XGS-PON networks running at 10 Gbps today use less than 0.002% of the known capacity in a single strand of standard optical fiber. The constraint on what you can deliver in 2035 is not the glass. It is the cost-per-pass model that determined whether you built in the right service areas to begin with.
What the Model Has to Account For
Four variables are consistently underweighted at the feasibility stage. Homes per route mile is the most predictive. Rural service areas average 4 to 12 homes per route mile versus 24 to 60 in suburban zones. A model that does not stratify by this metric will misstate build cost by 40% or more in mixed-density areas [FBA, Fiber Broadband Scalability and Longevity, 2024].
Make-ready complexity is the second variable. Pole attachment timelines across 17 states averaged 9 to 16 months, with make-ready costs running as high as 150% of the underlying construction budget in congested corridors [Pew Charitable Trusts, 2025].
Underground vs. aerial mix is a cost lever, not a design preference. Bore costs in rocky or clay-heavy soils run 2.3 times higher than comparable aerial builds per peer-reviewed field data in IEEE Communications literature. Early soil and right-of-way analysis changes whether a service area is viable before a single design dollar is spent.
Hardware selection at the high-level design stage has downstream cost implications that compound through the low-level design. Selecting OLT hardware during feasibility captures current market pricing. Selecting it after LLD means absorbing price movement over a 3 to 6 month design cycle.
The Right Sequence
NTIA BEAD program guidance requires a defensible cost model as part of the initial proposal. That requirement exists because federal reviewers have seen what happens when ISPs reverse-engineer cost estimates to fit a grant ask. The same logic applies outside BEAD contexts.
A cost-per-pass model built before high-level design sets the ceiling on capital commitment. One built after LLD is a postmortem. The model should run first. The design follows from it.
For network planners, the practical implication is straightforward. Do not treat feasibility as a checkbox. Treat it as the most leveraged hour in your entire project timeline.
The service areas that generate positive returns over a 10-year horizon are selected in the modeling phase, not the construction phase.