Summary

  • NBN Co's wholesale documentation is described as separating a recurring access charge per service from a dimensioning charge on connectivity virtual circuit capacity, which makes busy-hour headroom a purchasing decision the retail provider makes and the regulator can observe per provider.
  • Reported coverage of the pricing plan accepted in late 2023 says that separate capacity charge is being phased out for many services by bundling capacity into the access price, particularly at higher speed tiers — a change reported, not read, with no services-in-operation or capacity figure obtained and no current per-megabit rate asserted.

A wholesale broadband tariff that bills per service and per megabit is not one price with two names. The per-service charge buys the line, the port and the right to serve a household. The capacity charge buys the width of the pipe those households share at the hour they all use it. NBN Co's wholesale agreement material and pricing roadmap are described as distinguishing a monthly recurring access charge from a dimensioning charge applied to connectivity virtual circuit capacity (wholesale pricing roadmap).

The regulatory counterpart exists too: the Australian Competition and Consumer Commission publishes an NBN Wholesale Market Indicators Report whose tables are described as covering services in operation and the connectivity virtual circuit capacity acquired by retail service providers (ACCC wholesale market indicators). Put the two descriptions together and you have the two halves of one decision — how many services a retailer holds, and how much capacity it has bought to carry them.

What the record supports, and at what strength

Eight pages were retained for this article on 18 October 2026. Exactly one returned a fragment of its own text: a paywalled business column, cut off mid-sentence — "NBN Co has unsurprisingly decided to return to the drawing board with its Special Access Undertaking (SAU) given that the Australian Competition and Consumer Commission (ACCC) was almost certainly ..." (Business Spectator column). Everything else reached this desk as a summary written about a page that could not be read as text.

That distinction is not pedantry. A summary can be accurate and still not be quotable, and a structural claim carried only by summaries has to be labelled as reported rather than presented as a clause. This article does that throughout, because the alternative — quoting a tariff that was never read — would be a fabrication dressed as reporting.

A decade of teaching retailers to buy width

The mechanism has a history. Trade reporting states that in November 2015 NBN Co began trialling a discount pricing model intended to push retail service providers to buy more capacity for their customers (iTWire, November 2015), and that in April 2016 it introduced a tiered structure for the capacity charge with rates reported as low as A$11.50 per megabit, falling with the volume acquired (iTWire, April 2016).

Read that as design, not as a price list. A capacity charge that falls with volume is an explicit attempt to move a purchasing decision: it makes thin provisioning relatively expensive and generous provisioning relatively cheap, and it hands the retailer a dial it can turn as its subscriber base and its busy hour change. The A$11.50 figure belongs to April 2016. It is not a current rate, and nothing in the retained record says what the charge is today.

The phase-out, as reported

The later pages point the other way. Coverage of NBN Co's new pricing plan is described as saying the transition phases out the separate capacity charge for many services by bundling capacity into the wholesale access price, particularly at higher speed tiers (iTnews report). The regulator's page on the 2023 variation to the Special Access Undertaking is described as a major reset of how wholesale access is charged, including a transition toward simplified pricing structures for that capacity (ACCC variation page). A separate media release page is described as recording acceptance of the revised undertaking in October 2023, with the framework applying from the 2023–24 period (ACCC media release).

The two dates disagree, and this article will not pick one. One page's summary says November 2023; another says October 2023. Both are summaries; neither is the instrument. What they agree on is direction and period.

What moves when the megabit is bundled

This part is analysis, not observation, and should be read as the author's construct.

A separately priced capacity dimension distributes a judgement. The wholesaler sets the price per megabit and the shape of the discount; the retailer decides how much to buy for the subscribers it has. Contention at eight in the evening is then partly a purchasing outcome, and published per-provider capacity figures are the trace of that outcome — a way to see whether one retailer buys materially more headroom per service than another.

Fold capacity into the access price and the same judgement relocates. The wholesaler now chooses, tier by tier, how much capacity to assume per service, and the retailer's remaining lever is which tier it sells rather than how much headroom it provisions. The advertised speeds on the plan page do not change. What changes is where the assumption behind the evening experience is made, and how visible it is.

A second-order consequence follows from the same logic. If the bundling is concentrated at higher speed tiers — as the reported coverage suggests, though "particularly" is a summariser's word and not a schedule — then the capacity decision moves to the wholesaler precisely at the tiers where busy-hour contention is engineered rather than avoided. Readers on those tiers would still see a peak rate printed beside a price.

What this article cannot say is whether any of that has reached the measured data. No reporting period, services-in-operation count, capacity total or per-provider split was obtained. The regulator's series is described as existing; its numbers were not read.

What would change this account

Four checks, each of which the record could satisfy against the thesis rather than for it. A published undertaking clause or price schedule showing a separate per-megabit capacity charge still applying to a named access tier after the transition would defeat the bundled-tariff claim. An indicators edition that keeps reporting capacity acquired per retail provider on the same basis after the transition would show the purchasing decision survived the price change. A retail provider disclosing a separate capacity cost line after the transition would show the dial still exists at the retail layer.

And evidence that bundling applies only to entry-level tiers would invert the conclusion about where the judgement went.

The boundary, dated

Every factual statement above rests on material retrieved on 18 October 2026. On that date the regulator's pages, the wholesaler's pricing page and the trade report returned summaries rather than readable text; the one verbatim fragment is an incomplete sentence from a paywalled column. No services-in-operation figure, capacity total, per-provider split, busy-hour utilisation measure or current per-megabit rate was obtained, and none is asserted here. The acceptance date of the undertaking variation is in conflict between two retained pages and is presented as a conflict.

The account is therefore documentary. A two-part tariff was described. A volume-discount design on the capacity dimension was reported, with a dated price. A phase-out of that dimension was reported. Whether the phase-out has altered what retailers buy, and whether a published high speed now rests on a wholesaler-set capacity assumption, are open questions with named tests rather than conclusions.

That the ratio cannot be shown is itself the finding worth carrying. The instrument that made a provisioning decision legible is reported to be going away, and the reporting that would replace it has not appeared.