When Engineers Were Pushed Out, South Africa Went Dark

South Africa did not lose its grid overnight. It lost it through policy, personnel decisions, and a creed that treated representation as a substitute for competence. By the time the country reached rolling blackouts on a national scale, it had already spent years pushing out the people who knew how to keep the lights on.

The argument that once flared between Elon Musk and Don Lemon over diversity and standards looked, at first, like the usual television skirmish. In South Africa, the country where Musk was born, it became a matter of public ruin. A once reliable power system was recast as a vehicle for ‘deracialisation’, the engineering corps was thinned out, and the bill arrived in the form of stalled factories, ruined small businesses, dark suburbs, and a national utility forced to beg many of the very engineers it had driven away to come back.

Elon Musk Accuses South Africa of Blocking Starlink Over Race-Based Ownership Laws

Elon Musk has repeatedly claimed that Starlink is barred from South Africa because of the country’s Black Economic Empowerment laws, which he has called racist. The SpaceX founder, born in Pretoria, has said the company cannot secure a license “simply because I am not black.”

The application of race-preferential policies was seen as a net good for post-apartheid South Africa!

At issue is a requirement under the Electronic Communications Act. Telecommunications license holders must have at least 30 percent ownership by “historically disadvantaged groups,” a category that primarily covers Black South Africans and others disadvantaged under apartheid. These rules sit within the Broad-Based Black Economic Empowerment (B-BBEE) framework introduced after 1994 to promote economic redress. Starlink needs individual network and service licenses from the Independent Communications Authority of South Africa (ICASA), which include this ownership threshold.

Musk has described the mandates as “openly racist” and claimed authorities offered arrangements he characterized as bribery to create the appearance of Black control, which he refused. South African officials reject this framing. They say the rules apply equally to all operators, foreign or domestic, and that Starlink has never submitted a formal license application. Communications Minister Solly Malatsi has confirmed ICASA has received no applications from SpaceX or Starlink for the necessary licenses or spectrum.

One possible alternative under discussion is Equity Equivalent Investment Programs (EEIPs), which would allow companies to meet empowerment goals through skills development, enterprise support, or other investments instead of selling equity. Malatsi has pushed for policy changes to more fully recognize EEIPs in telecoms licensing. However, ICASA has stated that this would require amendments to the Electronic Communications Act itself and has declined to alter its regulations pending legislative reform.

Starlink operates in many other African countries without equivalent ownership rules. As of July 2026, the 30 percent historically disadvantaged ownership requirement remains in force for individual licenses. The dispute continues to highlight tensions over whether B-BBEE policies deter investment or remain essential for post-apartheid transformation.

The Question That Was Really Being Asked

The fight that broke out around Musk and Lemon was never only about a contract or a talk show. It was about a larger and older question that modern politics prefers to hide in soft language: when a system is complicated and failure is costly, does it matter more who sits in the job, or whether the person in the job can actually do it? That question sounds abstract in a studio. It stops sounding abstract when a power plant fails, or a substation burns, or a city’s lights begin to flicker in the evening and then go dark for hours at a time.

South Africa answered that question in the most expensive way possible. The country did not run out of coal. It did not lose its grip to a foreign enemy, a drought, or a once-in-a-century storm. It eroded its own capacity from within. A state-owned utility that had once been admired as the most advanced electrical system on the African continent was gradually turned into a test case for a political idea: that the race profile of the workforce mattered more than the accumulated knowledge that held the machinery together.

That idea was not introduced in a single dramatic decree. It arrived as policy language, hiring rules, promotion criteria, and corporate scorecards. It arrived in the bureaucratic vocabulary of transformation, a word that can mean many things and often conceals more than it reveals. In practice, on South Africa’s grid, it meant that the people running the power stations were no longer evaluated first on their ability to keep the system stable. They were being judged on whether they aligned with the ruling party’s demographic ambitions.

That is how a country can fail without admitting it has chosen failure. The lights stay on for a while. The speeches continue. The annual reports are printed. And then, when the old hands are gone and the younger replacements have not had enough time to learn what the old hands knew, the machinery begins to break in ways that no policy document can repair.

The Grid South Africa Once Had

The collapse matters because South Africa once had something worth losing. In the late 1990s, Eskom operated a grid that was not merely functional but formidable. Installed capacity stood at more than 38,000 megawatts. Energy availability was consistently above 85 percent. Reserve margins were comfortably above 20 percent. In plain English, the system had room to breathe. It was not running on the edge of collapse, nor was it one bad afternoon away from disaster.

