You’ve sat in the dark. You’ve thrown out the spoiled food in your fridge after Stage 6 load shedding. And now, the very utility that plunged your life into chaos wants you to believe it is the future landlord of the global AI boom.
Eskom is apparently sitting on a power surplus. Thanks to a combination of delayed economic growth and a few broken power plants finally sputtering back to life, they have extra megawatts. Their master plan? Sell that surplus to hyperscale data centers—the massive server farms running ChatGPT, AWS, and the entire global AI infrastructure.
It sounds brilliant on paper. Take the AI gold rush, hand it to a struggling state-owned enterprise, and watch the foreign investment roll in. But there is a glaring, multi-billion-dollar flaw in this strategy.
You can’t burn down a house for a decade and then sell the ashes as prime real estate.
The tech giants building these data centers aren’t selling sneakers. They are running mission-critical, availability-sensitive operations. A single five-minute outage costs them millions in lost revenue and breached contracts. They don’t just need cheap power—they need firm, uninterrupted, flawless power. They need a grid that never blinks.
Eskom’s entire brand is synonymous with blinking.
We all want this to work. We want South Africa to be the digital gateway of the Southern Hemisphere. We want the national pride of turning our historical misery into a strategic asset. But hoping global tech giants will anchor their AI infrastructure on a utility famous for rolling blackouts isn’t a strategy. It’s a delusion.
The hyperscalers know the truth. They aren’t going to bet their billions on Eskom’s promises. They are already planning to bypass the utility entirely. They will co-locate directly with Independent Power Producers (IPPs), build their own solar and wind farms, and install massive battery storage facilities. They will use Eskom’s grid only as a backup generator—paid for only when their own primary systems fail.
A power utility is only as valuable as its last blackout. And Eskom’s track book is a horror story.
If the tech giants build their own off-grid generation, South Africa might still get the physical data centers. We might get a few construction jobs. But the massive, ongoing revenue stream that was supposed to rescue Eskom? That vanishes. The utility becomes a marginal player in the very boom it hoped to host.
This isn’t just about a missed business opportunity. It’s a test of whether an emerging-market utility can convert a temporary operational breather into a real strategic advantage. Can Eskom prove it has actually changed? Or will years of broken promises keep the real money on the sidelines?
You don’t get to be the landlord of the digital economy when your own tenants have to bring their own flashlights.
Eskom doesn’t have a surplus of reliable power. It has a temporary pause in the crisis. Until they can prove they can keep the lights on for 365 days without a hiccup, the AI gold rush will pass them by. And they will find themselves exactly where they started: alone in the dark.
FAQ
Q: Can't Eskom just sign contracts guaranteeing uptime to data centers?
A: Hyperscalers don't care about penalty clauses; they care about physics. A financial payout doesn't stop a multi-million-dollar server crash. They will build their own power generation before trusting Eskom's grid.
Q: What does this mean for South Africa's economy?
A: If tech giants build off-grid, South Africa gets the physical infrastructure but loses the massive revenue stream that was supposed to save the national utility. It's a localized boom that leaves Eskom behind.
Q: Is Eskom's power surplus even real?
A: It's a temporary breathing room, not a structural overhaul. It's the result of delayed economic growth and emergency repairs, not a sudden mastery of energy generation.