Constraints

These two do not appear in the register below, and the omission is deliberate. A risk register lists contingencies with mitigations attached — items a reader can weight, discount and move past. Consent is not a contingency. There is no mitigation for a refusal, only redesign around it. The register that follows assumes both preconditions are cleared, and asks what could still fail.

Traditional Owner partnership and FPIC. Without genuine partnership on Country the corridor cannot exist. Not “should not” — cannot. The site’s position is that any TO group whose Country the corridor would cross has an absolute right to refuse, and the corridor redesigns around refusal rather than pressing through it. This is why Gate 0 exists. See Country for the principles, the illustrative NNTT table, and the engagement sequence.

Cultural heritage protection. Every state and territory has its own framework. Some are stronger than others. Cultural heritage assessment ahead of engineering — not parallel to it — is the position that survives external scrutiny. Where heritage protection and route efficiency conflict, the route changes.

Five constraints go to the thesis rather than to a node. If any one of them cannot be cleared, the delivered-cost advantage does not survive, and the corridor has no argument.

Storage cost for industrial firming. The thesis rests on firmed 24×7 supply at A$80–120/MWh. That figure is a function of installed storage cost, the duration required to ride through consecutive low-irradiance days, and the solar oversizing ratio. If installed BESS cost does not fall on the trajectory the staged build assumes, or if the required duration is longer than modelled, firmed cost rises toward the coastal delivered range of A$130–190/MWh and the advantage closes. Nothing else on this site survives that.

Anchor customer at an economic price. The buyer set for a training-scale tenancy is perhaps ten to twenty organisations globally. Each runs its own site selection process. Each is currently deploying at gigawatt scale in jurisdictions where power is cheaper to secure and faster to energise. Sovereign and research demand is more durable, less contested and less exposed to the capex cycle — and smaller. A first node that depends on winning a frontier training tenancy is underwritten by the least winnable segment.

Development capital for pre-site risk. The staged model depends on a first tranche of capital that funds land options, a lodged connection enquiry, heritage survey and customer development before any of it is de-risked. Most of the institutions named in an infrastructure capital stack do not hold mandates for that stage. The capital that does — strategic, corporate development, and catalytic — is a different market with different diligence and different terms.

Demand for latency-tolerant capacity at inland sites. Oxford Economics Australia, forecasting for AEMO’s 2025 ESOO and 2026 ISP, projects data centre load outside Sydney and Melbourne as minimal to FY50: 1.2 TWh in South Australia, 0.8 TWh across southern NSW and the ACT, 0.6 TWh in Western Australia. Inland is not a category in that forecast. A single 200 MW node draws 2.1 TWh a year — more than every forecast non-metropolitan data centre load in the country combined. The same source models efficiency-adjusted hyperscaler AI energy demand peaking shortly after 2030, approximately when a first node would energise.

Two readings are available and both are live. The forecast is built from network and industry data describing loads that have already declared themselves; it does not model a globally traded workload selecting an Australian inland site. Or it is correct, and the demand is not there.

The addressable pool is global rather than domestic. Roughly 45% of global data centre demand is latency-tolerant — near 35 GW in 2025, approaching 100 GW by 2030. One node is under 0.2% of the 2035 pool. That reframing does not remove the constraint. It converts it from market size to competitive position, against West Texas, Alberta, the Nordics, the Gulf, India and Johor.

Diverse long-haul and international fibre. Latency is settled by arithmetic. Fibre carries a round trip of roughly 10 microseconds per kilometre, which puts an inland site 1,000 km from the nearest capital at 13–15 ms — the same order as Sydney to Melbourne, a separation no operator treats as remote. Capacity, ownership and physical path diversity are not settled. A training tenancy moves datasets in and checkpoints out continuously and requires multiple physically diverse high-capacity paths; a single path is a single point of failure and ends the negotiation before price is discussed. Inland long-haul fibre follows rail and highway corridors, and public sources do not establish what capacity exists on the candidate corridors, who owns it, or whether two independent directions are available. Serving globally traded workloads adds a second dependency on international capacity.

Some published objections to comparable projects do not appear here because they do not survive scrutiny.

• “It’s too big to fund” — global infrastructure capital markets deploy at this scale annually
• “Industry won’t move inland” — hyperscalers already segment inland training regions from coastal inference regions
• “Solar is intermittent” — the corridor is designed around storage-firmed dispatch, not merchant solar
• “Nothing this ambitious ever gets built” — Sun Cable, CopperString, WGEH each disprove this at different scales

None of these are strong constraints on the corridor’s technical or commercial case. Where they appear in commentary, they usually indicate the writer has not looked at the delivered-firm-cost or workload-fit argument on The Case.