EV Charging Services: From the Local Vehicle Fleet to Site Economics

A motorway charging site, an apartment car park and a delivery depot serve different journeys. Their customers arrive at different times, have different alternatives and buy different services. A growing electric-car fleet matters to each, but it becomes a useful investment signal only after those local conditions are understood.

The analysis below follows a defined vehicle fleet into charging occasions, delivered energy and site economics. It separates operators, installers and software providers so that the demand model matches the business being considered. The evidence window ends in December 2025, and the site calculation is an illustrative scenario.

The vehicle fleet behind the charging catchment

The IEA's Global EV Outlook 2025 reports that electric-car sales exceeded 17 million in 2024 and represented more than one-fifth of new-car sales. The report includes battery electric and plug-in hybrid electric cars. These are historical sales observations published in 2025, rather than completed full-year 2025 results released later. The unit is vehicles sold during a year; it is not charging energy or supplier revenue.

Scroll horizontally to compare all columns.

Historical observations from the IEA's 2025 edition. No exact residual sales share is calculated from a rounded lower-bound statement. Full-year 2025 outcomes fall outside the evidence available by this edition's cutoff.

The sales count, sales share and accumulated fleet answer different questions. Charging services may respond to vehicle adoption, but their units can be sites, installations, managed ports, sessions, kilowatt-hours, subscriptions or supplier revenue. A large vehicle inflow does not establish how much charging a specific operator or contractor will sell.

Vehicle mix matters as well. A battery electric car relies entirely on charging for traction energy, while a plug-in hybrid can operate using electricity and liquid fuel. Battery capacity, charging power, efficiency, annual distance, and driver behaviour vary within both groups. Counting every electric car as the same charging load would discard a definition the source makes explicit.

The headline is therefore a reason to investigate charging demand, not a conversion factor. Preserve the original geography, period, vehicle class, powertrain coverage, and measurement unit before adding any commercial assumption.

Annual sales must become an in-use fleet before they become demand

Charging demand comes from vehicles operating during the period, rather than only vehicles newly sold in that period. The IEA's 2025 report puts the end-2024 global electric-car fleet at almost 58 million, around 4% of the passenger-car fleet. That stock measure is separate from the new-sales share and accumulates sales cohorts after retirements and other movements. A local study still needs an active catchment fleet.

A local stock bridge normally starts with registered or active vehicles at the beginning of the period, adds new and used-vehicle inflows, and removes scrappage, exports, deregistrations, and long-term inactivity. It keeps battery electric and plug-in hybrid vehicles separate where their energy behaviour differs. Commercial vans, taxis, ride-hailing vehicles, and heavier vehicles may need their own frames because a passenger-car sales series does not describe them.

Geography also changes across the bridge. A car can be sold in one administrative area, registered in another, and regularly charge near a workplace or corridor elsewhere. National sales may be useful for a national infrastructure question but too broad for a site catchment. A city registration count may still miss through-traffic. Match the vehicle population to the service area rather than choosing the most available headline.

Time creates another distinction. A late-December sale contributes almost no vehicle-days to that calendar year's charging demand. A model that multiplies full-year sales by a full-year energy value overstates the first-year contribution unless it uses an average in-service fraction. Cohort the additions by month or apply a clearly supported timing adjustment, then carry surviving cohorts into later years.

Choose the service, customer and infrastructure unit

“Charging services” can describe several revenue pools with different buyers, units, and economics:

• a charge point operator may receive driver or fleet payments for energy, time, occupancy, or a session;
• a site host may receive rent, a revenue share, parking income, or increased retail activity;
• an installer may earn project revenue for civil, electrical, grid, and commissioning work;
• a maintenance provider may earn scheduled and corrective service fees;
• a network service provider may earn software, connectivity, transaction, or managed-service fees;
• a mobility service provider may manage customer access, payment, subscriptions, or roaming;
• an equipment supplier may earn hardware revenue; and
• a utility or energy retailer may earn regulated or contracted electricity revenue.

These amounts can sit in the same transaction chain. Adding the driver's full payment to the operator's revenue, the electricity supplier's charge, the payment processor's fee, and a software provider's fee can count one session several times. Decide whether the model measures end-customer spending, gross ecosystem transactions, net supplier revenue, or the obtainable revenue of one business. Use that boundary consistently.

Separate new-build work from installed-base services. Hardware and installation are generally tied to additions, replacements, and upgrades. Software, networking, maintenance, roaming, and energy delivery may recur while ports remain active. A sales forecast for new vehicles does not state either the number of new charging projects or the installed base under management.

The counterparty must be named too. A household buying a charger, a property owner commissioning shared parking infrastructure, a fleet procuring a depot, a government awarding a corridor concession, and a public operator buying network software are different markets. Their budgets, procurement cycles, contract values, and supplier sets cannot be represented by one undifferentiated “EV charging” multiplier.

Count locations, ports, connectors, and power consistently

Infrastructure counts are definition-sensitive. The U.S. Department of Energy's Alternative Fuels Data Center distinguishes a station location, an EV charging port, and a connector. A location is a physical place. A port can charge one vehicle at a time even if it offers multiple connector types. A connector is the part plugged into the vehicle. Counting plugs as simultaneously usable ports can overstate service capacity.

