A decade of autonomous haulage at scale across the Pilbara has produced a story that is almost too neat. Utilisation rates climbed year after year, the technology matured, the financial case kept compounding. Through 2025 the slope of that curve looked like it would continue.

In the first quarter of 2026 it flattened. Across the three major Pilbara operators with knowable autonomous fleet data, utilisation gains have effectively stopped.

Where the bottleneck has moved

For the first six or seven years of autonomous haul at scale, the constraint was the technology — speed limits, conservative obstacle avoidance, slow ramp speeds. As those constraints eased, gains came easily.

The constraint is now the mine itself. Ramp geometries designed for human-driven trucks are not optimal for autonomous fleets. Bench widths, swap points, queueing locations, fuel logistics — all of these were designed around an operator model that no longer exists. The technology has outgrown the mine layout.

The interaction with electrification

The autonomous truck fleet renewal cycle now overlaps with the early electrification push. Battery-electric and trolley-assisted haul trucks behave differently from diesel autonomous units. Their charging and pantograph contact points are new constraint elements the previous fleet did not have.

Operations running mixed fleets — partial autonomous, partial electric autonomous, partial electric manual — are reporting utilisation behaviour that does not fit any clean model.

What is working anyway

The newer pits at two of the major operations have substantially reshaped their pit-floor logistics around autonomous truck flow patterns. Those operations are seeing utilisation comparable to what older pits reached three or four years ago — but only because the design was deliberate.

The retrofit path, where autonomous trucks are brought into a mine designed for manual operation, is approaching its ceiling.

What this means for new operations

If you are commissioning a new mine in 2026, the design choices that matter are not the obvious ones. Pit haul road geometries, swap points, maintenance bay locations, queueing zones, data infrastructure for fleet management — these determine whether the autonomous fleet runs at 70% utilisation or 85%.

The harder question

Underneath the technical plateau is a harder question about where productivity goes next. The easy gains from removing the operator from the cab are largely done. The next gains require systems thinking: re-engineering the whole logistics flow of the pit, not just the trucks themselves.

That is a different conversation. It requires mining engineers, autonomous systems specialists, and operations teams to work in a way that has not traditionally been part of the planning process.