Rarely does a power system change so much in so little time. In little more than a decade, Chile came to have one of the most renewable power systems in the world: during daylight hours, much of its demand is met with hydro, solar and wind energy at a variable cost close to zero. It is an achievement worth appreciating, and it brought with it a phenomenon that until recently was marginal: curtailment, that is, the available renewable energy the system cannot put to use. In 2025 it amounted to one in every six solar and wind megawatt-hours.
It is worth starting with a clarification: curtailment is endemic to highly renewable systems. Any system that commits seriously to variable sources lives with some degree of surplus, and what matters is how much, when and where. Seen this way, curtailment tells us that the capacity to produce clean energy has advanced faster than the system's capacity to consume it, store it or move it from one hour to another or from one place to another; rather than a problem in itself, it is a gauge of the system's flexibility.
In 2026 that flexibility received a significant boost: the battery fleet in commercial operation and testing doubled within a few months. Curtailment fell, but considerably less than that expansion would suggest. To understand why, we need to look at how a system with this make-up works, hour by hour.
A system that produces more than it can use at midday
The first feature of a deeply renewable system is that supply stops following demand: the resource sets it. During the day, solar generation rises until it exceeds what the system can consume, and the surplus appears precisely in the hours of highest output. As the solar and wind fleet expanded, curtailment went from an occasional event to part of daily operation.
This series hides a relevant nuance. In recent years the volume curtailed kept growing, but the share of available energy left unused stabilised at around 16%. In other words: the system is not losing its capacity to integrate renewable energy; it is growing, and curtailment grows with it. The underlying question is therefore not how much energy is curtailed but what fraction, and whether that fraction actually gives way as the system adds flexibility.
In 2026 the cumulative curve parts from the previous year during winter, only to run parallel again with the arrival of spring. With twice the storage, the annual trajectory barely changed shape. That is what needs explaining.
The shape of the day
Let us first think of curtailment as a timing problem. At night the system barely curtails: demand comfortably exceeds the available renewable generation. The surplus is concentrated in daylight hours and peaks in the early afternoon, when solar output is still high and demand has not yet begun its climb towards the evening peak.
Year after year the curve grew taller. In 2026 it also changed shape: the middle of the day gives way while the morning and evening hours gain weight. It shows how the system has begun to shift energy from midday to other hours, and how the surplus widens towards the edges of the solar day.
Storage reaches system scale
That shape of the day is precisely the problem storage is meant to solve. What a battery does, in essence, is move energy from one hour to another: it takes it when there is too much and delivers it when the system needs it. Barely three years ago batteries were a marginal resource on the grid, with 84 MW in operation; today they exceed 7,000 MW in commercial operation and testing. Few technologies have scaled at that pace in the history of the system.
The fleet operates as expected. Batteries charge in the hours of surplus and discharge into the evening peak, displacing generation that would otherwise be thermal. In energy terms they are already a significant player in the system's daily balance.
Where the new energy went
If batteries charge precisely when the system curtails, why did curtailment fall so little? The answer is that the system did not stand still while storage grew. Between January and August close to 2,900 GWh of additional solar and wind generation came in, and most of that energy ended up, precisely, charging batteries. Over the same period curtailment fell by 4%.
This does not detract from storage; quite the opposite. Without batteries, all that new generation would have been added to curtailment. What the data shows is that, in a system still adding renewable energy at such a fast pace, flexibility must grow at least as fast as variable generation for the surplus to actually shrink.
The breakdown by technology adds a further piece. Midday solar curtailment fell clearly: that is where the battery meets the problem it was designed for. Wind curtailment, by contrast, increased, and it did so across the whole day.
This contrast shows both the reach and the limit of storage. Where the surplus is concentrated in a few hours, the battery shifts it. Curtailment, however, is structurally a system-wide phenomenon: overproduction relative to what the system consumes and is able to manage at each moment. In those hours, units that cannot reduce their output remain dispatched —thermal plants running at technical minimum and hydro plants bound by water use and, in the summer season, by irrigation agreements— and the variable generation that does not fit on top of that base ends up being curtailed. Wind, which also blows in hours when batteries are already full or discharging, faces the same ceiling.
The seasons
The year-to-date total hides marked seasonality, probably the most revealing part of the year. In summer, with the solar resource at its peak, 2026 curtailed more than the year before despite the new storage. In winter the relationship reversed sharply, owing to significant cloud-cover events brought by the arrival of El Niño. With spring, the gap has narrowed again.
