Ask who buys powered land and most people say hyperscalers. That is the last buyer in the chain, not the first one.
Powered land investment is capital deployed to bring land to, or hold land at, a solved electricity equation before anything goes vertical — interconnection position, substation capacity, entitlements, fiber. The name on the final deed is usually a hyperscaler or an operator. The party that absorbs the first three to seven years of queue risk, deposit risk, and county-meeting risk is almost never them.
That gap is the business. Hyperscalers want shovel-ready. Someone else does the utility politics.
who is actually at the table
Powered land has at least six distinct classes of participant, and they enter at different points on the same curve. Sorted by when they show up:
| who | what they buy | what they are really taking on | when they show up |
|---|---|---|---|
| Hyperscalers | Shovel-ready sites, campuses, sometimes the developer itself | Compute demand risk — not entitlement risk | Late, or very early via acquisition |
| CRE assemblers and brokerages | Fragmented parcels near transmission, plus the local relationships | Assembly risk, option and carry cost, seller psychology | First |
| Infrastructure PE and private credit | Land plus power preparation, underwritten as infrastructure | Queue risk, upgrade cost allocation, hold-period demand risk | Early to middle |
| Utilities and ISOs | Nothing — they gate | Reliability, upgrade cost causation, ratepayer exposure | Continuously, and decisively |
| Parcel intelligence vendors | Data, not dirt | Screening accuracy | Before anyone |
| Counties, communities, economic development | Nothing — they approve or they don't, then host the load for thirty years | Tax base, water, noise, traffic, farmland identity, for the life of the campus | At the hearing, which is where deals actually die, and every year after it |
Two rows in that table hold veto power and neither one is a buyer. That is the structural fact most powered-land underwriting still treats as a footnote.
The scale on the demand side is real and worth stating precisely. Amazon, Alphabet, Meta, and Microsoft have guided to roughly $700–750 billion of combined 2026 capital expenditure, depending on how you align Microsoft's fiscal year with the calendar and how you treat its reclassification of certain leases off the capex line. That is the pull. It is not a promise that any specific parcel gets energized on schedule.
the seven plays
Powered land investment is not mysterious once you see it as seven recognizable strategies, distinguished mostly by how much of the electricity problem you agree to own.
| # | play | what you actually do | hold | typical exit | capital intensity |
|---|---|---|---|---|---|
| 1 | Power-ready land banking | Acquire or option before the utility confirms capacity; control the optionality early | 3–7+ years | Sale to hyperscaler, operator, or powered-land developer | Medium — carry plus interconnection deposits |
| 2 | Entitlement and interconnection | Buy raw land; advance zoning, environmental review, queue position, substation design | 3–7 years | Sale of an entitled, interconnected site at a premium | High — legal, engineering, community work |
| 3 | Powered shell / pad-ready | Deliver graded pads, power stub-ups, fiber, minimal vertical | 5–10 years | Sale to tenant or operator | Very high |
| 4 | PropCo land, OpCo operator | Land entity holds the dirt and the power rights; operator builds and leases | Long | Ground rent plus appreciation | Split by joint venture |
| 5 | Pre-leased campus finance | Secure an anchor tenant before or during construction, then finance against the contract | 7–15 years | Bonds or securitization against contracted cash flow | Very high |
| 6 | Hybrid power arbitrage | Site where on-site generation and storage unlock interconnection years sooner | 3–5 years | Powered land sale with generation in place | High — generation and storage capex |
| 7 | Geographic rotation | Move out of opposition-heavy markets into pro-growth jurisdictions | Varies | Entitled land sale in the new corridor | Medium |
Plays 1 and 2 are speculative land banking with an infrastructure underwriting screen bolted on. Plays 3 through 6 are infrastructure development wearing a real estate coat. Play 7 is an admission that the hearing is a real risk and the cheapest mitigation is a different county.
Every one of these carries the same unpriced item: the land is not only a pad. It is a watershed, a working farm, a remainder that stays alive whether or not the interconnect closes.
buying the pipeline instead of the parcel
The clearest 2026 signal that hyperscalers are no longer content to be the last buyer: Alphabet closed its acquisition of Intersect Power on March 10, 2026, for $4.75 billion in cash plus assumption of debt — a deal announced December 22, 2025. Google took Intersect's digital power business, the part built to co-locate data center load with new generation and storage. The grid-tied clean energy assets were spun out into a separate independent power producer, IPX Power, with majority backing from TPG Rise Climate. Across both transactions, roughly $12 billion of enterprise value changed shape.
Read that as a capital-stack move, not a news item. Alphabet did not buy a site. It bought the capability that produces sites — origination, generation, interconnection sequencing. When the terminal buyer acquires the pipeline, the middle of the stack gets compressed, and the returns available to a pure "buy dirt near a line and wait" strategy get thinner.
when the bond market shows up
The second 2026 signal is on the debt side. In April 2026, Meridian Arc HoldCo — an entity under a joint venture between Fluidstack and Next Frontier, the powered-land venture reported to have been launched by Coatue Management — priced $5.7 billion of 6.25% senior secured notes due 2031. Proceeds fund the acquisition of roughly 140 acres in New Lebanon, Sullivan County, Indiana, two turnkey data centers totaling 430 MW of critical IT capacity, and an on-site electrical substation. Capacity is 100% leased to Fluidstack on a 15-year triple-net base term, and Fluidstack's lease obligations are guaranteed by Google. Fitch assigned a BB rating with a Stable Outlook on May 14, 2026, noting roughly 92% debt-to-cost against about $505 million of sponsor equity at closing, an aggressive construction schedule, no guaranteed maximum price contract, and limited data center experience at both sponsor and contractor. According to LCD it was the largest AI-driven financing completed in the high-yield market to date as of that April 2026 pricing.
