Skip to main content
    Behind the Meter #5

    The data centre headline is three times too small

    The AEMO Draft 2026 ISP forecasts NEM electricity demand nearly doubling by 2050. The political conversation has decided it's a data centre story. The actual numbers say electrification is 2.6 times bigger. Here's why the framing matters for any portfolio sitting on a five to ten year horizon.

    5 min read
    The data centre headline is three times too small

    Edition #5 — May 2026


    Welcome back to Behind the Meter. The last fortnight has been one of the busier ones on the policy calendar in a while. AEMO put out the Draft 2026 Integrated System Plan, the AER landed its final Default Market Offer for 2026-27, and the federal government quietly closed the largest renewable auction Australia has run. Each of those headlines is being reported as its own story. They are actually the same story, told three ways.

    I want to spend the take on the ISP, because the way the data centre piece of it is being framed is wrong, and the framing matters for how anyone with a 2030-shaped portfolio thinks about demand.


    The Take: The data centre headline is three times too small

    The line everyone has picked up out of the Draft 2026 ISP is that NEM electricity consumption nearly doubles by 2050, from 205 TWh to 389 TWh. The political conversation has decided the reason is data centres. Every panel I have sat on this year has had someone frame the next decade of grid investment as a data centre problem. The actual ISP numbers make it clear how wrong that framing is.

    Data centres are forecast to add about 29 TWh of demand by 2050. Electrification, mostly buildings and industry switching off gas and process heat, plus transport plugging into the grid, is forecast to add roughly 76 TWh. That is 2.6 times the data centre number. If you want to argue that the grid is being rebuilt for any single demand trend, the honest answer is that it is being rebuilt for the heat pump, the induction stove and the electric ute, not the hyperscale GPU cluster.

    Why does it matter which story we tell. Because the two demand types do very different things to the grid. Data centres are concentrated and clustered around a small number of network nodes, mostly in NSW and Victoria, close to existing fibre and transmission. Electrification is the opposite. It is distributed across every commercial site, every warehouse, every factory and every house in the network, and it doesn't arrive in 200 MW chunks you can plan a substation around. It arrives as a million separate decisions that, in aggregate, reshape network load profiles and connection economics for every site that wasn't planning to upgrade.

    The portfolio implications are different too. A data centre demand story leads you to large user tariffs and behind-the-meter generation. An electrification story leads you to feeder-level capacity, peak coincidence at the substation, and whether your commercial sites can physically draw the load they need without a network augmentation queue that doesn't move for two years.

    The supply side carries the same signal. The ISP models 120 GW of grid-scale wind and solar, 32 GW of grid-scale batteries, 14 GW of flexible gas, 12 GW of pumped hydro, and 6,000 km of new transmission on top of the existing 44,000 km network to get there. The annualised capital bill is $128 billion in today's dollars, and most of it gets recovered through network and renewable charges on the bills your clients are paying right now.

    So the take, put plainly. If you are advising on portfolios with a five to ten year horizon, model the demand story you actually face. For most commercial and industrial portfolios, that is an electrification story. The questions to run are: which sites have the headroom in their existing connection to absorb electrification without an upgrade, which ones are first in the queue when that load hits the substation, and which states recover the transmission build through the tariff first. The data centre story is real, but it is one piece of a much larger reshape. Don't let it be the whole conversation.


    Quick Hits

    • The final DMO landed on 26 May with reductions across every region. The AER's final Default Market Offer for 2026-27 confirmed the draft direction: meaningful price cuts for small businesses on standing offers across NSW, South East Queensland and South Australia from 1 July, driven by lower wholesale contract prices and increased renewable and battery output through the evening peak. The point I keep making about the DMO is that it is a ceiling, not a target. If your clients are sitting on standing offers a year out from a contract renewal, the DMO drop is the signal to act, not the result.

    • CIS Tender 7 came in 56% above target. Tender 9 is now open. The federal government awarded 19 projects totalling 7.8 GW under Capacity Investment Scheme Tender 7, against an indicative target of 5 GW. Eight of those are hybrid projects with batteries attached, adding 7.9 GWh of storage. Tender 9 opened on 25 May for another indicative 5 GW. The CIS oversubscription matters because it tells you the underwriting cost of new renewable capacity, and therefore the price of the PPAs that will clear off the back of these projects. Forward curves have to absorb it.

    • The UK price cap is rising 13% from July, almost entirely on gas. Ofgem confirmed on 27 May that the energy price cap for typical dual-fuel households rises to £1,862 a year, up from £1,641, with gas bills up 24% and electricity bills up around 5%. It is a useful cross-market data point for anyone building a procurement view of how a Middle East gas shock transmits into different markets. The same shock is in the AER's draft DMO caveats here, but the structural transmission to the Australian retail bill is much weaker because gas is a smaller share of the wholesale stack and contract-level coverage runs longer.


    From the blog: If the electrification thread above is live in your portfolio, the tariff optimisation piece I wrote on network charges and demand pricing walks through how the network side of the bill actually flows when load profiles shift. It is the piece I keep sending to consultants asking how to model an electrified site against an existing tariff.


    The headlines tell you the grid is being rebuilt for AI. The ISP tells you the grid is being rebuilt for everyone plugging in. Those are different problems with different answers, and the second one is the one most commercial buyers are going to live with. Worth reading the draft directly rather than the coverage.

    Until next fortnight,

    — Cohen

    Know someone wrestling with a connection upgrade or an electrification call right now? Send them this. They can subscribe at utilified.com/behind-the-meter. I'm also on LinkedIn if you want to argue with any of it.

    If something in here sparked a thought, hit reply. I read every one.

    Ready to Transform Your Operations?

    Discover how Utilified can help optimize your utility management processes.