As one might expect, in winter the daily surplus is small enough for the battery fleet to store almost all of it. In the months of highest radiation, by contrast, the surplus is several times larger and batteries fill up before the afternoon is over: in October they charge about 19 GWh a day and, even so, the system curtails in the order of 26. A four-hour battery handles a four-hour surplus very well; faced with a longer one, it covers only part of it.
Hydrology adds another layer. The winter of 2026 was wet: hydroelectric generation rose sharply and thermal output fell significantly. When water is plentiful, the margin of thermal generation the system can displace narrows, and hydro itself begins to record curtailment.
What lies ahead could weigh more than what has been observed. Given the snowpack accumulated in the basins, a very marked melt season is anticipated, with levels of hydroelectric output not seen in recent years, coinciding moreover with the months of highest solar radiation. Whether curtailment actually falls in 2026, then, remains to be seen.
Geography
The Chilean grid is long and narrow, and curtailment has an address. For years it was concentrated in the north, where most of the solar fleet was built: in 2025 Antofagasta and Atacama accounted for four of every five GWh curtailed. In 2026 that distribution began to change. The north curtails less than the year before, while from Coquimbo southwards curtailment is rising in every region. Batteries, by contrast, remain concentrated almost entirely in the north, which accounts for more than 90% of their charging.
The pattern in both charts is consistent with how storage operates. A battery charging in the hours of greatest surplus relieves its whole electrical neighbourhood, not just the plant it is attached to. Its effect therefore shows up at system scale —at the solar midday and in the area where the fleet is concentrated— and, for the same reason, the remaining curtailment tends to grow where there are no batteries.
Zero prices
Curtailment is the energy the system does not take. A zero marginal cost is its market counterpart. Both are sides of the same condition, and the second makes it possible to follow the phenomenon precisely across the territory.
A few years ago zero-price hours were a rarity; today they are part of the system's normal functioning for much of the year. In September 2026, Charrúa recorded a zero marginal cost in 62% of the hours, the highest in the series for any of the reference busbars.
The value of energy
In the months of greatest surplus, the spot revenue captured by solar generation falls to minimal levels: in September 2026 it was 1.9 USD per MWh, far below the system's average marginal cost.
For a solar plant with no contract and no storage, curtailment reduces the volume it sells, and zero prices reduce the value of the volume it does deliver. The second loss is considerably larger than the first. It is also the clearest economic signal that the system needs flexibility: midday solar energy is worth little precisely because it is abundant, and that difference in value between hours is what makes storage profitable.
Flexibility as the central challenge
The analyses in this column converge on a single idea. The battery shifts energy in time, and does it well. But the size of the surplus does not depend only on how much energy can be shifted: it also depends on how much generation remains dispatched without the ability to reduce its output.
Let us think of midday as a sum. On top of a base that cannot be reduced —thermal at technical minimum, hydro committed by water use and irrigation— sit the sun and the wind. Every megawatt of storage raises the bar of what the system can absorb; every megawatt of inflexible base lowers it. Thus, as long as the base and variable generation grow together, storage will reduce the surplus in the hours it can cover, but will not eliminate it.
That, in the end, is the good news. A system that curtails renewable energy is a system that already produces more clean energy than it can use at certain hours. The focus thus moves from how much is built to how what has been built is operated: it is not about producing more, but about making better use of what is produced.
What to watch from here
The pipeline of batteries under construction, close to 4,900 MW according to the CNE registry, is equivalent to two thirds of the current fleet and, like the existing one, is concentrated in the north of the country. Its effect will depend as much on its size as on what happens to the rest of the system. Among other aspects, it is worth following:
- Whether flexibility grows faster than variable generation. This year, storage absorbed most of the renewable growth. For curtailment to fall in a sustained way, the system's flexibility must advance faster than new generation.
- How the inflexible base evolves. Curtailment is a system-wide phenomenon: it depends on how much thermal generation runs at technical minimum and how much hydro is committed by water use and irrigation. There is as much room for improvement there as in storage itself.
- Where new flexibility is located. A battery's effect shows up in the area where it operates. With curtailment growing in the centre-south and the pipeline concentrated in the north, location matters as much as capacity.
- Snowmelt. The accumulated snowpack points to an exceptional hydro season, coinciding with the months of highest solar radiation. It will be the toughest test the battery fleet has faced so far.
- The monthly rate, rather than the volume. Volume grows with the fleet, and the year-to-date total mixes very different seasons. The monthly rate over available energy is the series that shows whether the system is gaining flexibility or simply growing.
The starting point is promising. The system has already shown that it can integrate renewable energy on a scale few anticipated. What is needed now is for the way it is operated, its rules and its prices to rise to the level of the infrastructure already built.