South Africa passed 300 consecutive days without load shedding at midnight on 12 March 2026. For most finance teams that ended a four-year emergency, and it is tempting to treat the energy line on the budget as a problem that has been solved. It has not. The keeping-the-lights-on crisis has eased. The cost-of-electricity crisis has not, and it is the one that keeps compounding whether or not the grid stays up. If you are building a multi-year budget on the assumption that the worst is behind you, the number you are planning around is almost certainly too low.
The reason is that tariff risk and supply risk are two different things. One was a story about whether power was available. The other is a story about what that power costs, and that story runs on a separate, slower, more predictable curve that does not care how many days the grid has held.
The increases have not stopped, they have only slowed
It is easy to read the recent run of approvals as relief. NERSA granted Eskom direct-supply increases of 18.65 percent for 2023/24, 12.74 percent for 2024/25, 12.74 percent for 2025/26, and 8.76 percent for 2026/27, effective 1 April 2026. The annual number is clearly decelerating, and that is the figure that makes the headlines.
The annual number is also the wrong thing to budget against. What lands on your bill is the compounded effect. Run those four increases together and a unit of electricity that cost R1.00 in early 2023 costs roughly R1.64 by April 2026, a rise of about 64 percent in four years. A slower rate of increase on top of an already much higher base is still a meaningful rand increase every single year. Deceleration is not reversal.
And it has not stopped. A further increase of 8.83 percent is already approved for 2027/28, effective 1 April 2027. So the next NERSA increase is not a single event to absorb and forget. It is the next step on a curve you can already see, which means a credible energy budget has to be built out two to three years, not one.
The part of the bill a battery cannot touch
There is a structural change in the FY2027 schedule that matters more to a budget than the headline percentage, and it tends to go unnoticed because it sits inside the tariff mechanics rather than on the front page.
From FY2027, the fixed portion of the Generation Capacity Charge rose from 20 percent in FY2026 to 30 percent in FY2027, with the balance still recovered through the energy charge. In plain terms, more of what you pay Eskom is being moved out of the per-kilowatt-hour energy rate and into fixed and capacity-based charges. This matters because a solar-plus-storage asset earns its keep by reducing energy you draw and by shifting the timing of what you do draw. It is very good at the energy charge. It can do nothing about a fixed charge you pay regardless of consumption.
The practical implication for planning is direct. As the tariff rebalances toward fixed and capacity components, the share of your bill that any behind-the-meter asset can erode gets smaller, and the share you simply have to carry gets larger. A budget that assumes solar will keep clawing back the same proportion of every future increase is over-optimistic. The asset’s leverage is concentrated on the energy charge and on demand, and the energy charge is shrinking as a fraction of the whole.
Demand charges are where the next increase bites hardest
For a commercial site on Megaflex or Miniflex, the demand charge is frequently the most volatile line on the bill and the least understood. It is billed against your peak draw, not your total consumption, so a single half-hour of high simultaneous load in a month can set a charge you then pay against for the entire period. Every NERSA increase lifts that rate alongside the energy rate.
This is the line where active management does the most for a budget, and it is worth being precise about how. The asset cannot lower the demand-charge rate, which is set by NERSA and Eskom. What it can do is keep your billed peak down so the rising rate applies to a smaller number. In Soluno’s instrumentation the optimiser places a very heavy penalty on grid import above the site’s notified maximum demand, which in service terms means the dispatch is tuned to hold peak draw under that ceiling rather than letting an unmanaged battery breach it. As the demand rate climbs with each increase, the value of holding that peak down climbs with it. The same discipline is worth more in 2027 than it was in 2025, purely because the rate it protects against is higher.
The tariff you optimised against in 2024 no longer exists
There is a second reason a budget built on yesterday’s assumptions drifts, and it is not about the percentage at all. In the restructuring NERSA approved on 18 February 2025, the shape of the time-of-use tariff changed. The morning peak was reduced from three hours to two, the evening peak was extended from two hours to three, and off-peak rates were equalised across the seasons.
For a battery, the windows are the whole game. Stored energy is only worth what the gap between the cheap window and the expensive window makes it worth, and those windows moved. High-demand season weekday peaks now run 06:00 to 08:00 in the morning and 17:00 to 20:00 in the evening; in the low-demand season they shift to 07:00 to 09:00 and 18:00 to 21:00. The high-demand season itself is defined as 1 June to 31 August, with the low-demand season covering 1 September to 31 May.
If an asset is still charging and discharging against the pre-2025 windows, it is arbitraging into a tariff that was retired. The savings line in your budget quietly underdelivers, and nobody flags it, because the system still works and the bill is still lower than it would be with no asset at all. The gap is invisible precisely because it is a gap between two good outcomes, not a failure.
How to build the budget
Pulling this together, a few disciplines separate a credible multi-year energy budget from a hopeful one:
- Budget the compounded curve, not the annual rate. Plan against the cumulative rise across the next two to three years, including the 8.83 percent already approved for 1 April 2027, rather than treating each increase as a one-off shock.
- Split the bill into what the asset can move and what it cannot. Model the energy charge and the demand charge as the lines your solar-plus-storage asset works on, and carry the fixed and capacity charges, now a larger share, as a cost that rises with each increase no matter what.
- Treat the demand charge as a managed number. Budget your billed peak as something held down by active dispatch against the notified maximum demand, not as a figure that floats with whatever your busiest half-hour happens to be.
- Re-baseline savings against the current tariff structure. The percentage your asset saved under the pre-2025 windows is not the percentage it saves now. Use a saving figure derived from the tariff that is actually in force, ideally one shown in rand against a counterfactual grid-only bill rather than asserted.
None of this requires new hardware. It requires that the asset you already own is being managed against the tariff that exists this year, and that the budget reflects both the increases that have landed and the one already on the calendar.
The question for the finance team
Load shedding easing is real, and the residual supply risk has not vanished either: Eskom’s Winter Outlook 2026 still describes a high-risk case in which deep unplanned losses would bring back Stage 2 to 6 cuts during the mid-winter peak. But supply is not the line that erodes your margin year after year. Tariff is. So the question worth putting to your finance team before the next increase lands is simple: are we budgeting against the tariff we pay today, on the curve we can already see through 2027, with a clear split between the cost our asset can recover and the cost it cannot? If the energy line in the budget is still anchored to a number from the load-shedding years, it is anchored to a problem that has moved on, and it is the cost, not the outages, that will quietly run ahead of plan.
Sources and further reading
Want to know what your solar asset could save?
How Much Do I Save?