In the first half of 2026, 17 independent power producer projects reached commercial operation in South Africa, adding 1.92 GW. According to data from the Power Futures Lab at the University of Cape Town's Graduate School of Business, this was the highest capacity addition in any half-year, surpassing the previous annual record of 1.47 GW set in 2016.
More telling is the figure of 124: Nersa registered 124 generation facilities in the first quarter of 2026/2027 alone, providing 804 MW and attracting investments of R20.2 billion. The energy supply system has long ceased to be monopolistic, but most organizations purchasing energy have not yet adapted to the consequences of this change for their financial reporting.
The policy change occurred in two stages. In 2021, the generation licensing threshold was raised from 1 MW to 100 MW. In January 2023, amendments to Appendix 2 of the Electricity Regulation Act completely abolished the need for licensing for facilities transmitting energy to consumers through third-party networks, regardless of size; registration replaced permission.
Market Development in Cape Town
Cape Town is leading the development of these changes. In May, Growthpoint, trader Etana Energy, and the city itself announced the first joint transmission of renewable electricity from multiple facilities into the city's municipal grid. Energy produced by the Boston hydroelectric power station near Clarens in the Free State, jointly owned by Serengeti Energy and Growthpoint, is transmitted through the Eskom network into the municipal grid and distributed among five Growthpoint buildings, rather than being tied to a single facility. Growthpoint plans to expand this practice to more than 30 of its properties in the city.
In February, Discovery Green CEO Andre Nepgen estimated that about 80% of new private renewable generation over the previous year was channeled through traders, not direct bilateral contracts. The buyer now purchases energy from the market, not directly from the generator.
Metering is No Longer the Problem
Ten years ago, most companies could not track their consumption because readings were collected manually, at most monthly, and were often approximate. Technology has solved this problem: cheaper sensors, low-power broadband networks, and cloud platforms have made detailed metering commonplace.
However, deployment remains a challenge. Eskom's smart meter rollout for load shedding lines targets 577,347 meters. As reported by the Business Times, 513,022 had been installed by September 11, reaching about 89% of this target, although the goal was set for March 2026. The broader smart metering program will run until 2029. Some municipalities are also taking steps: Cape Town allocated R1.1 billion for the first phase of advanced water metering, covering approximately 83,000 meters between April 2026 and February 2029, as part of a city plan that includes over 680,000 meters.
Thus, metering is solved as a technical issue, but unevenly implemented. The cost of the measured commodity is not decreasing. Nersa approved an 8.76% increase in Eskom's direct tariff for 2026/2027 and over 9% for municipal distributors starting July 1, following the adjustment of Eskom's asset regulatory base by R54.7 billion. Water supply is moving along a similar trajectory. Water tariffs in Johannesburg rose by 12.5% from July 1, after increasing by 13.9% the previous year, and the city proposed increasing the fixed water charge by approximately 66% for households in 2026/27—a charge levied regardless of actual consumption.
Combining these two trends, it can be seen that one entity can receive energy from the municipal grid, from Eskom, from its own rooftop generation, and from a trader aggregating several independent producers elsewhere in the country, within a single month, using a single connection. Each source has its own commercial terms. Overlaid on these are usage periods depending on time, seasonal tariffs, public holiday rules, demand charges, network usage fees for transmitted energy (transportation fees on Eskom and municipal lines), and municipal surcharges. A meter can record all of this, but few organizations are capable of pricing it, allocating it to a tenant or cost center, and explaining the resulting outcome to any inquiry.
The pricing reform reflects this gap directly in the bill. The revised electricity pricing policy, published for comment in August 2026, requires bills to contain a full breakdown of tariff components for each consumed service, detailing energy, demand, network charges, fees, and municipal surcharges, and establishes a five-year transition plan to cost-reflective tariffs. The policy clearly states that the degree of bill separation depends on the established metering, as metering determines the limit of what can actually be measured.
Who Bears the Cost of This Gap
Before an organization can assess all of this, two things must happen, and neither is happening quickly: data capture and reconciliation. Implementing advanced metering is not a cheap upgrade. Cape Town's own water data—R1.1 billion for approximately 83,000 metering points—suggests a cost of around R13,000 per point, including installation, connectivity, and auxiliary infrastructure. The cost of electricity metering differs, but the private portfolio faces similar arithmetic without a municipal balance sheet.
Metering competes for capital with projects generating current quarterly revenue and loses, leading to assets operating years beyond their service life while the tariffs they measure continue to rise. Reconciliation is a more serious problem. Data arriving every 30 minutes from a portfolio using multiple suppliers is not an answer in itself. Tariff structures, lease terms, and occupancy must be integrated into one model, exceptions must be checked before billing, not after a tenant dispute, and the result must be justifiable to the tenant, auditor, and board of directors.
Most software used in South African industrial and property operations was designed for a simpler task, as that was the only thing that existed until recently: one supplier and a fixed tariff is an spreadsheet task. Four suppliers, a transmission agreement, and time-dependent usage periods are another story entirely.
AI Will Be Proposed as the Shortest Path
Applying artificial intelligence to an entity still assessing readings, or to a settlement process without a structured tariff model, will lead to faster versions of the same unreliable answer. Automating a process that no one can explain makes it less explainable, not more.
However, AI performs real work in subsequent stages. Consumption deviating from an asset's history can be flagged before it appears on the bill. A trader's monthly report can be line-by-line checked against meter readings—this is reconciliation that few finance teams can perform manually. Demand can be forecasted relative to time-of-use periods so that load is shifted before the peak, not explained after it. The order is: measure, then model, then automate.
This leaves the question facing the market: someone must take on the function of reconciliation, and no obvious candidate fits perfectly. The municipality is also a seller, making it an inconvenient arbiter of a disputed bill. The trader has a commercial interest in the number presented by them. The client is the party with the greatest incentive to get it right and usually the fewest means to do so, especially at smaller scales where justifying a dedicated energy manager is difficult.
Independent settlement is a recognized role in mature electricity markets, but it is almost absent here because before the 2021 and 2023 licensing reforms, there wasn't much to settle. The revised electricity pricing policy will force this issue to be addressed. Once the bill must show tariff components line-by-line, every party in the chain needs a number they can defend, and the five-year path to cost-reflective tariffs means that dispute amounts will grow every year. This turns the issue into a commercial one concerning market structure, not a technical one concerning meters.