That reliability was not a gift from nature. It was the product of a concentrated body of technical knowledge built over decades. About 4,200 registered engineering professionals worked inside the system. They were not interchangeable bureaucrats. They were the people who understood high-pressure boilers, live-line maintenance, turbine bearings, transmission faults, and the small signs that tell an experienced engineer a large machine is about to fail. They knew what a healthy plant sounded like. They knew which sounds meant trouble. They knew how to read the system the way a physician reads a pulse.

The utility’s strength also mattered beyond its borders. South Africa was exporting roughly 1,500 megawatts a year to neighboring countries. It was earning foreign exchange. It was powering mines, heavy industry, and a manufacturing base that made the country one of the most significant industrial economies on the continent. There was a time when South Africa’s electricity system was a source of strategic advantage rather than national embarrassment.

That world was not an illusion. It was real, and it was built by people whose careers were shaped by long apprenticeships and brutal technical standards. Some had spent 15 or 30 years learning how the system behaved under pressure. They worked through the political isolation of the apartheid years, the sanctions period, and the economic uncertainty that followed. Their accumulated expertise formed one of the most concentrated stores of power-engineering knowledge in Africa.

The modern temptation is to describe that sort of achievement as accidental, or to pretend it can be replaced by good intentions and accelerated training. It cannot. Complex systems are unforgiving. They reward experience because experience reduces error. They punish ideologically driven shortcuts because shortcuts eventually show up as outages.

When Policy Became Staffing Doctrine

The first decisive shift came not through a blackout but through a policy vocabulary. In 1998, the government published an energy white paper that explicitly framed electricity supply as part of a broader project of ‘deracialisation’. The idea was not that the grid should be left untouched by the country’s democratic transition. No serious person could argue that the post-apartheid state should simply preserve every old pattern in amber. The problem was the method: instead of treating engineering competence as the first requirement and social redress as a separate question, the state began to blend the two until the second overwhelmed the first.

The Employment Equity Act of 1998 and the Broad-Based Black Economic Empowerment Act of 2003 gave that impulse legal and bureaucratic form. By 2004, Eskom’s shareholder compact required that 60 percent of all new engineering appointments come from designated groups. Promotion pathways that had once taken 15 years were compressed to 7 or 9 years. Executive bonuses were tied to transformation scorecards and performance. The utility began to reward compliance with demographic targets, not merely reliability, cost control, and plant availability.

That shift sounds neat on paper. It never is in practice. In a power system, promotion schedules matter because the people moving upward carry institutional memory with them. They know the history of a plant, the recurring faults, the false alarms, the suppliers, the modifications, and the shortcuts that worked and the shortcuts that failed. Compress the pipeline too aggressively, and you do not simply change the face of the institution. You break the chain by which knowledge moves from one generation to the next.

The policy papers dressed this in the language of fairness, but the operational logic was blunt. If the old workforce did not fit the desired demographic profile, it had to be reduced. If the best-qualified people were the wrong people by the new political standard(white), then the standard itself took precedence over the machinery they kept running.

That is why this story matters beyond South Africa. Ideology rarely announces itself as ideology. It arrives as a moral correction. It says a system is unfair, and often it is. But if correction becomes a substitute for competence, the machinery does not care about the rhetoric. Boilers still need people who understand boilers. Transmission systems still need people who can diagnose transmission failures. If those people are no longer selected for mastery, the consequences are only delayed, not avoided.

Critics of race-preferential policies note that the pattern is longstanding. They cite the Bolsheviks’ campaign against the kulaks—prosperous farming landlords—which liquidated a productive class, crippled agricultural output and contributed to famine. A contemporary parallel, they argue, appears in U.S. elite university admissions, where Asian-American applicants were shown to require substantially higher test scores and academic credentials than other racial groups to secure places at Ivy League schools.

The Exodus No Press Release Could Hide

The damage did not appear all at once. It appeared in departures, early retirements, severance packages, and quiet resignations. Between 1994 and 2002, at least 1,027 white technical professionals left Eskom. Starting in 2000, the utility spent about 1.8 billion rand on voluntary severance and early retirement packages to accelerate the process. The corporate phrase was ‘creating space’. The human meaning was simpler: make room by emptying out the people who know how to do the work.

The union Solidarity, which represented many of those leaving, later described the pattern in blunt terms, saying Eskom had a long history of getting rid of the most skilled people. That accusation was not the complaint of people resisting all change. It was the warning from people who could see the breakdown of technical competence in real time.

