This piece pulls together a few of our recent observations about the current boom in battery storage, and draws some parallels to a story the NEM has already lived through. Whether history is repeating itself is an open question, but these similarities are worth examining — and so are some important differences.

Retelling a familiar story

From around 2008, attractive state government feed-in tariffs and other subsidies triggered the first wave of rooftop solar adoption, with utility-scale solar farm development following a few years behind. A dramatic fall in panel manufacturing costs through the mid-2010s turned early adoption into something closer to a stampede. 

Two distinct but related phenomena slowly developed.

The rooftop story changed the shape of demand, as behind-the-meter solar carved a deep trough into middle-of-the-day grid demand — as a result, the industry became familiar with the term ‘the duck curve’.

The utility-scale story was more about investment risk. As more capacity was installed on both sides of the meter, a growing number of headwinds weakened individual project returns. In 2018, we coined the term ‘solar correlation penalty‘ as a shorthand way to explain what was happening.

Every solar farm in eastern Australia sits within a relatively narrow band of longitude, meaning they all share roughly the same daily production shape. Millions of rooftop systems also share this same profile. As more panels were installed in front and behind the meter, the strong returns that early solar farms had enjoyed were steadily eroded.

The investment risk for utility-scale projects was compounded by a structural imbalance between large and small solar. On a per-watt basis, utility-scale solar is almost always cheaper to build, yet utility-scale farms bore the full force of increasingly negative spot prices during high-generation periods, while residential rooftop systems are generally not directly exposed to wholesale prices at all. The result was an uneven playing field: it was the large solar farms that started to get curtailed, while behind-the-meter systems quietly eroded the market they depended on without facing the same consequences.

By late 2022, most banks were unwilling to finance new standalone solar farms, leaving tens of gigawatts of proposed solar capacity sitting in relative development purgatory. Today a number of large operating solar farms have been on the sale block for many years without finding buyers.

Boom and bust cycles happen in almost every market, but they are particularly cruel in energy. Heavy fixed costs, long construction lead times, and assets that cannot be unwound once built mean that by the time the damage becomes common knowledge, it is usually too late.

Is a similar narrative unfolding with batteries?

This brings me to the main contention of this piece: are we already seeing the first signs of what we’re going to call a ‘battery correlation penalty’? 

As we see it, the majority of batteries being installed in the NEM — both in front of and behind the meter — share similar durations, typically one to two hours, and almost all appear to cycle on a diurnal basis to maximise their return on investment. All else being equal, as more systems connect, logically these highly correlated operating patterns will weigh on financial performance across the fleet.

From late 2023, we began to see this saturation in the FCAS markets, where a growing number of new entrants steadily eroded what had been a reliable source of battery revenue.

In our recent review of utility-scale battery performance using our GSD2025 data, we noted that as more assets have entered the market, price-setting has started to split significantly — a sign that parts of the bid stack are becoming more crowded during peak periods.

In 2025, the utility-scale battery fleet in Victoria set prices above $300/MWh slightly more often than they did in the previous year, but that role was divided by more assets.

Source: GSD2025 Data Extract

We are also watching closely whether the large-versus-small asymmetry that played out in solar is beginning to repeat itself. Concerns along these lines have already been voiced within the industry and were examined in the Nelson Review, which highlighted the risk of grid-scale BESS being disadvantaged relative to smaller behind-the-meter systems that remain shielded from the wholesale price signals that should, in theory, be coordinating storage dispatch.

Evidence of battery saturation in the main energy market may have already started to accumulate. In a short analysis piece last month we noted that intraday two-hour price spreads this past summer were materially lower in four of the five NEM regions, with the distribution of daily spreads also appearing more compressed throughout the season, not just at the extremes.

This past summer saw 2hr price spreads in Queensland at their lowest in a number of years, with less spikes and a more compressed distribution of daily spreads across the summer.

Source: NEMreview

Of course, a single season is not a sufficient sample size to draw conclusions in a market with so many complex underlying price drivers — so price spreads are worth watching closely in the months and years ahead.

This time it’s different

Counterarguments can be made that the outlook for battery returns remains strong, and some have genuine merit. Data centres and electrification will add new load. But there is significant uncertainty around the timing and scale of each of these trends — and the solar story taught us that structural tailwinds rarely arrive on the same timeline as capital deployment.

Coal closures are sometimes cited in the same breath, but the dynamic there is more nuanced. Each station closure will likely bring a step change of temporary relief for battery returns — but in a market adding capacity at the current rate, that relief is likely to be quickly eroded as new storage connects to fill the gap.

The current battery buildout does differ from the solar buildout in a few ways — but not necessarily in ways that soften the outcomes.

Firstly, one key difference is that the battery buildout is now moving at a faster rate than the solar buildout ever did, and that rate is projected to accelerate. More than 7.5 GW of utility-scale battery capacity is already operating or in commissioning in the NEM, and around the same amount of capacity has reached financial close and is expected to connect in the next 24 to 36 months.

The battery fleet’s capacity has ramped up faster than the wind or solar fleet did at a comparable size. With at least 7.7 GW of additional batteries still coming, the growth of batteries will reach a much faster rate than at any stage of the solar buildout.

Source: NEMreview

To me, the more striking figure is behind-the-meter. In the first nine months of the federal government’s home battery subsidy scheme, somewhere in the vicinity of 4 to 5 GW of residential battery capacity has already been installed. Like rooftop solar before it, the vast majority of this capacity is effectively being dispatched before utility-scale storage assets — quietly shrinking the parts of the duck curve that grid-scale batteries target.

Secondly, the head of the duck is smaller than its belly. There is an enormous amount of energy that has been removed in the middle of the duck curve — and it took more than a decade of cumulative solar deployment to erode it. However, the price impact of that was relatively minor. For example, in 2015 midday prices in Queensland averaged around $40/MWh; in 2025, the average was around $0/MWh. The evening peak is a narrower window of time, but the marginal cost design of the NEM means that collapsing a price spike requires far less additional supply than pushing midday prices to zero ever did.

Lastly, the public holds more of the risk this time, and it holds this risk ex-ante. Early solar development was effectively underwritten by the Renewable Energy Target’s LGC scheme, where asset underperformance was almost entirely a private investment problem. The current buildout is different — roughly a third of battery capacity that has reached FID in the past three years is underwritten through the CIS, LTESA or VRET schemes, which transfer a portion of long-term revenue risk onto public balance sheets. Additionally, there is a much larger pool of battery projects that currently hold one of these government underwriting contracts but are still in the process of reaching FID.

Source: AEMO Services, Victorian Government, ASL, DCCEEW

Final remarks

With geopolitical tensions driving price uncertainty across global energy markets — and Q2, increasingly the most volatile quarter in the NEM, just getting underway — it may seem foolish to question the returns of any energy asset right now. But this piece is less about the short term and more about whether investment fundamentals are as straightforward as the current pace of the battery buildout implies.

When the economics of standalone solar first started to turn, there were enough people with commercial interests involved to create a kind of sentiment inertia in the industry — a collective reluctance to say out loud what the numbers were beginning to quietly suggest. As a result, many investors, developers, policy makers and other key stakeholders were slow to react and adjust their decisions accordingly.

So it is worth asking — with more seriousness than the question currently seems to attract — whether history may repeat or rhyme.

What are your thoughts? Do you agree or disagree? Leave a comment below.