At a glance

  • Over 18 months TransGrid received more than 10GW of data centre connection enquiries in NSW, with about 6GW progressing to a formal application, a block of demand it likens in scale to a new Renewable Energy Zone.

  • In its submission to the NSW data centre inquiry, TransGrid proposes that developers fund the network upgrades that unlock their capacity, alongside measures such as take-or-pay and exit fees, so existing consumers are “no worse off”.

  • Oxford Economics, in research commissioned by AWS, estimates that six of every seven megawatts of Australia’s data centre connection requests is “phantom demand”; 44GW of national requests filters to about 7.9GW of likely projects and roughly 6GW of needed capacity.

  • TransGrid’s own Western Sydney studies find about 2.8GW of additional load could be unlocked there if data centres fund specific network augmentations.

  • NEXTDC chief executive Craig Scroggie reads the change as making capital, not the queue, the test of which projects are real, one that favours operators who can fund and build over those optioning connections.

  • The next signals are the AEMC’s connection rule for large loads, AEMO’s 2026 Integrated System Plan and Electricity Statement of Opportunities, and the inquiry’s report, due by 30 September 2026.

10GW of enquiries, about 6GW progressing

TransGrid frames data centres as a major economic opportunity that NSW should actively support, with the stated aim of unlocking network capacity faster rather than slowing it down. Its submission to the NSW data centre inquiry, lodged on 2 April 2026, sets out the scale it is managing. Over the past 18 months the operator received more than 10GW of data centre connection enquiries, with around 6GW progressing to a formal application. Individual facilities run from 250MW to more than 1,200MW, averaging about 650MW, and TransGrid likens the roughly 6GW in its pipeline to building a Renewable Energy Zone the size of Central-West Orana.

That demand sits on top of an already large capital task. An operator at the recent DCD Connect APAC investment forum in Bali put the cost of bringing a single gigawatt online in Australia at around A$50 billion. As we set out in our look at output per watt, the binding constraint in Australia is grid power rather than capital, which is what makes how that power is allocated the central question.

Six of every seven megawatts is phantom demand

The headline queue is not the same as real demand. In research commissioned by AWS, Oxford Economics found that of the 44GW of data centre connection requests AEMO received for its 2025 planning inputs, six of every seven megawatts is best described as “phantom demand” that will not reach the grid. Filtering for projects unlikely to proceed and for duplicate applications cuts the 44GW to about 7.9GW of prospective projects, and utilisation factors bring the capacity likely to be needed under AEMO’s central scenario to around 6GW.

The gap reflects developers competing to secure future capacity, often lodging the same project at multiple sites or well ahead of a firm commitment. For a network planner, that is the core problem: building transmission against requests that mostly evaporate would load cost onto consumers for capacity that never draws power.

A balance-sheet test for grid access

TransGrid’s answer is to make commitment financial. Its submission proposes that new large loads bear the costs of connecting to and using the network through mechanisms including application fees, long-term agreements with take-or-pay charges, prudential security, exit fees, and user-funded augmentations to unlock additional capacity. The guiding principle, which TransGrid calls non-negotiable, is that existing consumers are no worse off and do not subsidise new large loads.

The Western Sydney studies show what that looks like in practice. TransGrid’s preliminary network analysis finds about 2.8GW of additional load could be supplied there, but only if data centres fund a defined set of augmentations, including double-circuiting a key line into Sydney West, added voltage support and a new transformer at Kemps Creek, and if committed storage and other actionable projects proceed. Customer-funded connection assets are not new in themselves, since TransGrid’s contestable arm Lumea already builds them, but extending the principle to the shared augmentations that unlock capacity changes the calculus for proponents.

Alongside funding, TransGrid wants reliability built into connection agreements, including access to onsite backup generation, storage or demand management, fault ride-through obligations, and indicative load profiles before connection. Whether large loads stay relatively inflexible or can be encouraged to support the grid is, on AEMO’s current reading, still an open question.

What capital-funded access reprices

Writing on LinkedIn, NEXTDC chief executive Craig Scroggie read the shift as moving the industry from a forecast model to a commitment model, where augmentations must be funded upfront by the proponents who trigger them. In his account, that turns capital into the test of which projects are real and reprices proven delivery: operators who can fund, build and hold customers become the scarce asset, while an option-and-flip approach to connections gets harder. Because capital is mobile, he argues, a constrained NSW signals opportunity to other states. These are Scroggie’s conclusions rather than settled outcomes, and regulators have yet to decide how augmentation costs are allocated.

NEXTDC’s own position illustrates the point. The operator is expanding its S4 campus at Horsley Park in Western Sydney and has framed its recent contracting as de-risking those developments ahead of potential strategic partnerships with private capital from 2027. That is the profile of a builder positioning to fund and hold rather than trade connections.

The supply side is moving in parallel, which keeps the picture from being purely a constraint story. AEMO’s Draft 2026 Integrated System Plan identifies around A$9 billion of actionable and future transmission projects within a roughly A$128 billion sector-wide investment task, and projects net savings for consumers from that transmission. TransGrid’s EnergyConnect interconnector is near completion and HumeLink is under construction. AEMO also notes that large new loads can help drive investment in new supply, with data centres increasingly signing power purchase agreements that underwrite new generation and storage.

What to watch

Several near-term signals will show how fast this firms up. The AEMC is finalising new connection standards for large inverter-based loads. AEMO’s 2026 Integrated System Plan and its next Electricity Statement of Opportunities, expected around August, will fold updated data centre demand into system planning. The NSW Public Accountability and Works Committee must report on its data centre inquiry by 30 September 2026. Across all of them, the question is whether customer-funded access and reliability obligations move from proposal to standard practice, and whether large loads end up helping the grid rather than only drawing on it.