The role of Data & AI within smart charging - takeaways from its panel event. The conversation around Electric Vehicle (EV) adoption has fundamentally shifted. In mature markets like the Netherlands, the question is no longer "Will I find a charger?" but rather "Will that charger deliver the power it promised, and is the grid ready to support it?" At its recent Amsterdam ClimateTech x AI, Enjins gathered founders, investors, and tech leads to discuss the topic of Data & AI in EV charging. Panellists from Fastned, Milence, and Chargetrip discussed the raw operational realities of scaling Europe's charging ecosystem. Here are the key takeaways from the event. Bouwe Neerbos March 18, 2026 From 99% to 100% uptime: closing the "Power Anxiety" Gap. For a premium Charge Point Operator (CPO) like Fastned, "good enough" is no longer an option. While the industry often talks about 99% uptime, that final 1% represents the difference between a seamless journey and a stranded customer. As Robin Wouters (Director of Product & Engineering at Fastned) noted during its panel: "The first 99% of uptime is easy; it's the last 1% that requires deep data integration and proactive engineering to ensure the driver never even sees a fault." This pursuit of perfection is happening as the primary consumer fear evolves. Range Anxiety - the fear of running out of juice - is being replaced by Power Anxiety. This is the frustration of arriving at a 400 kW charger only to receive 150 kW because of vehicle limitations, temperature, or grid curtailment. Addressing this requires bridging the "Expectation Gap" through better data transparency at the charger interface. Gideon van Dijk, CEO of Chargetrip, is tackling this head-on by moving beyond basic routing. By directing the right vehicles (and their owners) to the most compatible stations based on real-time site health and vehicle-specific charge curves, Chargetrip effectively solves both anxieties simultaneously. The scale of this influence is impressive: as of early 2026, Chargetrip's technology routes more than 20% of all European EV drivers, providing the orchestration layer necessary to match demand with actual power availability. Managing the grid: from static connections to intelligent systems. The "Power Anxiety" mentioned by Luis Hurtado (Milence) is rooted in a structural underinvestment in grid capacity. With local substations hitting their limits the industry is moving toward Local Energy Systems To navigate this, leading operators are deploying: * Peak Shaving via BESS: Using on-site battery storage to buffer high-power sessions without triggering expensive grid upgrades. * Cable Pooling & Dynamic Load Balancing: Sharing capacity between chargers or adjacent industrial loads. * On-site generation: Expand the sites with solar panels to generate electricity locally. Data and forecasting models play an integral role in the digitization and optimization of such local grids, to ensure more capacity is available than the grid connection allows. The logistics shift: predictability over pricing. For heavy-duty transport, logistics operators are sensitive to Total Cost of Ownership (TCO) and Service Level Agreements (SLAs). They don't need the cheapest kilowatt; they need a guaranteed 30-minute window at a specific power output to avoid penalties when being too late at their destination. For passenger cars, as Gideon van Dijk (Chargetrip) demonstrated, routing intelligence is evolving into a B2B optimization engine. By directing 30-40 GWh of energy demand annually, AI-driven routing can steer fleets toward underutilized sites, balancing the load across a network and preventing localized grid "brownouts." The three pillars of intelligence: data, ML, and Generative AI. Moving from a "real-estate play" to a high-performance energy business, like many EV companies, requires a sophisticated data & AI stack. Enjins categorize the innovation into three distinct pillars: I. The data foundation. To optimize physical assets, you must first master the real-time data they generate. This foundation requires a unified architecture where high-frequency telemetry is captured, stored, and made interoperable across the entire ecosystem. Beyond the charger itself, high-fidelity signals from vehicle OEMs, real-time weather forecasts, and grid pricing are critical for optimization. Currently, the industry suffers from "siloed data storage" between OEMs, CPOs, and grid operators. The strategic winner will be the player that successfully integrates these disparate streams to build a comprehensive data set that is a fundament for further optimization and automation. II. Machine Learning. Machine Learning is the workhorse of operational efficiency, moving networks from reactive to predictive. Use cases are scaling rapidly, from individual site physics to network-wide orchestration, like: optimally charging and discharging your BESS to ensure energy capacity at the right time. Or predictive maintenance through training models on historical charger log files. An ML algorithm can identify signature patterns that precede a hardware failure (e.g., a cooling pump slowing down). This allows a CPO to fix the charger before a driver arrives to find it broken. III. Generative AI. While ML optimizes the "under the hood" mechanics, Generative AI is transforming how users and operators interact with complex energy data and systems. Concrete Example: Contextual Route Enrichment. Chargetrip showcased how LLMs allow users to query routes using natural language, for example, "Find a route for a heavy truck that includes overnight parking certified for dangerous goods and complies with EU rest regulations." This turns a complex multi-variable optimization problem into a simple conversation. Other use cases resides around the opportunity that charge point operators can automate human-intensive processes around deploying new sites: from legal to finance, and from design to project realization. "We are all in the same boat. Sometimes it feels that we are solving unique issues, but challenges are common in the industry" Luis A. Hurtado Munoz Director of Technology

We are driven by impact, and so is our vision on ML. We don’t want a data scientists’ laptop to be a graveyard for brilliant models.