AI goes to school in the bush
Where a data centre can find the best location for behind-the-meter gas generation in Australia
The escape hatch from the metropolitan connection queue is a behind-the-meter gas peaker — and this article is about where the molecules are.
If it’s all about speed to electrons, the opportunity is to locate directly on-top of the source of firmed generation - which is to be found in the northern states of Australia.
For the southern states, and it is not binary as this work will need doing also, the answer will likely be about engineering an energy product that links battery storage with virtual power plants to soak up the abundant rooftop solar – which is future FTB analysis.
However, in the interest of quick electrons, data centres want a systematic, homogenous offering that can be rolled out in any global jurisdiction and therefore would respond well to looking at the negative space emerging in energy system rules and policy.
As the regulator is going to make data centres spend on generation and firmness regardless – why not do it on your own terms?
AEMO’s 2026 Integrated System Plan re-affirms the critical function of gas-powered generation forecasting a total of 17 GW of gas generation is required by 2050 – more than double today’s ~8 GW fleet – to compensate for ‘dark and still’. Only last month, South Australia went from 100% renewables to its worst four-day wind drought in seven years, requiring gas to cover for over 50% of the state’s energy demand.
This article looks at the question that follows, where are the molecules that support a behind-the-meter gas proposition for large loads?
Let us look north to Queensland and the Northern Territory where the volume of gas reserves is staggering.
Queensland – but look beyond the default of Surat and Wallumbilla
Queensland LNG producers’ control around 70% of total 2P reserves in the East Coast Gas Market, and volumes of gas exported internationally via Curtis Island represent around 75% of annual consumption in the ECGM1.
There is substantial pipeline investment, such as APA’s East Coast Gas Grid Expansion Plan Stage 3A is committed, adding 58 TJ/d of capacity from Wallumbilla to Moomba with a combined 132 TJ/d of north-to-south capacity by the winter months of 20282.
AEMO’s southern supply adequacy assessment depends on the timely completion of these expansions.
However, exposure to the southern states is a disadvantage. Every new southbound pipe connects Queensland molecules to higher-priced southern demand and worsens a behind-the-meter proposition.
So where to look more closely in Queensland?
The North Queensland Gas Pipeline system – Moranbah to Townsville – received higher production forecasts from recent field performance reviews, and AEMO concludes there is sufficient committed and anticipated supply to meet forecast demand on the NQGP until 2043. No other region in the regulator’s gas statement of opportunities report gets a two decade greenlight.
Because the NQGP is an isolated system – connecting Townsville to the Moranbah Gas Plant, it is assessed separately from the Queensland LNG balance in the ECGM and therefore structurally insulated from netback pricing.
Layer on top of that, the ISP committed project CopperString3, cutting new high-voltage transmission through the same corridor and Moranbah-Townsville becomes the only place in the NEM where AEMO assessed gas-adequacy and a new high-voltage backbone arrive together.
Also worth a look at Mount Isa.
The APA Group Diamantina Power Station is a 242 MW combined cycle gas turbine plant not connected to the NEM as it is an isolated gas-fired system with natural gas delivered to the site by APA’s Carpentaria Gas Pipeline. The power station is located at the end of CopperString’s Western Link, at Mount Isa Substation. Connecting directly at Diamantina Power Station would be a very interesting proposition.
The Northern Territory – substantial gas resources but very limited transmission
The Northern Territory has world-class hydrocarbon volumes but its 132 kV grid is far too small to enable a 100+ MW load to plug in. For a data centre, the line capacity would need substantial investment to up-rate to a 330kV or 500kV to meet large load demand – and no plans exist for doing so.
Therefore any ‘Beetaloo AI Zone’ is not connecting to the grid, it is building its own behind-the-meter generation.
And Beetaloo gas exploration is now committed, no longer just inferred:
Beetaloo Energy’s Carpentaria Pilot Project has reached final investment decision and changed from anticipated to committed status, with an initial expected production of 9.5 TJ/d from mid-2026 which will increase to 25 TJ/d, or 9.1 PJ/y, from 2027.
Tamboran’s Shenandoah South Pilot has progressed from uncertain to FID, with up to 40 TJ/d expected from mid-2026.
APA’s Sturt Plateau Pipeline committed and under construction since November 20254.
For reference, a 100 MW data centre would require ~26 TJ/d to serve the flat load.
It would be worthwhile for a data centre to investigate an offtake, however note Beetaloo is currently limited by capacity of the Northen Gas Pipeline and the Carpentaria Gas Pipeline. APA’s proposed North to East Australia Pipeline via the 342 km Bulloo Interlink is still an uncertain project and not proposed for development within a five-year window.
Note the Federal Government’s is progressing a domestic gas reservation scheme5, which would require gas exporters to supply a proportion of production to the domestic market. The scheme is under consultation and was explicitly excluded from the 2026 GSOO’s modelling. If legislated as proposed, it re-prices every location in the table.
What this means for investors
The three locations in this article offer three different trades:
Moranbah to Townsville offers certainty and greater infrastructure, so the deal turns on securing a field expansion commitment.
For Mount Isa, it offers gas generation potentially outside the offset mandate which would add hundreds of millions in forced generation and storage capex to any NEM connected facility.
Beetaloo is about locking in cheap gas, because at scale, it is trapped. No infrastructure that improves that is proposed within the next five years. Build at the wellhead and you are the producer’s premium customer.
We maintain the firming-options and location screen behind this analysis. If you are pricing a gas-anchored data centre proposition and need the fuel story stress-tested, that is the work we do privately: info@followthebottleneck.com
Not investment advice. This article uses public information only and its contents is general in nature.
AEMO 2026 Gas Statement of Opportunities
Ibid
CopperString | Powerlink
AEMO 2026 Gas Statement of Opportunities
Domestic Gas Reservation Scheme to secure supply for Australian market | Department of Industry Science and Resources