Compute center deals sized to get approved.
We bring you standalone sites that pay their own way on power and meet the checks towns are writing into their rules.
Communities are blocking large data center campuses. Opposition now decides which projects get built.
Contested projects fail about half the time, and banks now weigh community sentiment as a credit factor. Most new local bans target sites over 20 to 25 MW, which leaves room for standalone sites below that line.
Projected global use in 2030, up from 415 TWh in 2024. The US accounts for almost half of the growth.
IEA, Energy and AI (2025)
Inference is moving to regional sites built where the power is. EPRI and Nvidia are piloting 5 to 20 MW sites at substations with spare capacity, and CBRE sees urban infill of 2 to 20 MW gaining momentum.
IEEE Spectrum on the EPRI pilots; CBRE, North America Data Center Trends H1 2026
Of contested data center projects canceled in 2026. Banks now weigh community sentiment as a credit factor.
Heatmap; Reuters, Aug 2026
Criteria we work to
Community opposition is now a credit risk. These are the criteria we work through with each operator before an introduction, each one something a third party can check.
Sources: Gallup, Heatmap, Dominion Energy and local ordinances, 2026.
| The site | Why it matters |
|---|---|
| Sized to substation headroom and local rules, typically 5 to 15 MW, and not split from a larger campus | Most new bans target sites over 20 to 25 MW, and splitting campuses invites tighter rules. |
| Uses verified spare capacity at an existing substation, with no new transmission lines | No new grid build that lands on local bills. Above 15 MW, only where the substation proves it. |
| Pays its own interconnection and grid upgrade costs upfront, and signs a contracted minimum demand with the utility | Virginia's large-load class starts at 25 MW. Below it, standard utility terms still allow both, so the site can choose to pay its own way. |
| Closed-loop or non-evaporative cooling | Keeps water off the list of objections. |
| Grid and storage, with disclosed backup at every site; fuel and runtime follow local permits | Compute needs firm backup, and towns accept it best when it is sized, permitted and disclosed. |
| Set back from homes, schools and parks, with noise limits | Distance triggers restrictions even for sites below the size line. |
| Public disclosure before rezoning or the first hearing | Secrecy turned land-use disputes into recalls in 2026. |
| No special tax abatement | Tax breaks with few permanent jobs drive recalls and repeal efforts. |
The largest data center market in the US, inside PJM where data center load now drives power bills. Loudoun, Prince William and Culpeper have moved data centers from by-right approval to public hearings, so every site we work on plans for that. It is also where we are already working with communities on resilience.
We are piloting a resilience evaluation layer, which turns this checklist into a scored record, and a resilience investment layer, so a site's benefit to the community can be assessed and financed alongside the power.
What you get
We only bring you operators that meet the criteria you give us, with the answers to your first questions already in hand.
Mid-size and edge operators need mid-market amounts on shorter terms than the hyperscale deals in the market today.
When an operator pays on a first tranche, the next one goes to a borrower with a payment record.
Before an introduction, we work through the criteria towns now write into their rules: size, power, water, setbacks and disclosure.
These questions come from a team that ran supply chain risk and built ESG ratings for 9 million private companies at Dun & Bradstreet.
- Customers and contracts: who buys the compute, at what price and for how long
- Financials: revenue, burn, current debt and what it costs
- Equipment and site: chips on hand or ordered, lease or ownership, power
- Team: operating history of the people running the sites
- Community fit: size line, power source, cooling, setbacks and disclosure
- Credit support: guarantees, reserves and collateral the operator can offer
How it works
- 1Book a call
Twenty minutes on what you fund, your ticket size and terms, and what rules a deal out.
- 2We qualify operators
We work through each operator's contracts, financials, equipment and credit support.
- 3We introduce the ones that fit
You see the operators that match your criteria, with what we found.
- 4Close and follow on
We support the deal through close and the next tranche.
Who it fits
- Private credit and infrastructure funds that want data center exposure
- Equipment lenders financing compute and power equipment
- Family offices that fund energy and digital infrastructure
- Capital partners who prefer mid-market tickets and shorter terms
- Mandates limited to investment-grade borrowers
- Mandates that only take hyperscale deals
Questions capital partners ask
- Does Zero Circle lend alongside us?
- No. We bring you operators that meet your criteria. The credit decision stays with you.
- Do you finance GPUs?
- We lead with power and with revenue backed by customer contracts. GPU-only tranches are outside what we underwrite today.
- How does Zero Circle get paid?
- The operator pays us: a short retainer, and a fee tied to the close. We cover the details on the call.
- Have you closed a compute deal?
- Not yet. Zero Circle helps mid-market energy projects get to financing, and compute is new for us. The first introductions will come from the operators we are speaking with.
Tell us what you fund.
Twenty minutes on your ticket size, your terms and what rules a deal out.
Or write to contact@zerocircle.eco.