The exodus continued. Between 2008 and 2018, Eskom lost 47 percent of its registered professional engineers. That is not a trimming exercise. That is gutting. Almost half the technical brain trust vanished in a decade. The veterans who had spent their lives around coal-fired boilers, switchyards, and control systems were pushed out, retired early, or went elsewhere. In many cases, they were replaced by accelerated learnership programs that emphasized compliance and abbreviated modules, not the kind of deep, stubborn apprenticeship that complex infrastructure requires.

The state told itself this was progress. In reality, it was institutional amnesia.

A power station is not just a place where people show up and switch things on. It is an ecosystem of memory. One technician remembers how a certain valve behaves under a particular load. Another knows which supplier’s replacement part lasts and which one fails early. Someone else can hear a turbine bearing beginning to go before the instruments register the fault. That sort of knowledge does not sit neatly in a file. It lives in the heads and habits of experienced personnel. Remove too many of those people and the organization loses more than headcount. It loses pattern recognition.

By the time the utility began to realize what had been lost, the loss itself had become the problem. Handover protocols collapsed because so many people were leaving at once. Knowledge transfer turned into an administrative fiction. There were still meetings, still memos, still tidy organizational charts. But inside the plants, the continuity was gone.

The Decay Inside the Plant

The first signs of decay were easy to miss if you were sitting in a boardroom. They were not easy to miss if you were the one trying to keep the boilers running.

Eskom’s internal audits showed coal plant availability dropping from 85 percent in 2008 to 68 percent within about a decade. Tube leaks on critical boiler piping rose from around 1,200 a year in 2009 to more than 3,800 by 2018. Weld failure rates on high-pressure piping tripled. Maintenance overtime jumped by 217 percent between 2012 and 2019. Planned outage completion rates fell below 37 percent. The average time needed to repair a high-voltage transmission fault stretched from 14 hours in 2005 to 67 hours by 2020.

Those numbers describe an institution becoming slower, more brittle, and less able to recover from ordinary wear. They also show a vicious chain reaction. A boiler tube leaks. An unscheduled outage follows. The crew assigned to the repair is less experienced than the old crew would have been. The repair takes longer and may not meet spec. The unit returns to service derated, not fully restored. The remaining units must work harder to carry the load. That extra strain accelerates wear on the rest of the plant. Another fault follows, then another, and the margin between control and crisis disappears.

This is how infrastructure dies in a democratic state. Not with one cinematic explosion, but with the steady accumulation of misjudgment, delay, and political self-deception. By 2007, the reserve margin had already fallen below 5 percent. That means the grid was running with almost no buffer. One failure could trigger another. A system that had once been planned with comfortable margins was now operating on a knife-edge.

Once that happens, language becomes part of the problem. Officials talk about turnaround plans, interventions, and transformation journeys. Contractors are called in at daily rates exceeding 25,000 rand per head. Temporary fixes are treated as a strategy. Repairs that should have been routine become specialized crises. The system is no longer being maintained; it is being managed by improvisation.

What makes this so destructive is that the deterioration was predictable. The engineers knew. The technicians knew. The unions knew. The figures in the internal audits told the story long before the country’s politics did. But once the utility had made demographic targets a central measure of success, acknowledging the operational trade-off would have meant admitting that the policy itself was doing damage. Bureaucracies do not enjoy that kind of confession.

Darkness, Damage and the Two Countries That Emerged

By 2008, South Africa had its first sustained nationwide blackouts. By 2014, the problem had hardened into stage three and stage four load shedding, meaning thousands of megawatts were being cut from the grid on purpose at peak times to prevent total collapse. By 2022, the country was losing more than 11,800 gigawatt-hours a year to load shedding. In 2023, the system broke in a way no one could pretend away: 24,638 gigawatt-hours were lost to deliberate blackouts, more than 100 times the 192 gigawatt-hours lost in 2018.

The human cost was immediate. There were weeks when parts of the country sat through up to 12 hours of darkness a day under stage 6 load shedding. Households bought candles, paraffin stoves, and portable generators. Families that could not afford backup systems cooked, charged phones, and tried to work around a public utility that could no longer guarantee basic service.

The economic cost was larger still. The Council for Scientific and Industrial Research estimated damage at 2.9 trillion rand in a single year, roughly $170 billion. The Reserve Bank said load shedding was slicing between 1.8 and 2.4 percentage points off quarterly GDP growth. The National Treasury later put cumulative economic damage above 5 trillion rand in lost output. Those are not abstract sums. They are the difference between investment and retreat, between expansion and shutdown.

