Summary

  • Arcep’s draft MCO guidelines and accompanying cost model would estimate operating-maintenance costs separately for each publicly initiated FttH network and use those estimates as a reference for recurring wholesale access tariffs.
  • The model is a negotiation and regulatory-analysis tool, not a binding price order: a little over half of the networks in Arcep’s simulation were below modeled cost, with most of those networks showing a modeled coverage gap of €0.50–€2 per active access per month.

Arcep’s proposal addresses a problem that becomes more visible as France’s publicly initiated fibre networks move from construction toward long-term operation: what should recurring wholesale access charges reflect when the relevant operating and maintenance costs are difficult to compare across networks?

Published on 15 October 2026, the draft guidelines and associated cost model seek to make that cost basis more explicit. The immediate object is not a national tariff. Instead, Arcep proposes a minimum recurring monthly tariff method for each public-initiative fibre network, designed to support discussions between infrastructure operators and commercial operators under existing access contracts and to provide Arcep with an analytical reference when applying existing tariff principles.

That distinction matters. The draft has no binding force. It does not amend an existing contract, and it does not prevent a party from proposing another calculation method where that method can be justified under the applicable regulatory principles. Any contractual change still requires negotiation and case-specific examination. The practical significance of the model therefore lies less in automatic enforcement than in establishing a common accounting language for a dispute that might otherwise turn on incompatible cost calculations.

The proposal follows a consultation conducted from March to May 2026. Arcep says it adjusted the model in response to feedback, including by reducing its estimate of curative-maintenance interventions, increasing indirect maintenance-cost rates, and increasing costs associated with maintenance and wholesale-commercialization information technology. Those changes illustrate what the model is intended to do: turn operational assumptions into explicit parameters that can be examined, challenged and revised.

A network-specific calculation, not a national price

The reference calculation begins with a 25-year horizon, the projected active-line base of the network and modeled annual MCO costs. For the recurring-tariff comparison, the methodology applies a 1.27 multiplier and adds Orange civil-engineering rental together with an initial-deployment financing component. Arcep’s 25-year approach uses an 8% real cost-of-capital rate to derive the multiplier.

The active-line projection is particularly important because the denominator affects the resulting cost reference. Arcep’s proposed estimate draws on INSEE data: principal residences receive a 0.9 weighting, secondary residences a 0.25 weighting and establishments with at least one employee a 0.97 weighting, with the stated lags of three or two years. This is a modeled projection of target active lines, not a record of observed subscribers.

That difference makes the benchmark an economic model rather than a simple accounting exercise. A network with different expected take-up, territorial characteristics or operating drivers can produce a different result. The proposed framework is explicitly intended to balance simplicity and precision rather than reproduce every operational act. Some support costs are allocated across modeled activities, while revenues associated with some cost items are not netted because contractual practices differ.

Arcep’s simulation gives a useful indication of the range without converting it into a universal outcome. Based on infrastructure-operator data, a little over half of public-initiative FttH networks had recurring tariffs below the calculated costs. For most networks in that group, the modeled increase required for cost coverage was between €0.50 and €2 per active access per month.

That figure needs careful interpretation. It is a modeled wholesale access-contract adjustment for most of the under-recovering networks in the simulation. It is not a consumer broadband price increase, and it is not a tariff that Arcep has ordered every operator to adopt. Nor does the simulation establish that final renegotiated tariffs will reproduce the model output.

What the model leaves outside

The boundary around MCO is as important as the calculation inside it. The model concerns operating and maintenance costs; it is not a comprehensive benchmark for every cost incurred over the life of a fibre network.

The accompanying explanation says burial and diversion costs can vary by more than tenfold between territories. They are therefore not included in the generic model because they are irregular, difficult to predict and dependent on decisions by local authorities. Such costs require a different evidentiary treatment rather than being silently absorbed into an average MCO figure.

Complex connections and possible resilience investment are also expressly outside the MCO cost work and require separate discussions. Asset-renewal investment is identified for cost tracking, but no general benchmark amount had been established at publication. Initial deployment financing is treated separately within the reference calculation rather than making MCO a proxy for total lifecycle expenditure.

This separation prevents the proposed benchmark from being read as a complete measure of financial sustainability. France’s fibre rollout had reached 95.3% of premises eligible for fibre and 85.0% of fixed broadband subscribers using fibre as of 30 June 2026, according to Arcep. Those national figures provide context for the maturity of the rollout, but they do not demonstrate that each individual public network covers its costs.

From disagreement over costs to evidence over assumptions

The model’s institutional effect may therefore be greater than its formal legal force. Where parties previously relied on different approaches to recurring costs, a common network-specific methodology gives each side a defined set of assumptions to inspect.

For infrastructure operators, the relevant task is to demonstrate how network characteristics and operating drivers feed into the proposed calculation and where costs fall outside the generic model. For commercial operators, the corresponding task is to examine the projected active-line base, cost assumptions, allocation choices and other inputs rather than treating the benchmark as self-validating.

Arcep occupies a different position. The model can provide a consistent analytical reference when considering whether a recurring tariff is compatible with existing tariff principles, while retaining room for alternative methods that are properly justified. The result is a framework for comparison rather than a single prescribed answer.

The consultation remains open until 16 November 2026 at 18:00 Paris time, with Arcep directing particular attention to section 3.1.4 for consultation responses. The next stage is consequently still one of evidence and review. The central question is not whether the model produces one number, but whether its assumptions provide a sufficiently credible and usable basis for network-specific tariff discussions.