Routing
Phase 03The framework's core decision: the validated blueprint goes to whichever certified hub wins on cost of production, energy price, logistics, tariffs, risk, and proximity to demand. Wages are not one of the columns: not because production costs the same everywhere, but because the thesis moves the cost driver to robots, uptime, and engineering. This is the thesis in a single step. Routing here is manufacturing routing, not logistics routing. It decides where production should happen, not how boxes travel.
The economics choose, the hub accepts
The order bound to a tier, never to a factory: this is where that pays off. Routing weighs the blueprint's requirements against every eligible certified hub and asks one question: which tier-and-hub combination gives the best total cost, flexibility, risk, and portability for this product? Not "how cheaply can this one factory make it". That is the old model's question.
Certification is what makes the comparison honest: a hub's scorecard states what it can run and how well. Acceptance is not a handshake; it reserves real capacity against the blueprint's stations, tooling, and quality requirements. The hub readiness scorecard is the first public sketch of exactly this instrument: Layers 01 and 05 working together, certified capability making the capacity promise routable.
| Candidate | Certification | Production cost | Energy | Logistics | Tariffs | Capacity | Call |
|---|---|---|---|---|---|---|---|
| Hub A · Norway | Tier 3 | Unmeasured | Low, renewable | 2 days, road | None (EEA) | 3 weeks | Selected |
| Hub B · Poland | Tier 3 | Unmeasured | Moderate | 3 days, road | None (EU) | 9 weeks | Loses on capacity |
| Hub C · Vietnam | Tier 3 | Unmeasured | Moderate | ~28 days, sea | Exposed | Open | Loses on logistics + risk |
The BOM can flip this table: if the critical long-lead sensor came from Asia and dominated landed cost, Hub C could win. The BOM is a routing document, not a parts list. Wages appear nowhere in the columns. That is not a claim that production costs the same everywhere; the thesis calls that reading "too absolute". It is a claim that the driver moved: where the labor rate used to be central, "robot/capacity cost, uptime, energy, and engineering become more central" (thesis §4). Hubs still differ on what it costs to make the thing. They differ on a different variable. Which is why the third column is specified and empty. See what has to be proven below. The call above is made on the five decision columns that can be read today; the unmeasured one could change it.
What routing reads
Ten inputs, straight from the thesis §9 table. An eleventh is added here out of §4: the thing that replaced the wage rate. None of them is a wage rate, and one of them is what replaced it.
Near the material first, near demand over time
While the network is thin, products are produced near the highest shipping-cost or most critical material and component: the BOM's heaviest constraint anchors the choice.
As certified hubs and supplier clusters mature, production moves closer to final demand. The scalable path is conversion, not construction: existing factories already have buildings, power, people, machines, and industrial culture. Moduloa adds the framework that makes them comparable and routable.
Routing is not a one-time event. When tariffs shift, a war closes a lane, energy prices diverge, or demand moves, a Tier 5 blueprint can be redeployed to another certified hub. That optionality is precisely what the higher tiers ask customers to pay for; whether they will is an open claim the register tracks as P-09. Lower tiers trade that flexibility away for speed and lower upfront engineering, and the model says so honestly.
What has to be proven
Cost of production is what §4 moves toward the centre when the labor rate moves out, and it is the one column of the comparison above with nothing in it.
Routing is the part of the model that needs a network to be real, and today the network is zero hubs. The comparison table above is the shape of the decision, not a screenshot of a working system. What the register tracks: P-09 (customers pay a premium for manufacturing optionality), P-23 (the first hub exists and proves the humanoid-native operating model by 2031), P-24 (the network scales by converting existing factories, by 2036), and P-25 (the framework operates as a global capacity network, by 2046). See the register →