The Chamber of Mines said gold and platinum operations lost 47 billion rand in revenue in 2022 alone because underground ventilation and pumping systems could not run continuously. Automotive assembly plants in the Eastern Cape reported a 22 percent year-on-year production decline. The Small Business Institute said about 47,000 small businesses failed between 2020 and 2023, many because refrigeration systems failed, payments could not be processed, or payroll could not be met in time.

The poorer the household, the harsher the burden. Statistics South Africa found that during peak periods in 2022, 84 percent of urban and peri-urban households were dealing with outages averaging 8.7 hours a day. Some families were spending as much as 12 percent of their monthly income on paraffin and portable backup just to preserve food and maintain a few hours of basic light. The Department of Basic Education estimated more than 2.1 million learners lost about 180 instructional hours a year because they could no longer study after dark. Provincial health departments recorded a 22 percent rise in preventable complications linked to power instability.

Then there was the other country inside the first one. In wealthier suburbs and commercial districts, rooftop solar surged by 2024. Anyone who could pay for a 150,000 rand installation could buy back a measure of normal life. The tariff had risen by roughly 190 percent over the decade, so the people who still relied on the public grid were paying more for a service that worked less. Those who could afford to exit the system left it. Those who could not were left holding the cost of the collapse.

That is the irony of elite-friendly crisis management. The wealthy build exits. The poor keep paying into a system that has already failed them.

The State Went Back to the People It Had Dismissed

By 2019, long before the blackout crisis reached its worst level, Solidarity had already assembled a database of former Eskom engineers and specialists who had been pushed out. The union offered to send them back. The list eventually reached the minister, and it was no small offer: 705 names, representing about 500 years of combined experience.

By 2022, the politics of refusal had run out of road. On 14 July that year, Pravin Gordhan wrote back to Solidarity. He thanked the union for the offer and spelled out what Eskom needed: power station engineers, mechanical and electrical specialists, nuclear expertise, system maintenance people, plus senior artisans and plant operators for coal and nuclear stations. He asked for the names to be sent so they could be deployed immediately.

That letter was an admission without the usual political honesty. The country needed exactly the skills it had spent two decades devaluing.

The final shortlist that went back to Eskom contained 300 turnaround specialists. Together, they carried 5,500 years of power-sector experience. Fourteen had PhDs in engineering or related fields. They held more than 400 accredited qualifications. Some were internationally sought-after project managers. A few offered to come back and work for free. About 70 percent of them were generation experts, the very kind of specialists needed most urgently.

This was not sentimental nostalgia for the past. It was a practical recovery of lost capacity. Eskom’s internal performance measures later suggested that each rehired veteran engineer restored an average of 180 megawatts of dependable capacity within 9 months. That is the sort of number that makes ideology look expensive. One experienced engineer can matter more than a dozen abstract strategy documents.

The political backlash showed how far the argument had drifted from reality. Gordhan was attacked from inside his own African National Congress for accepting help from a conservative, all-white union. A former parliamentary chair accused him of racism. The National Union of Mineworkers said he should stop interfering in Eskom’s recruitment. Yet the reality was stubborn. The grid was broken. The people who could repair it had been pushed out. The state had no choice but to go back and ask.

Recovery, But Not Redemption

By November 2022, Eskom had launched a digital skills crowdsourcing portal, a recruitment mechanism that sidestepped the demographic filtering that had drained the utility in the first place. Participation climbed to roughly two-thirds of engineers who had left after 2018. The movement was quiet but unmistakable. A country that had spent years pushing skilled people away was now trying, in a hurry, to pull them back.

The results were visible by 2025 and 2026. Eskom marked a full year without a single hour of load shedding as of 15 May 2026. The last rotational blackout had been implemented on 15 May 2025. Diesel costs at the utility had dropped 84 percent year on year. Energy availability had climbed from a humiliating low of 56 percent in late 2024 to around 67 percent in early 2026. System availability was reported at about 98.9 percent. President Cyril Ramaphosa said the country had ‘put load shedding behind us’.

But a working grid is not the same thing as a healthy state. The recovery did not erase the deeper problems. Cable theft and illegal connections were still costing about 17.5 billion rand a year from Eskom, with roughly 12 percent of the electricity generated disappearing into informal networks. In parts of Gauteng and KwaZulu-Natal, load reduction remained in place because local infrastructure had been overloaded by years of theft and deferred maintenance. The system was breathing again, but not yet healed.

