Is computing on your own power worth it? Independent feasibility and design for energy and compute
We answer that from your meter data rather than from assumptions, and we tell you when the answer is no. Every engagement opens with a paid scoping diagnostic, credited in full against the work that follows. Fixed scope, deliverables published in full, delivered by the principal.
Access Status
The discovery call is free and takes forty-five minutes. Melanie arranges it; fees are quoted privately afterwards, in US dollars.
- Discovery call
- Forty-five minutes, free, and an honest answer at the end of it — including when the answer is that there is no engagement here.
- The paid gate
- Every engagement opens with a paid scoping diagnostic, credited in full against the work that follows.
- Design, not build
- We produce feasibility studies, designs and build-readiness packs. Construction, procurement and operations belong to somebody else.
Who this is for
Three kinds of people write to us. If you are not one of them, the second column further down will tell you so faster than a call will.
You generate your own power
A solar array, a hybrid plant, a site that spills surplus it cannot sell. You want to know whether a computing load is a sensible thing to point at that surplus, what size it should be, and what it earns on the hours it actually runs — not on the hours a spreadsheet assumes.
You run computing that could flex
Some of your work is latency-bound and some of it genuinely is not. You suspect the second category is worth something — to a utility, a generator, or your own bill — and you need that suspicion turned into a defensible number before anyone commits capital.
You pay people for measured work
A pool, a co-operative, a shared facility — anywhere a participant’s share is computed from a measurement. You want to know whether that measurement is where you think it is, and whether a participant could check your arithmetic without taking your word for it.
How an engagement runs
Four steps, in this order. The second one is a gate, not a formality — nothing starts without it.
Discovery call — 45 minutes, free
You describe the site or the workload. We tell you whether there is an engagement here, including when there is not. No deck, no follow-up sequence, no charge.
Scoping diagnostic — 2 to 3 weeks, paid
The gate. We build the energy and load profile of the site or workload as it actually is — from your meter data, bills and telemetry, not from assumptions.
You get a go, no-go, or go-differently verdict with the reasoning shown. If the answer is no, that is the deliverable. The fee is credited in full against an engagement commissioned within ninety days.
The engagement — 6 to 10 weeks
One of the three below, or the settlement review. Scope is fixed before it starts, the deliverables are named in the contract exactly as they are named on this page, and the principal does the work.
Afterwards — optional
An advisory retainer if you want a hand on the tiller through implementation, with a six-month minimum. Or nothing at all: most engagements end with a document and a contractor who can price it.
Fees are quoted privately, in US dollars, after the discovery call. We do not publish rates, day rates, or percentages — the shape of the work is on this page in full, and the number depends on your site.
The three engagements
Open one to read what you actually receive. These are the same words that go in the contract.
Flexible-load and utilisation assessment for inference infrastructure
For operators whose computing could be a controllable resource, and who need an honest number on what that flexibility is worth. 6–10 weeks.
Flexible-load and utilisation assessment for inference infrastructure
For operators whose computing could be a controllable resource, and who need an honest number on what that flexibility is worth. 6–10 weeks.
- Telemetry and metering architecture — what to measure, where the boundary sits, and how it reaches you.
- A utilisation and deferability model — which of your work is latency-bound and which is not.
- Controllable-load architecture — how load is shed, ramped and verified, and the failure modes when it is not.
- A measurement and verification plan naming IPMVP Option B or C, with the baseline, the normalisation and the non-routine-adjustment process defined — so that any future savings claim is measurable by someone other than us.
- A costed implementation plan with a build-versus-buy split.
- Excluded, in the contract: any efficiency or savings guarantee, operations, thermal design, and any representation about interconnection or utility-programme qualification.
Renewable-powered compute design
Design a computing load that fits the generation you have, rather than a generation asset that fits a load somebody assumed. 6–10 weeks.
Renewable-powered compute design
Design a computing load that fits the generation you have, rather than a generation asset that fits a load somebody assumed. 6–10 weeks.
- Generation curve against proposed compute curve, hour by hour, with surplus and deficit quantified.
- Sizing — what computing the generation can actually carry, and the honest utilisation factor rather than the flattering one.
- Surplus-absorption design — what runs when generation exceeds demand, and what stops when it does not.
- The storage-versus-flexible-load trade-off, with the arithmetic shown rather than asserted.
- A capital and operating model with sensitivities, and a named list of what would kill the project.
Solar-hub siting and design for edge compute
Our focus. Site feasibility and design for a solar-powered hub that hosts computing. Design and feasibility only — we do not build it. 8–12 weeks for a first site.
Solar-hub siting and design for edge compute
Our focus. Site feasibility and design for a solar-powered hub that hosts computing. Design and feasibility only — we do not build it. 8–12 weeks for a first site.
