Last month, energy regulator Nersa published a consultation paper on a two-year Eskom pilot tariff for “cryptocurrency mining and other flexible-load customers”, with a benchmark price of about R1.20/kWh. Comments close on October 23, public hearings follow on November 3 and Nersa is due to publish its decision on December 7.
The proposal exists because Eskom now has more power than it can sell at certain times of the day. It estimates that the surplus could reach 5-7GW at times, mostly during the day, when solar output peaks, and late at night, when demand falls away.
Eskom says more than 5TWh of energy could be curtailed in 2028. Curtailment means telling wind and solar plants to stop generating because nobody needs what they produce, and it isn’t free: under the government’s renewable energy procurement programme, certain contractual costs remain payable even when output is curtailed.
Eskom’s sales, meanwhile, fell 6.2% to 178TWh in FY2026. The modelling behind the pilot lists rooftop solar, wheeling and the termination of the Mozal aluminium smelter’s supply agreement among the forces reshaping the load.
Yet the same paper cites projections showing a significant decline between 2026 and 2030 in firm capacity, the power that can be relied on at any hour. South Africa may have too much electricity at noon and still be tight at 6pm.
What Eskom needs, then, is a customer that will take large amounts of power when there is too much of it and get off the grid quickly when there isn’t. Most big industrial users can’t do that. Factories and smelters are built to run around the clock, and stopping them is expensive. Bitcoin miners are among the few that can.
A bitcoin mine is a warehouse of specialised chips that do nothing but crunch the cryptographic puzzles that secure the network. It can also be switched off almost instantly when the grid needs the power back, and in the US that flexibility has become a business. In the first quarter of 2026, Riot Platforms earned $21m in power credits, partly by selling contracted power back to the grid and partly through demand-response programmes in Texas and the Midwest, taking its all-in power cost to US3c/kWh.
Measured against that, not every South African miner thinks R1.20/kWh is cheap enough. One told TechCentral the price must fall below $0.05/kWh, about 82c, for the numbers to work.
Two mining companies have expressed interest, with combined initial demand of about 10MW and potential expansion to 500MW. Even the optimistic figure is a tenth or less of the surplus Eskom describes, so bitcoin on its own won’t solve curtailment.
That makes the second question in Nersa’s paper the most important one. It asks whether crypto mining should be treated as a distinct customer category, or whether the tariff should be open to any customer that meets defined technical and operational requirements.
Nersa should choose the second option. Data centres, green hydrogen electrolysers, cold storage and electric vehicle fleet depots can all shift some demand into hours of abundance, and a tariff built around how customers use power, whatever business they are in, would be harder for any single industry to capture and open to all of them.
Eskom needs a customer that will take large amounts of power when there’s too much, and get off the grid when there isn’t, but a miner could still decide the peak rate is worth paying.
The paper says only that participants “may be required” to take part in demand-response programmes, and asks stakeholders whether telemetry, response times and penalties for nonperformance should apply. All three should be mandatory. Eskom is offering a discount in exchange for flexibility, and it needs to know the load will come off when the system is under strain.
The R1.20/kWh benchmark applies to the highest-voltage customers in the southern region, and the pilot would adjust zone differentials to steer new flexible load there. In effect, it uses price to bring demand to the power. Moving power to where demand is requires transmission lines, and those are running late.
On October 1, the National Treasury and the department of electricity & energy pushed the final request for proposals for phase one of the Independent Transmission Projects programme, which covers about 1,164km of privately built lines, to the second quarter of 2027. It had been promised by the end of the third quarter of this year.
The reasons for the delay are defensible: bidders want more clarity on bankability, and the credit guarantee vehicle meant to support the projects is still being set up. But the timetable has slipped again.
So, by all means, Nersa should let the bitcoin miners in to soak up extra capacity, provided it’s on strict terms and with no cross-subsidy from other customers.
McLeod is editor of TechCentral