Two things are worth extracting. First, 430 MW across about 140 acres works out near three megawatts per acre, which sits inside the 3–4 MW per acre planning ratio that Hines Research and CBRE Investment Management publish. Two caveats on that arithmetic: the ratio is a planning heuristic rather than a design spec, and 430 MW here is critical IT load, not utility interconnection capacity — the grid-side number a utility has to serve is larger. Acreage is a poor proxy for either.
Second, and more important: a pre-leased, credit-guaranteed campus can now raise high-yield debt at real scale and be analyzed with project-finance tools — Fitch rated it under its infrastructure and project finance criteria. That is not the same as being project finance. Fitch was explicit that the financing terms and debt provisions are weaker than typical project finance protections, citing merger and joint-venture flexibility without rating affirmation, reinvestment latitude, and the option to regear loan-to-cost from 92% up to 95% after commencement. The signal is that lenders will now underwrite contracted, shovel-ready capacity against a predictable takeout instead of treating it as pure speculation. The signal is not that the risk was engineered out.
That is a signal, not a product we sell. We do not underwrite, arrange, place, or participate in that debt. We are reading the same tape you are, and the tape says the market is separating sites with a solved electricity equation from sites with a nice map.
what none of that capital is buying
Run the seven plays end to end and notice what never appears as a line item. Every play prices megawatts, acres, entitlement progress, and takeout. None of them prices the half of the parcel that will never hold a pad — the riparian edge, the recharge area, the working ground on the far side of the setback, the basin the cooling plan draws from.
That half does not disappear when the highest and best use flips to data center. It keeps producing clean water, flood attenuation, and habitat for the same county that has to vote on the project. It is also, quite often, the thing the hearing is actually about.
We are not Coatue. We do not run a multibillion-dollar vehicle buying land next to power, and there is no powered-land fund here to pitch you. If what you want is the pad, the campus, or the interconnect construction, the honest referral is to the firms that named the asset class and staff that work — CBRE Investment Management, Hines, JLL, and brokerages with real data center practices.
We broker land. We hold the other half of the same parcel.
the remainder as an entitlement variable
Powered-land underwriting usually books three variables: power deliverability, development execution, and commercial durability. There is a fourth that shows up as a schedule risk and gets treated as a soft cost — whether the community grants a license to operate.
Projects with enforceable, measurable, funded commitments on water, noise, and farmland tend to clear. Projects with a slide deck of promises tend not to. That is not our opinion about civic virtue; it is a pattern in approvals, and we have written the mechanics of it once already rather than repeat them here: how to get a data center approved faster.
The instrument side takes two lines, because the distinction matters. In ensurance, a certificate funds a named natural asset directly — a specific wetland, a specific stretch of rural open space — and because it is held onchain, the funding is checkable rather than merely announced.
A certificate can be attached to a condition the county records. It does not become that condition. The enforceable instrument stays the county's: the permit condition, the development agreement, the recorded covenant. What the certificate adds is evidence that the money behind the condition actually moved, which is the part a promise cannot supply.
Our own stage, plainly: 26 certificates are listed on /specific as of September 2026, and live agents include rural-open-space.ensurance, inland-wetlands.ensurance, and water-abundance.ensurance. Small volumes. We find, broker, option, entitle, and subdivide infrastructure land and fund the living remainder through those agents. We do not build hyperscale pads, arrange interconnection construction, move a queue position, or lower anyone's wholesale power cost. Queue reform and cost allocation are FERC, ISO, and utility problems. We answer the living-system half.
what is in it for each seat
| if you are | the powered-land question you are already asking | the half this adds |
|---|---|---|
| Infrastructure PE or a land assembler | Can I energize on schedule and exit into demand? | The remainder is a schedule variable, not a donation. Funded commitments are the pattern that clears, and the remainder is where the funding has to land. |
| Private credit or a capital provider | Is the takeout contracted and the basis defensible? | Water and community commitments that are funded rather than promised are diligence-grade, not reputational. |
| A data center developer or operator | How do I stop losing years in one county? | A checkable commitment on the living half of the site you are already buying. |
| A generalist allocator | Is powered land an asset class or a moment? | The living systems on and beside these corridors are the durable position underneath a cyclical one. |
Powered land is a subset of infrastructure land, not the whole of it. Batteries, advanced manufacturing, fleet charging, cold storage, and industrial outdoor storage all buy power as a real estate attribute. Data centers are the loudest buyer in 2026, not the only one — and every one of those uses sits on a parcel with a living side.
next step
One conversation, about one named corridor. Not a fund pitch.
If you allocate to powered land or infrastructure land, bring a corridor you are already looking at and we will tell you what the living half is worth funding and how a commitment on it becomes checkable. If the answer is "nothing here," we will say so.
- infrastructure investors → — how the remainder fits an infrastructure underwriting screen
- capital providers → — blended and yield-seeking structures for the living half
- talk about a specific corridor → — bring the parcel, we will bring the basin