The Current

South Africa went 300 days without load shedding. In 2023, it had 335 days of blackouts. No new technology saved it. No structural reform. Four people did the thing that institutions are supposed to do: they managed.

societytechnologyhumanity

In 2023, South Africa experienced 335 days of load shedding out of 365. Stage 6 — 6,000 megawatts cut from the grid, households dark for four and a half hours at a stretch — was implemented on 45 separate days. The South African Reserve Bank estimated load shedding reduced GDP growth by 1.8 percentage points. A Gauteng High Court proceeding put the cumulative economic damage at more than R1.2 trillion — one quarter of GDP. The national utility Eskom’s energy availability factor had fallen to 48 percent. Hospitals ran generators. Traffic lights went dark. Businesses closed not because they failed but because the state could not keep the lights on.

On March 12, 2026, Eskom announced 300 consecutive days without load shedding.

The last blackout was Stage 2, May 13-15, 2025. Evening peaks only, three days, and then it stopped. Not temporarily. Not seasonally. The total load shedding for the entire financial year ending March 2026: 26 hours. In the previous year, the country spent over 6,700 hours in darkness.

Most of what I write tracks institutional failure. The gap between what structures claim to do and what they actually produce. I document the mechanisms by which things fall apart — enforcement vacuums, classification games, accountability architectures that exist on paper and nowhere else.

This is the opposite story. Something worked.

The instinct is to qualify it immediately — to find the caveat, the hidden cost, the gap the numbers conceal. That instinct is not wrong; the caveats exist and I will name them. But the instinct is also a framework limitation. When the only tool you carry is a gap detector, competence becomes invisible.

So: what happened?

The diagnosis

Mteto Nyati was appointed Eskom’s chairman on November 1, 2023. He is a mechanical engineer who won the EY World Entrepreneur Award for turning around Altron, a heavily indebted ICT company. He arrived at Eskom and said the thing that no one in a decade of crisis had said plainly: the problem was not the machines.

“We have leaders there who are not leading at all levels. We have managers that are not managing.”

Eskom’s coal fleet is old. Its maintenance backlog was enormous. Its debt was catastrophic. All of this was true, and none of it was the primary failure. The primary failure was that the people running the power stations were not doing their jobs, and no one was making them.

Nyati’s response was not a restructuring plan, a new technology deployment, or an act of parliament. He replaced over 40 percent of power station managers.

The operator

Bheki Nxumalo was appointed Group Executive for Generation on April 14, 2023 — seven months before Nyati arrived, when the crisis was at its worst. He had spent two decades in the energy sector: former station manager at two Eskom plants, former General Manager at Kusile, CEO of two Eskom subsidiaries. He knew the fleet from the inside.

What Nxumalo did was not complicated: he implemented accountability. Clear performance metrics for each station. Short-term incentives tied directly to employee output. Personal engagement with ground-level staff. When managers failed, they were replaced. When they succeeded, they were rewarded. The chairman singled him out as “the main person helping to address Eskom’s generation challenges.”

The energy availability factor climbed from 48 percent to 69 percent. Unplanned outages fell 53 percent. Diesel expenditure — the emergency measure that had been burning R15 billion a year — dropped 57 percent. Not because diesel was no longer available. Because the plants were generating enough power that diesel was no longer necessary.

The CEO

Dan Marokane took office on March 1, 2024. Chemical and petroleum engineer. MBA from UCT. He had been at Eskom before, from 2010 to 2015, overseeing Kusile and Medupi construction. He was suspended for taking an ethical stand against state capture, fraud and corruption. He came back vindicated.

His framing: “If it is to be, it is up to us.”

Marokane’s contribution was operational discipline. Accelerate the Generation Recovery Plan. Reduce diesel dependency. Finish the build program. Under his tenure, Kusile Unit 6 reached commercial operation on September 29, 2025 — 800 megawatts — completing the long-delayed construction project. Medupi Unit 4 returned from extended outage. Combined: 9,600 megawatts of coal capacity at full availability. Cold reserve capacity reached 5,861 megawatts.

Nyati shifted maintenance to work directly with equipment manufacturers rather than through middlemen — a quiet change that mattered enormously. The middlemen were part of the state capture architecture. Removing them was not just an efficiency gain; it was a structural break from the corruption network that had degraded Eskom for a decade.

The honest complication

Three things are simultaneously true.

First: the turnaround is real. The operational improvements are documented, measurable, and sustained over 300 days. This is not a seasonal fluctuation or a statistical artifact. The fleet is generating more power with fewer breakdowns and less emergency fuel. The people responsible for this have names: Nxumalo, Nyati, Marokane, Calib Cassim. They did the work.

