capital region are attributed to a CBRE Korea report as cited in industry press dated 14 July 2026 and reflect cumulative filings through March 2026; we did not independently verify the underlying dataset. Site-specific conclusions on grid status, permitting stage, or contract structure require an engagement, not an article.
Power grid status, renewable supply structure, and permitting stage vary by site, substation, and timing relative to the new legislation. We have completed ESG and E&S due diligence on data center assets in the Seoul capital region and in Busan — these should be scoped from the target's own filings, not from sector-wide figures.
Start a confidential discussionLast reviewed: 7 September 2026 · Dime Works