Does the nearby acreage bring its own power?
When I saw the article that Georgia Power announced the approval of OpenAI’s Effingham County contract in August. It seemed the project expected approximately 3,200 MW of demand, with up to 1,000 MW covered by demand response under a 25-year agreement. That demand response allows the utility to reduce electricity delivery during specified periods, covering at most around 31% of expected demand here. Project announcement · Contract approval
If you’re buying land nearby, you still need to know how much power your parcel can get and when it can arrive. So how can you do that?
Start with the serving utility’s large-load or economic-development team. Give them the parcel location, the load you’re planning, when you need it and whether your operation can handle interruptions. Ask which load studies are required, how long they take and who pays for the upgrades. Try to get the findings in writing before using them to justify the purchase price.
What happens when power is interrupted?. What happens to the computing work when delivery is reduced? Can it wait, can it move somewhere else, or does the operator have customers expecting the service to be available?
Those answers change which power agreements a developer can accept. Expect flexibility to matter more as the Georgia pipeline grows, with continuous load bearing facilities competing for a narrower set of suitable sites. How often reductions can occur, how long can they last and the notice required will determine whether an agreement fits the workload.
Now let’s analyze the electricity bill into the picture when comparing land:
Assume 40 MW IT facility at 1.25 PUE. PUE measures total facility electricity against electricity used by the computing equipment, so this example needs 50 MW at full load. At 90% utilization, holding PUE constant and consumption proportional to utilization, annual usage reaches about 394,200 MWh.
Assume one site costs $10/MWh more for electricity. You’re looking at another $3.94 million a year. A $10 million discount on the land disappears in roughly two and a half years of additional energy expense, before discounting cash flows.
You can probably negotiate a good price on the acreage and still end up with a bad deal. When comparing both sites using the same load, utilization and delivery requirements, you should include the utility costs each actually carries, part of the reason one looks cheaper may be that its costs may be underpriced.
But wouldn’t a campus that size bring infrastructure into the area or be required to build its own substation among other factors?
It could, but the next question is how much capacity remains available to another customer once the campus is served. A new substation or transmission upgrade may support further development, but the utility woul need to establish what that means for your parcel and proposed load and that takes time.
Expect buyer interest to concentrate around sites that can document service, manageable upgrade costs and delivery dates that fit a tenant’s requirements. Acreage marketed mainly on proximity will face more scrutiny as buyers move from browsing listings to financing construction.
If the seller wants to be paid today for power access that still needs to be established, you’re taking on that uncertainty. Keep the unconfirmed benefit out of the base valuation and negotiate purchase terms that account for the outstanding utility work.
Before agreeing to the premium, ask the utility for a written response on supportable load, the delivery schedule, any curtailment conditions and the costs assigned to the buyer. Put those answers beside the asking price, then decide whether the site works.
I put together a dashboard tracking all DC’s on Liquid Labor tracker and the supporting filings on the power research dashboard.
Uwe Cerron
https://liquid-labor.com/