RAW Charging | AI Transformation Index

Overview

RAW Charging deploys and operates EV charging infrastructure at hospitality, leisure, and retail destination sites across the UK and Europe, generating revenue through transaction-based fees paid by EV drivers at the point of charge.

Executive Summary

RAW Charging is a physical infrastructure play, not a cognitive-labor business — its durability against AI disruption is strong precisely because charging an EV requires physical assets that AI cannot replicate or bypass. The flip side is that AI transformation potential is genuinely low: growth is gated by capex and site acquisition, not by cognitive bottlenecks that AI could compress. For an infrastructure investor like Antin, the AI story here is at the margin — smarter maintenance, better pricing, lighter customer service — rather than a step-change value creation thesis; the investment case rests on infrastructure scarcity and EV adoption tailwinds, not AI transformation upside.

Durability Rationale

RAW Charging's durability is anchored firmly in the physical irreplaceability of its core function — no AI capability dissolves the need for electrons at a specific location, and the capex, grid infrastructure, and site partnership network constitute genuine barriers that software-native competitors cannot shortcut. The primary structural vulnerabilities are competitive (well-capitalised rivals with deeper balance sheets) and commercial (low end-user switching costs mean network effects are weak), not AI-driven. The moderate profit-pool migration risk from aggregator platforms and energy suppliers is real but manageable given RAW's asset ownership position.

Opportunity Rationale

AI's potential for RAW Charging is genuinely limited because the business is not a cognitive-labor business — it is a physical infrastructure business where value is created by steel, cable, and kilowatt-hours, not by analysts or knowledge workers. The AI-applicable functions (customer support, fault management, pricing optimisation, site analytics) exist but are peripheral to the core production function. There is no large manual cognitive workforce whose work AI could dramatically compress, no deep domain intelligence requiring proprietary AI amplification, and no throughput constraint solvable through automation rather than capital deployment.

Key Risks