The Spec Was the Signal

The constraint in AI data center development is not compute. It is power. More precisely, it is grid access — and grid access cannot be purchased at any price.

This structural reality reshapes how to interpret a tenant's behavior before a contract closes.

When the Buyer Changes the Brief

SGC Energy's initial AI data center design called for 40MW. Before any contract was signed, the anchor tenant requested an upward revision to 60MW. That sequence — buyer engaging with the infrastructure spec at design stage — is more informative than any financial projection.

In conventional commercial real estate, the developer builds the asset, then finds tenants to occupy it. Tenants respond to the finished product. When a tenant instead participates in shaping capacity parameters before committing, the nature of the project changes. Demand risk — the primary uncertainty in speculative development — is answered in advance. Not because the contract is signed, but because the intent is demonstrated at a level that carries operational cost for the tenant.

Requesting a capacity increase from 40MW to 60MW means the tenant sized its AI compute requirements, found the site capable of meeting that requirement, and then asked the developer to design for it. That is not how you behave if you are price-shopping or exploring alternatives.

Scale as a Qualification Test

60MW is not a pilot. Naver's AI Factory operates at 55MW as of mid-2026, with a stated target of 100MW. A tenant requesting 60MW places itself in the same capacity tier as one of Korea's largest internet infrastructure operators. At this scale, the specification narrows viable sites to a small number nationwide.

A tenant sourcing 60MW capacity is not selecting from a commodity market of interchangeable data centers. It is identifying a specific operator whose power infrastructure credentials can support institutional-scale AI compute — and then negotiating for the development to match those requirements. The request itself is the result of due diligence, not the beginning of it.

The Non-Purchasable Variable

Grid access is the binding constraint that capital alone cannot solve.

Interconnection queues in major markets are measured in years. Transformer procurement lead times extend beyond twelve months in tight supply conditions. Grid capacity allocation proceeds through regulatory processes that do not accelerate in response to developer urgency. A site without power infrastructure in place cannot purchase its way to readiness on a compressed timeline.

This creates structural differentiation. Operators who have resolved the power problem occupy a different tier than those who have not. The market of tenants approaching 60MW contracts is not selecting based on price alone — it is selecting based on which operators can actually deliver. A tenant choosing a specific site at this scale is, by implication, confirming that the operator has passed the power qualification test.

The moat is not the power itself. It is the years of infrastructure development and regulatory navigation that produced it — and which cannot be replicated by capital injection.

The Revenue Floor

On conservative assumptions — ₩3.5B to ₩4.0B per MW in annual capacity revenue — 60MW yields ₩210B to ₩240B per year as a floor. As of June 2026, that range does not include power resale income, which sits as an independent line item for operators with infrastructure configured for it.

The floor matters because it separates the capacity fee from the variable economics. Revenue from power resale fluctuates with grid conditions and pricing. The contracted capacity fee is fixed by agreement. When the two are valued separately, the floor establishes a minimum revenue baseline before any power economics are calculated.

At institutional capacity, the asset economics are visible well before the building is occupied.

How Anchor Tenants Compound

Securing a single major tenant at scale changes the risk profile of every subsequent transaction.

An operator who can demonstrate a verified tier-1 technology company as a 60MW anchor client changes the negotiating dynamic with every future prospective tenant. The due diligence question — "has this site supported production-scale AI workloads?" — is answered before it is asked. Underwriting risk for a new tenant drops because the reference case exists.

This is the compounding logic of infrastructure businesses. The first tenant validates the asset. The second assumes less development risk because the first exists. The third operates with lower uncertainty still. Each expansion cycle carries a lower risk premium as the reference list accumulates. Disclosed anchor identity accelerates this entire sequence — which is why tenant name matters beyond the revenue calculation.

The Disclosure Binary

The structure described above is sound regardless of the outcome at SGC Energy's June 9, 2026 investor briefing. The power moat exists. The demand-pull contract structure reduces development risk. The 60MW scale positions this project in the institutional AI infrastructure tier.

What the briefing determines is the timing of market recognition. If the tenant is confirmed as a major technology company on record, the re-rating mechanism activates — the reference effect begins, future expansion risk reprices, and the infrastructure moat is publicly validated. If the tenant is not disclosed or confirmed at the expected tier, the structural logic remains intact. The near-term catalyst does not.

Investor events at this stage exist precisely to resolve this binary. The underlying asset does not change. The timeline does.

Track infrastructure repricing in real time at Blackboard.