Sibanye-Stillwater is testing whether robotics and AI can enable safe gold mining below 3,000m in the Witwatersrand Basin. CEO Richard Stewart described the effort at the Joburg Indaba in Johannesburg on Oct.
7-8, 2026. The Witwatersrand Basin is South Africa's main gold-bearing rock formation. Stewart called safe mining below 3,000m the "quantum change" needed to reach more high-grade gold, including extensions around existing mines, according to Mining Insight News in its trial report.
Table of Contents
- Why is 3,000 metres the barrier?
- What changed for robotics?
- How would robots improve safety?
- What should automation engineers watch?
- Why are the trials starting now?
Why is 3,000 metres the barrier?
Ultra-deep means rock that is hot, stressed, and prone to sudden failure. Virgin rock reaches near 50°C at 3,000m and above 60°C near 4,000m.
Heat demands extensive cooling and ventilation. High rock stress also raises seismic and rockfall risk for people at the face.
What changed for robotics?
South Africa's gold mines use narrow-reef mining. That means machines must work selectively in thin, tabular gold seams, often less than a metre high.
Conventional automation previously failed in those conditions. Stewart said rapid advances in robotics and AI have "suddenly made some of this look possible," according to Mining Insight News in its trial report.
How would robots improve safety?
The central benefit is removing workers from the most hazardous faces. Robots would take over deep extraction where heat and rock stress are worst.
That change would directly affect underground crews at Sibanye-Stillwater's South African gold mines. People would move to supervision, maintenance, and remote operation rather than physical work at the face.
What should automation engineers watch?
Useful trials must prove operation in three hard conditions: Endurance matters as much as dexterity. A robot that can cut or handle rock for hours without overheating is more valuable than a faster machine that needs frequent recovery.
- Thin stopes with limited clearance and visibility
- Sustained heat, humidity, dust, and vibration
- Seismic ground movement and falling rock without human rescue nearby
Why are the trials starting now?
Current mines are nearing their planned ends. Kloof runs to 2026 after geology cut reserves, Beatrix to 2031, and Driefontein, the deepest at about 3,300m, to 2036, though high prices may extend them. Sibanye-Stillwater has started funding trials, with Stewart reporting early results are "starting to come up quite positively." A successful test would not start immediate production; Stewart stressed safety will decide whether deeper resources are ever exploited, so it would only open "a whole different discussion" around the Wits Basin, according to MiningMX in its Joburg Indaba report.