Nor had the policy question gone away. In May 2026, the Democratic Alliance tabled the Economic Inclusion for All Bill, an attempt to replace black economic empowerment with a non-racial framework. Ramaphosa, meanwhile, doubled down on BEE with new employment equity targets across 18 sectors, tied to a 2030 deadline. Companies with more than 50 employees still had to file five-year racial workforce plans or face penalties. Even within the ANC, some senior figures had begun to say publicly that BEE had been captured by a politically connected elite. The party itself showed no sign of giving up the doctrine.

That matters because the grid story is not isolated. South African freight rail has already fallen by 34 percent since 2018. Wastewater treatment is in critical condition across 64 percent of municipal facilities. The pattern that damaged the electricity grid is spreading into other systems where competence matters more than speeches. The country fixed one broken machine by calling back the engineers it had dismissed. It has not yet answered whether it will do the same elsewhere before the next crisis becomes irreversible.

The Culture of Repair Was Still Missing

Even the recovery has a catch that matters to anyone who has spent time inside a large public system. Restoring a power grid is not the same as restoring the culture that allowed it to decay. A utility can restore generation capacity, buy diesel, hire contractors, and prevent the worst blackouts. It can even boast, with some justification, that the emergency is past. But if the organization still treats maintenance as a negotiable expense and technical memory as an afterthought, the next crisis is already being assembled in the background.

That is why the South African case should not be reduced to a triumphal story about turnaround. A system that needs to beg back its senior engineers has already admitted that its internal pipeline failed. A system that relies on emergency recruitment, outsourced patchwork, and heroics from retired specialists has not solved its underlying problem; it has merely interrupted the collapse.

The deeper failure was not just in hiring. It was in the way the state came to think about institutions. Public bodies were increasingly measured by political obedience, symbolic targets, and compliance language rather than by whether the lights came on, the trains ran, the taps flowed, or the sewage plants stayed within spec. Once that habit takes hold, every department begins to resemble every other department. The names on the organization chart change, but the same mistake repeats: technical work is treated as if it were less important than political theatre.

That logic also helps explain why the grid story resonated so widely. It was never only about electricity. It was about what happens when a country starts to reward the appearance of justice while neglecting the boring, expensive, difficult work of keeping systems intact. The people who repair turbines, valves, substations, and relays rarely win applause. They do not fit neatly into slogans. Yet without them, the slogans become a luxury no one can afford.

South Africa’s grid is therefore both a warning and a test. If the state can relearn the old discipline of competence, then the recovery may spread. If it cannot, then the power crisis will simply be remembered as the first and most visible collapse in a broader national unraveling. Either way, the lesson is already written into the record.

The Bill for Choosing Politics Over Competence

The lesson is older than South Africa and larger than one utility. Complex systems do not run on slogans. They run on people who understand the machines, their failure modes, the margins, and the small clues that signal a fault is coming before it arrives. You can dress the staffing policy up in any language you like — equity, inclusion, transformation, redress — but if the policy treats skill as secondary, the system will eventually disclose the price.

Other countries have already paid related bills. Venezuela learned it when the political remaking of PDVSA helped drag oil production reliability down despite massive reserves. Zimbabwe learned it when politically driven restructuring hit agricultural output and state capacity. South Africa has now provided the most complete example in the world of what happens when a state-owned utility is turned into a social engineering project and then expected to function like a power plant.

That is why the South African case is more than a local embarrassment. It is a warning. The country was not short of money, intelligence, or mineral wealth. It had deep capital markets, serious universities, and a sophisticated industrial base. What it lacked, for too long, was the discipline to put operational competence ahead of ideological reassurance. The bill for that choice has already run into trillions of rand, tens of thousands of lost businesses, and years of damaged public trust.

And the question is still open. If engineers can be brought back to a power utility after the collapse is visible, why should the same logic not apply to water systems, rail lines, ports, hospitals, and municipal services before they fail? South Africa has paid enough to know that infrastructure is not an arena for moral signaling. It is a test of whether the people running it know what they are doing.

The Musk-Lemon dispute looked, in the end, like another culture-war quarrel played out for cameras. South Africa gave it a harsher answer. It showed what happens when an ideology about who should be in the room is allowed to outrank the question of whether the room still functions. The lights came back on because the engineers came back. The harder question is whether the country has learned enough to stop repeating the same mistake in the next sectors that matter.