If you own or manage a fleet of behind-the-meter solar assets, the most expensive assumption you can carry into a financial model is that exported energy earns the same rand as imported energy avoided. It does not. South Africa now has a working framework for moving and selling generated electricity, but the framework rewards on-site self-consumption and treats export as a heavily discounted afterthought. Before any export revenue line goes into a portfolio model, three regulatory questions have to be answered: are the sites registered, can energy be wheeled to where it is worth more, and what does the grid actually pay for what you push back. Get these wrong and a line you booked as an asset becomes a write-off.
This is a map, not legal advice. The rules below are the ones a portfolio owner needs to model around in mid-2026.
Registration is triggered by connection, not by export
The first thing to settle across a portfolio is which sites are registered and with whom, because the obligation is wider than most owners assume. NERSA has clarified that the trigger is having a point of connection to the grid plus installed generation capacity. It is not whether you export. A site that runs purely for self-consumption and never feeds a single kilowatt-hour back is still caught the moment it has both a grid connection and generation behind the meter.
The threshold that decides where you register is capacity:
- Systems of 100 kW or less with a grid connection point register with the relevant distributor, meaning Eskom or the licensed municipality that supplies the site.
- Facilities above 100 kW with a grid connection point register directly with NERSA.
For a fleet, this matters because registration responsibility fragments by site. A national portfolio can easily straddle a dozen municipal distributors plus Eskom direct supply, each with its own paperwork, each with its own connection conditions. The only site genuinely exempt is one with no grid connection point at all. Everything else in a portfolio carries a registration obligation, and an unregistered grid-tied asset is a compliance exposure sitting quietly on the balance sheet until a distributor audit finds it.
There has been some easing on the cost of compliance at the small end. Eskom waives the registration fee for systems up to 50 kW, and that waiver has been extended to 30 September 2026. For an installer rolling out a portfolio of small commercial or residential-scale assets, that window is worth planning registrations around rather than letting it lapse.
Wheeling: the framework now has teeth, with limits
Wheeling is the mechanism that lets a generator deliver electricity from its connection point to a separate consumption point across a network it does not own. For a portfolio owner this is the difference between a roof that can only offset the building beneath it and a roof that can supply energy to other sites in the same group. It is what turns scattered assets into a fleet that can move value to where the tariff is highest.
The regulatory ground shifted on 3 March 2025, when NERSA approved updated Regulatory Rules on Network Charges for Third-Party Transportation of Energy. These rules enable cross-jurisdictional wheeling between Eskom and municipalities, which is the hard part: a generator on one network supplying a consumer on another. Eskom also runs a Gen-wheeling reconciliation tariff for sites in a wheeling transaction on Megaflex, Miniflex and similar tariffs.
Two limits belong in any model that leans on wheeling:
- The wheeled-energy credit is not the full retail rate. It uses the WEPS energy rates, and it explicitly excludes network losses and excludes the Generation Capacity Charge portion embedded in the time-of-use energy rates. The energy you wheel is credited at less than the energy a consuming site avoids paying.
- Access is uneven. Eskom launched virtual wheeling as a commercial offering for low-voltage clients in early 2025, with Vodacom as the first at scale, but medium-voltage clients and licensed traders remain outside it pending NERSA finalising the electricity trading rules.
The practical reading for a portfolio is that wheeling is real and usable, but it is not yet a frictionless internal energy market. Whether a given pair of sites can wheel between them depends on their networks, their voltage levels and their connection conditions. That has to be confirmed site by site before any inter-site transfer is modelled as revenue.
Net-billing: export is credited, not matched
The single most consequential rule for export economics is how the grid settles what you feed back. South Africa uses net-billing, not net-metering, and the gap between the two is where most optimistic models break.
Under net-metering, one kilowatt-hour exported cancels one kilowatt-hour imported at the same retail price. South Africa does not work this way. Under net-billing, which Eskom administers through its Offset or Gen-offset tariff, import and export are treated as two separate transactions. The site still pays the full retail tariff for what it consumes and the full capacity charge for the demand it places on the grid. Exported energy is credited back separately, at a lower export rate based on the Gen-offset and WEPS rates, with losses and the GCC portion stripped out. You buy at retail and you sell at wholesale-minus, in the same billing period, on the same meter.
There is a further ceiling that caps how much that export can ever be worth. The FY2027 schedule states that the total exported energy credited per time-of-use period cannot exceed the active energy measured by Eskom in that same period. In plain terms, you cannot bank a net credit beyond your own consumption in each TOU bucket. Export beyond what you draw in that window earns nothing. A site that generates far more than it consumes during the day does not get paid for the surplus once the cap bites.
Two structural shifts make export economics worse, not better, the longer a model runs. From FY2027 the fixed portion of the Generation Capacity Charge rose from 20 percent to 30 percent, moving more cost into a charge that export credits do not offset and that a battery cannot arbitrage away. And the direct Eskom tariff trajectory, 18.65 percent in 2023, 12.74 percent in 2024, 12.74 percent in 2025 and 8.76 percent for the year from April 2026, has compounded to roughly a 64 percent rise across four years, with a further 8.83 percent already approved for the year from 1 April 2027. The retail price you pay to import keeps climbing. The export rate you receive does not track it.
What this means for the portfolio model
Put the rules together and a clear order of value emerges. The most valuable kilowatt-hour is the one consumed on site at the moment it is generated, because it avoids the full, rising retail rate. The next most valuable is the one stored and discharged into a peak TOU window, again avoiding retail. Wheeling to a sister site comes next, useful but credited below retail and constrained by network access. Export to the grid under net-billing is the least valuable, credited below retail, capped per TOU period, and shrinking in relative terms as the fixed charge grows.
For a fleet owner the disciplined position is to model on-site self-consumption and peak-shaving as the load-bearing economics, and to treat export as a residual that is nice to have rather than a revenue line the financing depends on. Any model that books export at retail-equivalent rates, or that assumes uncapped credit, is overstating the asset.
This is precisely where active stewardship earns its place against a static model. Soluno models full time-of-use tariffs, including import and export rates with both TOU and stepped rate types, the seasonal schedules that split high-demand winter from low-demand summer, and the notified-demand and maximum-demand charges that net-billing leaves untouched. Dispatch is then tuned to that reality: charging is biased into the cheapest windows and discharging into the most expensive, self-consumption is favoured over export precisely because export pays less, and a heavy penalty is held against any grid import above the notified maximum demand, protecting the demand-charge line that every tariff increase inflates. Across a portfolio this is run as a recurring discipline, not a one-time setting, so each site stays tuned to the tariff and the rules as they actually are this year, not as they were when the system was commissioned.
The question to settle first
Before you count on a cent of export revenue across a portfolio, settle three things in writing. Which sites are registered, and with the distributor or with NERSA. Which sites can actually wheel to which, confirmed against their networks rather than assumed. And what the export rate and the per-period cap really credit, modelled against the rising retail rate you pay to import. Answer those honestly and the portfolio model reflects the asset you own. Skip them and you are financing against revenue the rules will not pay.
Sources and further reading
Want to know what your solar asset could save?
How Much Do I Save?