Johannesburg’s City Power plans to introduce AI to tackle electricity billing and losses

Impact level

⚪ Background

⚠ Risk:A procurement plan does not establish that deployment will succeed, and it is not enough to conclude that electricity tariffs or supply reliability will improve.

What happened

According to a report published on October 12, 2026, Johannesburg’s municipal electricity utility, City Power, is recruiting external experts to use AI to check for irregularities in electricity sales, meter data, and collections reconciliation. Suppliers must integrate with at least two major banks, two retail groups, and cash-based electricity vending machines. Audit materials show annual local electricity losses of about 5.7 billion rand; the deployment and go-live dates and procurement amount are unknown. [1]

Involves:electricity meter、cash-based electricity vending machine

Seller impact

• InferenceAt the city finance level, reducing uncollected revenue could improve the utility’s cash recovery. But the R44 billion infrastructure backlog shows that billing fixes alone are unlikely to resolve network bottlenecks; whether any savings will be passed on through electricity tariffs is unknown. [1]
• InferenceFor local electricity-consuming businesses, anomaly detection and reconciliation could reduce cash-flow disruptions caused by billing errors and disputes. The impact depends on meter data quality and access to the relevant channels; it cannot be extrapolated to sellers in other countries. [1]

Evidence that would disprove the assessment

• Assessment 1：Check procurement progress over the next two weeks. If the proposal has not entered implementation, expectations of near-term improvement in collections will not hold.
• Assessment 2：After implementation, if billing error rates and dispute resolution times do not fall, or false positives increase, the assessment of operational improvement will not hold.