the acre next to the substation

In 2019 a Meta affiliate paid about $51 million for 414 acres in Licking County, Ohio — roughly $123,000 an acre. In 2025 the same affiliate paid $566 million for 514 acres in the same county — more than $1.1 million an acre. Farmland in Licking County trades for $10,000 to $15,000 an acre.

That second number is what people mean when they search data center land prices. It is a real price and it is fully earned. It is also not what the land is worth. It is what it costs to make the land stop being a farm by a certain date. Those are different facts, and this post holds both.

what data center land prices actually contain

what are data center land prices?

Data center land prices are what buyers pay for acres where the electricity equation is solved, or solvable on a schedule — interconnection, substation capacity, entitlements, fiber — under a capex deadline. In the 2025–2026 Ohio and Virginia corridors they ran from roughly $350,000 to $3.7 million an acre, against farmland at $10,000–15,000. The premium is for time to power, not dirt.

The acre is the delivery vehicle. The thing being bought is time. The pillar post, what powered land actually is, lays out the layers that make land power-ready; this post follows the price those layers command.

That is why the same county can carry a $12,000 acre and a $1.1 million acre a few miles apart. The expensive one sits where a buyer already has interconnection, a substation path, zoning, and fiber either in hand or within reach. The cheap one has a transmission line in the view and nothing else. The sibling post a map line is not powered land covers that screen; the short version is that proximity is not deliverability.

Three recent comps, with what each price actually bought:

The Devlin row is the clearest case. The land cost $51 million. The homebuilder won a residential rezoning, then went back and won a second rezoning to data centers with three substations on the plan, then held that entitlement through a two-year court fight. What Amazon paid $700 million for was the second rezoning and the substations — the manufactured right to put watts on that ground. None of that is in the soil.

the buyer who manufactured the watts

Hold the Ohio comp up against the series spine — a map line is not powered land — and it looks like a contradiction. In 2019 Meta bought proximity: acreage in a business park with transmission in the area, not a campus with a signed interconnection for gigawatts. Then it manufactured deliverability. AEP Ohio began evaluating new 345 kV routes into New Albany in early 2023, announced the two-line Vassell–Green Chapel project in December 2023, and won siting approval for the first line in June 2024, with construction running into 2027. In parallel the Ohio Power Siting Board approved three behind-the-meter gas plants for the Meta affiliate — 200 MW in June 2025, 200 MW in September 2025, and 250 MW with 116 MW of battery storage in June 2026 — none connected to the grid, all dedicated to the campus.

At hyperscaler scale the buyer can make the watts show up. That is what turned the 2019 acre into the 2025 acre. It is also why a landowner cannot read the comp as a promise. Nobody is going to site a gas plant behind your barn, and the utility's transmission plan was drawn for a customer that had signed for hundreds of megawatts, not for the parcel next door. Proximity became deliverability in Licking County because one buyer paid to make it so. Everywhere else, it is still a line on a map.

the clock, not the worth

Here is the part a land-price search will not tell you. A conversion bid is a clock. It says: by this date, on this parcel, the highest and best use flips from growing something to housing compute, and the price is what it takes to make that happen on schedule.

The farm is not worth $1.1 million because Meta paid it. The farm was already there — and honesty about what it is matters. Most of that corridor is tile-drained row crop: a working stock that has been simplified for a century, carrying less of its original soil carbon, hydrology, and habitat than the prairie and woodland it replaced. Degraded, and still real. It grows food, holds and slows water, and carries the option of being restored toward something richer. A pad does not degrade that option further. It ends it. Concrete, fill, and a fenced substation yard are the one land use from which there is no path back to a field.

Nobody in that transaction is the villain. The family that sells is doing arithmetic any of us would do at seventy times the ag value. The buyer is paying for months, because in this market months are the scarce input. Both are rational. What neither of them is paid to notice is which option closes when the clock runs out — and that is the only part of the deal that does not come back.

why the first quarter of 2026 was scarcity of power, not dirt

Sales of land earmarked for data centers came to roughly $3.3 billion in the first three months of 2026 — up 141% from the same quarter a year earlier, according to Avison Young data reported by Bisnow. Data center sites were about 30% of all development-site dollars, up from about 19% in 2025.

The number of data center sites sold in that quarter fell — from 20 to 16.

More money, fewer deals. That is not a country running out of acres. That is a country running out of acres with watts. Power-ready parcels — the ones with a substation path and a queue position that a utility engineer will sign — are the constrained supply, and their scarcity is what shows up as a land price. In key US markets the generation and transmission upgrades needed to serve a large new load take five to ten years; the price of an acre that skips most of that wait is the price of the wait. Data centers are the loudest buyer of watts as a real-estate attribute, not the only one — battery plants, advanced manufacturing, and fleet charging sit in the same queues at smaller scale.

