남아프리카 데이터센터 운영자들이 주택용 지붕태양광 소유자들보다 전력망 접근을 우선하도록 요구하고 있으며, 전력망이 정전에서 회복되는 과정에서 요금 및 형평성 갈등을 초래하고 있다.

For nearly a decade, South Africa endured rolling blackouts that its state utility failed to resolve. Businesses and homeowners stepped in, pouring billions into rooftop solar and driving private installed capacity from roughly 2,260 megawatts in mid-2022 to around 7,300 megawatts by late 2025. No government mandate directed this shift; market pressure did. Now, that hard-won stability faces a new stress test. A surge of data center investment—fueled by South Africa’s emergence as the continent’s dominant AI hub—is placing heavy, continuous demand on a grid that has only recently stabilized. Priority access to this grid is beginning to reshape the energy recovery ordinary citizens built.

Eskom’s decline was gradual, then total. Load shedding began in earnest in 2007, forcing South Africans to plan their lives around scheduled blackouts. Unable to maintain its aging coal fleet or build capacity quickly enough to meet demand, the state utility left waiting for a central fix as a futile strategy. The private sector stopped waiting. Manufacturers, retailers, businesses, and homeowners invested in rooftop solar because electricity became an economic necessity. This makes South Africa’s energy recovery unusual. Most large-scale transitions rely on subsidies, regulatory targets, or utility procurement programs. This one emerged from economic pain, producing one of the most genuinely market-driven energy shifts on record—and, as it turns out, one of the most vulnerable to being undermined.

South Africa already holds the continent’s dominant position in data infrastructure. It operates 56 data centers and accounts for roughly 1% of global AI data center capacity, with its ten largest facilities together carrying around 278 MW of load. That base is expanding rapidly. Approximately $1.5 billion in new data center projects are currently planned or underway, with the bulk concentrated in just the last two years. In Cape Town, four proposed facilities could collectively consume electricity equivalent to more than a third of the city’s current total power demand. Durban has announced plans for what could become the country’s largest AI-focused data center, requiring hundreds of megawatts. These projects bring real economic benefits—investment, employment, and a stronger foothold in the global AI economy—but they also place heavy, continuous load on a grid that has only recently found its footing. Eskom has already flagged further delays to its coal-closure schedule, creating genuine tension between two legitimate national interests.

Large operators are not simply plugging into the grid and drawing power. Some are building their own generation; Teraco, for instance, is constructing a 120 MW solar facility. Others are pursuing “wheeling” arrangements: long-term contracts that purchase electricity from utility-scale solar plants and deliver it through Eskom’s transmission network. On the surface, this appears constructive, as data center operators help finance new renewable generation rather than drawing more from aging coal plants. However, wheeling requires long contract tenures and balance-sheet strength that mid-size manufacturers or small businesses cannot offer. The divide is not about whether clean power exists, but about who can access the financial and regulatory mechanisms that unlock it. Consequently, the smaller businesses and households that collectively rescued the grid during load shedding now find themselves on the wrong side of that line, facing rising electricity costs, a growing patchwork of compliance requirements, and municipal charges moving in the wrong direction.

A recent Public Affairs Research Institute report quantifies the value of this second tier. By December 2024, Johannesburg, Ekurhuleni, and Tshwane together accounted for more than half of all rooftop solar installed nationwide. Most residential systems produce surplus electricity across a full 12-month period, but that surplus is curtailed because the infrastructure to absorb it is not in place. The report estimates that if municipalities purchased that surplus at a discount and stored it in leased battery systems for peak dispatch, the three metros could save close to R2.5 billion per year—roughly equal to their combined annual capital budgets. That represents a significant return on infrastructure that private citizens have already paid for and continue to maintain. Instead, policy is moving in the opposite direction. Municipalities are adding compliance costs, pushing installation owners onto more expensive time-of-use tariff structures, and requiring new metering at the owner’s expense. A survey cited in the report found that 40% of Johannesburg solar owners would feed surplus power back at zero tariff—just to avoid a fixed monthly charge of around R1,500 before any consumption at all. The policy response to that offer has been to make the terms worse, not better.

South Africa’s situation is not an outlier. The same collision between large-load data center growth and distributed solar capacity is already surfacing in California, Arizona, and PJM territory across the eastern United States. Ireland froze new grid connections near Dublin from 2021 until late 2025 because data center demand had grown to consume more than a fifth of the country’s electricity, lifting the block only when operators agreed to supply most of their own power from new renewables. What makes South Africa’s case distinct is that it has already lived through the consequences of grid failure and rebuilt from the bottom up. That experience provides an unusually clear view of what is at stake. The distributed generators being squeezed by punitive tariffs today are the same actors who kept the lights on when the utility could not. The policy choice now before South Africa—whether to integrate distributed generation or allow it to be taxed out of the market—will likely preview what other markets face within the next few years. If the country finds a way to accommodate data center growth without stranding the solar capacity its citizens built, it will have something genuinely useful to export: not just electrons, but a workable model. If it cannot, the second era of energy instability it risks may prove harder to solve than the first, because this time, the private capital that rescued the grid will have already walked away.

Carlos Albero Rojas serves as CEO. He is an engineer with strong expertise in technical and industrial topics, previously worked at international companies such as Siemens, and is multilingual.