- Site selection and resource assessment — irradiance, land, access, security, connectivity.
- Load design — what the hub hosts, what it serves, and what it earns.
- Energy architecture — array, storage, controllable load, protection.
- Thermal and environmental factors stated as constraints, with a specialist sub-consultant identified wherever design liability attaches.
- Capital cost, operating cost, tariff and revenue model, and the operating assumptions the whole thing rests on.
- A build-readiness pack a third-party contractor can price.
Settlement and metering correctness review
An addition to any of the three, or a standalone piece. For anyone paying participants on the basis of measured work. 3–5 weeks.
Settlement and metering correctness review
An addition to any of the three, or a standalone piece. For anyone paying participants on the basis of measured work. 3–5 weeks.
- Where your measurement boundary actually sits, as opposed to where the documentation says it sits.
- How work is accounted and paid, and whether a participant can reconcile their own share independently.
- The failure cases — what happens to the accounting when a meter drops out, a clock drifts, or a participant disconnects mid-interval.
- A remediation list, ordered by how much money each item can move.
What we do, and what we don’t
The right-hand column is here because it is the fastest way for you to find out we are not your firm.
What we can back
Everything we sell is downstream of infrastructure we wrote ourselves and can show you.
- A mining pool written in C with our own payout engine, a differential test harness that compares it against a reference implementation, and end-to-end tests on a private test network.
- An access gateway of roughly eighteen thousand lines of Python: one-time-code sign-in, roles and entitlements, audit logging, payout reconciliation, and mutual-TLS on the payout path.
- A public explorer that miners opt into — so what we say about measurement is visible rather than described.
- Solar installation delivered and documented in East Africa, at a tea factory and at health centres. A Boston firm with Kenyan roots, delivering in the conditions we came from.
That list is the ground the third engagement stands on. The solar was delivered in the field; the settlement infrastructure was built in-house — the layers a solar hub for edge compute is designed from. Designing yours, siting and feasibility first, is exactly what the engagement is for.
What we don’t do
Where your problem needs one of these, we name the constraint and identify a specialist. We do not absorb the scope.
- Thermal or computational-fluid-dynamics modelling.
- Transformer sizing, power-distribution selection, or electrical safety systems. These carry professional design liability and belong to a registered engineer.
- Operations, firmware tuning, or fleet procedures.
- Hardware procurement, or any claim of buying leverage.
- Power purchase agreements, utility interconnection, or curtailment operated against a grid operator.
- Uptime or savings guarantees.
- Building the hub. We design it and hand it to someone who does.
We also do not sell scale we do not have. Ask us what we run and we will tell you plainly; it is a correctness testbed, not an energy asset, and our value is in the correctness.
Start with the call
Forty-five minutes, free, and an honest answer at the end of it — including when the honest answer is that there is no engagement here.
Melanie takes the details and arranges the call: what you have, what you are contemplating, and roughly what you have to spend on finding out. One verified path, no separate inbox. Fees are quoted privately afterwards, in US dollars.
Why is the first step paid?
Because the honest version of it costs real weeks. Reading two years of meter data, reconciling it against bills and telemetry, and working out what a site can actually carry is the work. It is not a sales activity dressed as one.
A free version would be either unfunded, and therefore shallow, or funded by the engagement it recommends, and therefore incapable of recommending against itself.
The credit is the point. Commission the engagement within ninety days and the diagnostic fee comes off it in full, so you have paid nothing extra for the option to walk away. If we tell you not to proceed, you have bought a decision, and it was cheaper than the alternative.
Why aren’t your rates on this page?
Because the number depends on your site, and a published range would mislead in both directions at once. What we can publish without qualification is the shape: how long each step takes, exactly what you receive, who does the work, and what we refuse to do. All of that is on this page in full.
Ask on the call and you will get a figure, in US dollars, without a negotiation ritual attached. We quote in dollars only — one currency, wherever the work happens.
Can you deliver in East Africa?
We are a Boston firm with Kenyan roots and field experience there, and that is a large part of why the solar-hub engagement exists at all. Design and feasibility work travels: the study, the model and the build-readiness pack are produced by us wherever the site is.
Work that constitutes professional engineering services executed on the ground may require local registration and a signing registered engineer. Where that applies we say so before you commission anything, and we bring in a locally registered practice rather than quietly writing around the requirement.
How big an operation are you?
Small, and the work is delivered by the principal rather than handed down a bench. That is a genuine constraint on how many engagements run at once, and it is the reason there is a diagnostic gate rather than a pipeline of proposals.
It also means we do not sell scale. What we run ourselves is a correctness testbed, not an energy asset, and we will describe it in exactly those terms if you ask.
What is on offer is that the settlement and measurement infrastructure underneath it was built here, by the people who would be advising you.
