Strategy
Compute belongs where the demand is.
The thesis is not that AI demand will grow — everyone underwrites that. It is that the value of a unit of compute depends on where it sits, and that a uniform, metro-resident fabric can be financed, built and sold as a repeatable unit.
The thesis
Four propositions the model rests on
01
Inference moved to the edge
Training concentrates in a few very large campuses. Inference does the opposite: it follows the user, and every millisecond of round-trip is a product constraint. Capacity inside the metro is worth more than capacity far from it.
02
The center is capital-constrained
Hyperscale build queues are measured in years and priced accordingly. A distributed fabric of smaller, identical sites can be financed and delivered in parallel rather than sequentially.
03
Sovereignty is a requirement, not a preference
Regulated buyers increasingly need workloads to stay inside a defined jurisdiction, on identifiable hardware, under a known operator. That is a structural property of where the compute sits.
04
Uniformity is the financial product
One architecture repeated 101 times turns infrastructure into an underwritable unit: known bill of materials, known capacity, one seat price, one waterfall.
Commercialization
Capital to revenue, in six steps
Capacity is never sold before it exists. Each step gates the next.
- 01CapitalizeEach Hub LLC assembles its own capital against its own objective, clearing the 70% and 85% gates in order.
- 02DeployLong-lead equipment is released, sites are prepared and the nine-plane stack is racked to the same specification everywhere.
- 03CommissionNetworking, software and acceptance testing bring each site to a defined ready state before any capacity is sold.
- 04BenchmarkFounding and pilot workloads establish real performance and utilization figures on the deployed fabric.
- 05Open capacitySellable seats — 50% of built capacity — are made available to third parties at the modeled seat price.
- 06Sell & scaleChannel and direct enterprise demand fill capacity; utilization drives the Monthly Distribution Amount.
The sequence describes intended operating steps, not a schedule. No step is complete today: the fabric is planned and pre-revenue, and progress depends on funding and readiness conditions being satisfied.
Demand
Four channels, one price
Every channel sells the same product: a hybrid seat at $175 per month, in a named location.
Resellers
Technology solution providers, MSPs and other channel firms.
Direct enterprise
Enterprises contracting directly for distributed edge AI capacity.
Founding / pilot customers
Early workloads that benchmark the fabric and validate utilization.
Local investor network
Accredited purchasers in each market who may become customers, connectors or advocates.
There is no committed customer base and no executed capacity contracts. Demand channels are planned routes to market, and capacity may go unsold or sell below the modeled seat price.
Where this goes next
The economics follow from the structure
If the fabric is uniform and the seat price is one number, the return model is arithmetic rather than narrative.