The NBN: Australia’s Biggest Infrastructure Project and Its Most Expensive Compromise

The NBN is worth studying not as a telecommunications story but as a case study in how infrastructure decisions with thirty-year consequences are made on five-year political horizons. Every party to the decision behaved rationally within their incentives, the cheaper option genuinely was cheaper to build, and the total cost across the full asset life is now higher than the expensive option would have been. That pattern recurs across public infrastructure everywhere.

Why did Australia build a government broadband network?

Because the market had not delivered and regulating the incumbent had failed. Telstra owned the copper access network, competitors depended on wholesale access to it, and a decade of access disputes and ACCC determinations had produced neither adequate investment nor genuine competition. Broadband speeds lagged comparable countries.

Rather than continuing to regulate a vertically integrated monopolist, the government chose structural separation through construction: build a new wholesale-only network, buy out the copper, and prohibit the network owner from selling directly to consumers. That last constraint is the essential design feature, because it removes the conflict of interest that made regulating Telstra so difficult.

The wholesale-only model has broadly worked as intended. Retail competition in Australian broadband is genuine, prices are transparent, and no provider has a structural infrastructure advantage. The controversies concern how the network was built and what it costs, not whether the separation was correct.

What changed in 2013 and why does it matter?

The technology mix. The original plan ran fibre all the way to the premises for the large majority of homes, which is expensive and slow to deploy but requires no further upgrade for decades. The revised plan ran fibre to a street cabinet and reused the existing copper for the final stretch, alongside acquiring and reusing existing cable television networks.

The stated rationale was cost and speed of deployment, and on those terms it succeeded — the network was completed sooner and for less initial capital than the original design would have required. The counter-argument was always that copper in the final segment limits achievable speeds, requires ongoing maintenance, and would eventually need replacing anyway.

That counter-argument has been vindicated by subsequent policy. Australia has committed billions to upgrading fibre-to-the-node areas to full fibre, which means paying twice for the same connection. The total lifecycle cost of the cheaper option now exceeds what the original design would have cost, which is the standard outcome when capital expenditure is deferred on a long-lived asset.

How does NBN Co make money?

By charging retail service providers for wholesale access, through a combination of a monthly access charge per service and a bandwidth charge based on the capacity the provider purchases. That second component, historically known as the connectivity virtual circuit charge, has been the most contested element of the pricing structure.

The tension is straightforward. NBN Co must generate enough revenue to service the capital invested in building the network, which pushes wholesale prices up. Retail providers operating on thin margins argue that high bandwidth charges force them to under-provision capacity, causing congestion at peak times that consumers blame on the retailer rather than the wholesaler.

Pricing has been restructured repeatedly in response, and the underlying problem is unresolvable at the margin: the network cost what it cost, and someone must pay for it. Either wholesale prices stay high and consumers pay, or they fall and the taxpayer absorbs a larger write-down of the investment. Every pricing review is a negotiation about that allocation.

What does the NBN mean for competition?

It made retail broadband a commodity, which was the intention. Every provider buys identical wholesale inputs at identical published prices, so no retailer has a cost advantage from infrastructure. Competition therefore occurs on customer service, bundling, contract terms and brand — and on price, which compresses margins for everyone.

For Telstra this was a structural downgrade, converting a monopoly asset into a reselling business. For challengers such as TPG and smaller providers it was an opportunity, because access to customers no longer required negotiating with a competitor. Market shares in fixed broadband have shifted meaningfully as a result.

The unintended consequence is that value migrated to mobile. When fixed broadband offers no differentiation and thin margins, carriers invest where they still own the network and can charge for quality. That is why Australian telecommunications strategy is now almost entirely about mobile coverage, spectrum and 5G fixed wireless — the last of which competes directly with the network the government built.

Was the NBN worth it?

On the structural question, almost certainly yes. Separating the wholesale network from retail competition solved a problem that a decade of access regulation had not, and Australia now has a competitive retail broadband market with transparent pricing and universal availability including in remote areas that no commercial operator would have served.

On the execution, the assessment is harsher. The technology redesign deferred cost rather than avoiding it, the build took longer than either plan promised, and the capital invested is unlikely ever to be recovered in full at commercially acceptable returns. Treating part of the investment as a public good rather than a commercial asset would have been more honest from the outset.

The transferable lesson concerns how infrastructure is justified. The NBN was sold politically as a commercial investment that would earn a return, which constrained every subsequent decision about pricing and technology. Infrastructure that delivers broad economic benefit but not commercial returns should be funded and described as such, because pretending otherwise forces exactly the compromises that produced the multi-technology mix.

How does the NBN compare internationally?

Ambitious in structure and mixed in outcome. Very few countries chose full structural separation, and those that did — New Zealand being the closest comparison — have generally been satisfied with the competitive result. Most jurisdictions instead regulated access to incumbent networks, with the persistent under-investment that pattern produces.

On speed and technology Australia sits mid-table among developed economies rather than at the front, which is the direct legacy of the multi-technology mix. Countries that committed to full fibre earlier, including several in Asia and northern Europe, achieved higher average speeds sooner and are not now funding a second upgrade programme.

On universality Australia performs well. Satellite and fixed wireless components deliver service to remote premises that no commercial operator would connect, cross-subsidised by the urban network. That is a genuine public policy achievement and it is precisely the part that a purely commercial network would never have delivered.

What does the fibre upgrade programme involve?

Replacing the copper segment between the street cabinet and the premises with fibre, in areas originally built as fibre to the node. It is being rolled out progressively, generally on a demand-driven basis where households ordering higher speed plans trigger the upgrade for their address at no direct cost to them.

The economics are awkward to describe publicly. The work is funded from NBN Co’s balance sheet and ultimately from wholesale charges or the taxpayer, and it delivers the network capability that the original design would have provided in the first place. Framing it as an upgrade rather than a rebuild is politically necessary and analytically misleading.

For consumers the practical effect is straightforward: substantially higher achievable speeds and better reliability, with no need for the in-premises equipment changes that earlier technology transitions required. For anyone assessing infrastructure policy, it is the clearest available demonstration that deferred capital expenditure on a long-lived asset is a loan, not a saving.

What should businesses take from the NBN experience?

First, that structural questions and technology questions should be decided separately. The decision to separate wholesale from retail was sound and has held up; the decision about which technology to deploy was made on a shorter horizon and has not. Conflating the two produced a debate in which people argued about fibre when the important reform was separation.

Second, that a mandate to earn commercial returns on an asset built for public benefit will distort every subsequent decision. NBN Co’s pricing, technology and product choices have all been constrained by the requirement to recover an investment that was never going to earn a market return, and that constraint has been passed through to retailers and consumers.

Third, that transition costs are systematically underestimated. Migrating an entire country from one access network to another involved millions of individual connection events, each with the potential for service disruption, customer confusion and installation failure. The engineering was the easy part; the logistics of the migration consumed years and enormous goodwill.

Frequently Asked Questions