Where AI genuinely helps charging and fleet operations, and where it does not

AI is easy to oversell. For charging and fleet operations, the useful question is narrower: which decisions are repetitive, data-rich and recoverable if wrong? Those are where software earns its place. The others need a person in charge.

Where it works today

The IEA's Energy and AI report says AI can improve the forecasting and integration of variable renewable generation, and that AI-based fault detection can help rapidly identify and pinpoint grid faults. In oil and gas, it notes AI being used to optimise and automate production processes and to predict maintenance needs.

The IEA's 2026 report Modernising Grids in the Age of Electricity explains why these uses lead: lower-risk applications such as forecasting, maintenance, inspection and planning tools are scaling first because they improve decisions and workflows without directly controlling the power system.

Carried over to charging and fleets, that points to three honest uses:

• Forecasting demand and arrivals, so sites and depots are not caught short or over-built.
• Scheduling charging around a fleet's routes and the grid's limits.
• Spotting equipment faults early and planning maintenance, rather than waiting for a failure.

On the fleet side, the IEA's truck analysis observes that logistics operations follow predetermined schedules, and says lifting en-route charger utilisation may require approaches such as adaptive route planning.

Where it does not

The IEA is direct about the limits. Modernising Grids says accuracy is necessary but not sufficient: models must also be explainable, auditable and secure. It says the operator retains authority and accountability throughout, and that automated switching or self-learning operation requires validated limits, cybersecurity assurance and clear accountability before it can be trusted in safety-critical settings.

The IEA expects that, in the near term, AI is more likely to support engineering judgement than replace it. That is a sensible default for a charging operator too: software proposes a schedule or flags a fault, and a person who understands the site decides what to do with it.

The unglamorous constraint is data

Both IEA reports return to data. Energy and AI lists missing or inadequate access to data and digital infrastructure and skills as barriers, and Modernising Grids says models depend on data quality and continuity in ways that static hardware and software do not. For a charging network, that means reliable status from every charger, which brings the discussion back to open, well-implemented protocols.

Security belongs in the same sentence. Energy and AI notes that persistent digital and physical security concerns can outweigh potential efficiency gains, and Modernising Grids observes that operational data is sensitive. A system that needs every charger's data also needs a clear answer about who can see it and where it is kept.

Scalable is the group's AI software company — AI, automation and web systems. This is a general explainer. It describes no BFCEV site, truck, network, customer or result.