Nothing on this page moves a queue. Interconnection timing is a problem for the utility, the ISO (the independent system operator that runs the regional grid and its queue), and FERC, and pretending otherwise would be the fastest way to lose the engineer in the room. What the price signal does tell a landowner and an assembler alike is where the clock is loudest.

the multiple depends on what the acre was doing

Our land-pressure work uses a working range of 50–350× agricultural value for parcels that clear the powered-land screen near active corridors. It is a range from corridor comps, not an appraisal rule, and no comp in this post reaches the top of it. The Ohio parcels sit inside it: $1.1 million against $10,000–15,000 farmland is roughly 73–110×. The Virginia deals do not compute against farmland at all — one was a campus, the other was zoned for houses before it was zoned for halls — which is its own lesson. The multiple depends entirely on what the acre was doing the day before the bid arrived.

For a landowner, the range means two things at once. The neighbor's comp raises your assessment, your heirs' expectations, and every future conversation about the place. And the comp is only real for your acre if the watts are — a corridor price on a parcel with no power path is a rumor with a decimal point.

two land banks, same acre

Powered-land banking and nature land banking are the same trade with opposite exits. Both acquire or control acres before the market prices in what they can become. Both carry the land through a long, mostly yield-less hold. They differ on what "done" looks like.

Nature land banking here means acquiring or controlling land for its living function ahead of conversion. It is not mitigation banking — the regulated trade in wetland and stream credits sold to offset permitted impacts somewhere else. The two can touch; they are not the same instrument.

The shared row is the whole argument. The powered-land bid teaches the living side the same lesson it teaches the assembler: whoever is not at the table before the substation study is done will pay the post-study price or lose the acre. The nature side's advantage is that it does not need the whole parcel. It needs the living half funded before the pad half closes.

if you own the acre

The bid in your mailbox is a clock, so treat it like one.

One constraint first. You do not choose the pad; the buyer's site plan and the county's setbacks, stormwater rules, and buffer requirements do. A campus needs contiguous graded ground plus room for substations, laydown, and security setbacks, and a carve-out that breaks the plan can kill the deal rather than shrink it. Know what the buyer actually needs before deciding what you want to keep.

Three honest paths, none of them the moral of this post:

Sell the pad, keep the living acres. Most campuses need a fraction of a working parcel graded and served. The floodplain, the woodlot, the field the pad does not touch, and the drainage that keeps the pad dry are the remainder. They can stay in your hands or a land trust's, and they can be funded rather than left as leftover.

Option only what the buyer will pad. An option on, say, 120 acres of a 400-acre farm prices the conversion clock on the part that converts and leaves the rest in production — with the option premium carrying the place while the queue does its work.

Hold. Some families will, and the comp next door will not make that wrong. What it will do is raise the cost of holding, so the living acres should be earning something during the hold.

The farm spoke goes deep on the protection ratio, working-lands requirements, and the conservation easement as the legal instrument that keeps preserved acres in production. One paragraph here is enough: a campus can be structured so that the acres it develops fund the permanent protection of many more nearby, and the community can require that ratio before it says yes.

A plain gloss before we go further. ensurance funds a named place now, before it is lost — the opposite of insurance, which pays after. The mechanism for the acres you keep is concrete and, today, small. First, the remainder is identified on the plat — the floodplain, the woodlot, the field outside the pad. Second, it is held by you or by a land trust under whatever legal instrument fits. Third, it is routed to a named agent — for working ground, rural-open-space.ensurance; for a wet field or a creek margin, inland-wetlands.ensurance — an onchain account whose certificates are funded commitments to that kind of place. Fourth, certificates on that agent fund protection and stewardship of the remainder at current volumes, which are small. There is no per-parcel syndicate waiting behind this, and a certificate is not a $1 million-an-acre offer and never will be. It is how the living half of a parcel gets funded while the pad half gets paid for.

if you buy the option

For the assembler, the loss runs the other way. You have already priced the failure cases — the denied conditional use permit (CUP), the interconnection study that comes back at nine figures, the county that rotates you to Texas. What most powered-land underwriting has not priced is the remainder: the roughly 80 acres at Devlin that Amazon did not buy, set aside as open space in the rezoning proffers according to reports — which means planned on paper, unfunded, and with no one named to steward it. Land that is planned but unfunded is land somebody else will plan for, usually at the hearing.

Price the remainder in. Decide before the entitlement vote what the living half of the site does, who holds it, and what funds it. A remainder with a funded plan is a hearing exhibit; a remainder with a proffer and no money is a rezoning risk. The opposition series covers how enforceable commitments move approvals; the point here is narrower — the same acre that carries your power option also carries a wetland, a field, or a recharge zone, and the market only priced one of them.

What we do, stated plainly: we find, option, entitle, and broker infrastructure land — including power-adjacent tracts — and route the natural remainder to funded protection through ensurance. We do not build hyperscale pads, construct interconnection, or run a powered-land fund. Certificate volumes are small today. The living agents are live: rural-open-space.ensurance, inland-wetlands.ensurance, and the basin-level accounts behind them. If you underwrite powered land, the infrastructure-investor path is where that conversation starts.

next

If you are the landowner, start with the landowner path — it lays out the pad-versus-remainder decision before anyone is bidding.

If the farm is the question, data centers don't have to eat the farm is where the protection ratio lives.

If you want the remainder as an asset rather than a leftover, the last post in this series is keep the living acres.