Know whether a site is viable before you commit serious engineering money.
Concept-stage engineering for datacenter sites, in days rather than weeks: feasibility, layout, sizing, and the document your investors actually read. Every claim carries its source, or is marked as an assumption.
Looking for five design partners.
Each partner receives one complete concept package at no charge: whether the site is feasible, what fits on it, and what it would take. Give me a site you're evaluating and you get it back within a week.
In return: detailed feedback, and if the work delivers, a short non-binding letter of intent confirming that for similar projects you'd consider Geometral as a commercial supplier. No purchase commitment, and no obligation if the work falls short.
Typical turnaround 5–7 days · Typical scope 25–100 MW campuses, greenfield or retrofit · Cost none for design partners
The point of concept stage
In the last package I delivered, the brief I was handed contained a heat-recovery concept that could not satisfy basic energy balances and a hall distribution 2.5× beyond ring-main practice. Both were caught on the first pass, before a line of design was committed to them.
That is what this stage is for: finding the expensive problems while they are still cheap. The package itself was nine deliverables for a 50 MW greenfield campus, covering site plan, power distribution, cooling, compute, balances, standards checklist and the investor-facing document.
Why days, not weeks
The drawings, the balances and the checks are generated and verified by tooling built for this stage, rather than redrawn by hand for every site. The engineering judgment stays human. The repetitive drafting, checking and traceability don't get done twice.
Delivered so far
- 50 MW greenfield campus, concept stage
- 9 engineering deliverables, each true to scale at A1
- 66 adversarial-review findings resolved before issue
- 12 open flags and 3 named deviations, stated rather than smoothed over
- ±15 % concept accuracy, 5–7 day turnaround
Read the full case → — the brief, what failed verification, all nine deliverables, and what the utility said afterwards.
What you get
- Site feasibility. Grid capacity and connection route, land, regulatory flags, cost order-of-magnitude. Every claim carries its source, or is marked as an assumption.
- Concept layout. Halls, power, cooling, routes and zones, drawn to scale on your actual parcel over real imagery.
- Engineering basis and sizing. IT load, power, cooling, water: each number traced back to the basis that produced it.
- The investor document, plus a live link they open in a browser. No install, no account.
- Known unknowns. What still needs to be verified before engineering begins, and where the concept departs from standard practice, and why.
What I need from you
A 45-minute kickoff call, whatever site documents you already have, and the occasional short question by email while I work. A handful, not a stream.
- Site boundary: cadastral extract, KML/DXF, or just an address and a rough outline
- Target IT load (MW), and the intended build phasing if you have one
- Grid situation: available capacity, connection point, correspondence with the utility
- Water and cooling constraints, if any are known
- Jurisdiction, and any planning constraints you're already aware of
- For a retrofit: the existing building's footprint and clear height
Whatever is missing gets flagged as an assumption, not guessed. A package built on gaps that are named is useful. One built on gaps that are hidden is dangerous.
Scope
Concept stage only (HOAI LP 1–2 / Pre-FEED): the work that happens before a licensed engineer stamps anything. It doesn't replace that signature, and it isn't construction documentation.
Who you'd be working with
I'm Mark. I do the engineering myself, and I built the tooling that makes it fast. Geometral is a one-person studio with software leverage, which is why turnaround is days rather than weeks and why the first packages cost nothing: I would rather prove the work than describe it.
Start
Send one site you're evaluating. Within a week you'll know whether it's worth pursuing. If the work proves useful, we talk about the next one. If it doesn't, you owe me nothing.
Send me a site