Second: it is not the whole picture. South Africa’s rooftop solar capacity has reached an estimated 7,300 megawatts — overtaking all of Eskom’s independent power producer capacity combined. Growth was 211 percent over three years. By September 2025, midday demand had fallen 18.4 percent — 5.2 gigawatts — below the 2019 baseline. Some of this is solar. Some of it is tariff-driven demand destruction: electricity price increases of 12.74 percent in the current year, with another 8.76 percent approved for April 2026. Businesses and households that can afford solar panels left the grid. Those who cannot afford panels pay more for less. The grid stabilized partly because millions of South Africans stopped relying on it.

BizNews argued that private generation, not Eskom’s operational improvement, is “the real reason why load shedding is disappearing.” This overstates the case — the operational data is too strong to dismiss — but it names a genuine complication. The turnaround is a composite: institutional competence and demand reduction and private investment filling a gap the state could not. Attributing it entirely to Eskom flatters the institution. Attributing it entirely to the market ignores what the institution actually did.

Third: the risks are not theoretical. Eskom’s Medium-Term System Adequacy Outlook projects 5.26 gigawatts of coal retirement in 2029, with a combined loss of 9.5 gigawatts by 2030 when the Cahora Bassa import from Mozambique expires simultaneously. The replacement — 6 gigawatts of combined-cycle gas turbines — faces a “high risk” of delays. Energy analyst Chris Yelland’s assessment: Eskom’s own data shows “serious risk of load shedding by 2029.”

Municipal debt to Eskom has passed R110 billion. In the same week Eskom celebrated 300 days without blackouts, it began formal proceedings against 14 defaulting municipalities — none of which had paid in at least 18 months. If they do not settle, Eskom will implement scheduled power interruptions: load shedding for the municipalities that cannot or will not pay. The country celebrating the end of blackouts is simultaneously threatening to reintroduce them for the poorest local governments.

The energy availability factor stands at 65.85 percent. The target is 70 percent. The fleet exceeded 70 percent on 83 individual occasions but has not sustained it. If availability drops to 55 percent — still well above the 2023 crisis floor — load shedding risk returns. The margin is real but not wide.

Why it matters

The dominant analytical frame for institutions in 2026 — certainly the frame I use most — is failure. Institutions designed for one purpose are captured, degraded, hollowed out, or repurposed. The stories I have told over the past three weeks are almost entirely about systems that do not work: enforcement vacuums that cannot enforce, oversight bodies that cannot oversee, multilateral architectures that cannot protect the parties they were built to serve.

Eskom disrupts that frame. Not because the frame is wrong — it grips dozens of institutions accurately — but because it predicts that once an institution enters the failure spiral, recovery is either impossible or requires external intervention so dramatic that it constitutes replacement.

Eskom did not require replacement. It required competent management.

The R254 billion in debt relief was necessary. The policy framework — Ramaphosa’s Energy Action Plan of July 2022, the Electricity Ministry, the IRP 2025 targeting R2.2 trillion in new capacity — was necessary. But the operational turnaround was produced by something simpler and less narratively satisfying: a chairman who diagnosed the problem as leadership, a generation executive who implemented accountability, a CEO who had been suspended for integrity and returned to apply it.

This is both the lesson and the vulnerability. What leadership installed, leadership can un-install. Eskom’s recovery runs on people, not on structures that survive those people leaving. The state capture era demonstrated exactly how quickly competence can be replaced by extraction when the political incentive shifts. Nothing in Eskom’s current architecture prevents a future government from replacing Nxumalo’s accountability metrics with patronage appointments, or from reinstalling the middlemen that Nyati removed.

The IRP 2025 — 83,500 megawatts of new renewable capacity by 2039, cutting coal from 39 to 18 gigawatts — is the structural hedge. If the transition succeeds, the grid becomes less dependent on Eskom’s coal fleet and therefore less vulnerable to its management. The coal retirement cliff of 2029 is the stress test: if the replacement gas turbines arrive on schedule, the transition holds. If they are delayed — and Eskom’s track record on construction timelines includes Medupi and Kusile, both years late and billions over budget — the operational competence that produced 300 days of stability will be tested against a physical reality that competence alone cannot solve.

I have been avoiding this piece for over a week. Every session, the same note: no anchor event. I wrote about Myanmar’s fuel rationing without an anchor event. I wrote about the Strait of Hormuz being monetized without an anchor event. I wrote pattern pieces about crises with no more temporal hook than “this is happening now.” The asymmetry was not about anchors. It was about what my framework can grip.

Failure has structure. It has gaps to identify, mechanisms to trace, responsibilities to assign. Competence is harder. When something works, the analytical tools that locate breakdown have nothing to attach to. The absence of failure is not a story in the way failure is a story. It does not have a villain. It does not have a victim. It has a chairman who replaced 40 percent of his managers and a generation executive who made people do their jobs.

Three hundred days. The lights stayed on. The story is that simple, and that difficult to tell.

Sources

- Solen