Power and energy are different measures. Kilowatts describe an instantaneous rate; kilowatt-hours describe energy delivered over time. A 150 kW label does not mean a port continuously delivers 150 kWh every hour. The vehicle's acceptance rate, state of charge, shared site capacity, temperature, power management, downtime, and session profile can reduce realized output. Nameplate capacity is not throughput.

The UK Department for Transport's July 2025 release counts publicly available charging devices and distinguishes local geography and power categories. A device inventory, a simultaneously usable port count and a connector count have different denominators. Use the unit actually defined in the release and verify its fit to the operator dataset. This edition retains the July 2025 device series and does not substitute later revisions or a different charger-count definition.

Inventory status also needs a rule. Installed, commissioned, operational, publicly accessible, discoverable, available at the observation time, and used during the period are different states. A funded or announced site is not current capacity. A port that appears in an inventory but is persistently unavailable cannot support the same revenue as a working port. Preserve status dates and source methods.

Where and when the target vehicles charge

The same electric vehicle can create very different public or commercial demand depending on where it is parked and how it is used. A useful segmentation begins with the charging occasion rather than the powertrain label alone:

• Home-access vehicles may obtain most routine energy from a private connection but still use corridors or destinations.
• Drivers without reliable off-street access may depend more heavily on public residential or nearby charging.
• Workplace users may shift energy into long daytime dwell periods when charging is offered.
• Depot fleets can have concentrated, scheduled demand and may buy private infrastructure, software, maintenance, and energy services.
• High-mileage commercial vehicles may generate more energy demand per vehicle but require high availability and carefully located capacity.
• Corridor travellers need coverage and speed at particular route nodes rather than an even geographic spread.
• Destination users trade charging speed against dwell time at retail, hospitality, leisure, or parking sites.
• Plug-in hybrid drivers may use electricity differently from battery electric drivers and retain a fuel alternative.

Do not assign a universal public-charging share to each group. Housing form, parking rights, electricity tariffs, employer provision, fleet return-to-base patterns, road network, climate, vehicle efficiency, trip length, local price, and charger reliability all affect the choice. Evidence from one city or early-adopter sample may not transfer to another market.

The commercial denominator is often smaller than the total fleet. A corridor operator needs vehicles whose journeys pass the catchment and whose drivers are willing and able to use that site. A depot installer needs fleets approaching an investment event. A software supplier needs compatible ports whose owners will procure external management. Define “reachable vehicle” or “addressable port” for the specific offer.

Translate vehicle activity into energy and sessions

A demand-side energy model can be written as:

relevant in-use vehicles × distance per vehicle × energy per distance × share delivered by the target charging channel

Every term needs the same geography and period. Distance should represent the selected vehicle segment, not all cars when the target is a delivery fleet. Energy intensity should reflect the vehicle mix and conditions. The channel share should distinguish home, workplace, depot, destination, street, and corridor charging as required. Losses need a stated treatment so energy drawn from the grid is not confused with energy stored in vehicle batteries.

A session model uses a different bridge:

active ports × available days × sessions per available port-day × delivered energy per session

This model is useful for an existing or proposed network. “Available” should exclude relevant downtime, while sessions per day should use comparable sites and maturity. An average that combines newly opened, temporarily closed, destination, and highway sites may not describe any of them. Show the distribution or cohorts when utilization is skewed.

Reconcile the vehicle and site views. If catchment vehicles imply 10 GWh of annual target-channel energy but the planned sites imply 25 GWh at the assumed utilization, at least one input or boundary is wrong. If existing operator data show energy that the vehicle model cannot support, investigate through-traffic, fleets, duplicate vehicle counts, or a channel-share assumption before accepting the larger total.

Sessions, energy, and occupancy also answer different capacity questions. Two short sessions and one long session may occupy a port for the same time but deliver different energy. Revenue based on kilowatt-hours responds differently from revenue based on time, parking, a fixed session fee, or a subscription. Choose the activity measure that matches the charging contract.

Match demand with competing capacity

The table shows why a global sales observation cannot finish a provider forecast. “Available” means an official source supplies the stated indicator for its own scope. It does not mean that the indicator has the geography, customer, or commercial boundary required by a particular decision.

Scroll horizontally to compare all columns.

The unavailable cells are the main research agenda. Replacing them with an assumed percentage of global sales produces a precise number without an evidenced commercial path.

Treat public charging counts as supply evidence

The IEA's 2025 charging chapter describes more than five million public charging points at end-2024, with more than 1.3 million added during 2024 and growth above 30%. These are infrastructure observations available in 2025. They do not supply site-level energy sold, sessions, uptime, margins or an operator's contestable customer base.

A point count treats different power levels, locations and access conditions alike. A vehicle-to-point comparison can describe supply density only when its fleet and infrastructure definitions match. Adding rated power refines the capacity description, but still does not establish when drivers arrive or whether energy is sold. Treat both as infrastructure context, then measure operating performance separately.

Global averages also conceal geographic concentration. A network can be dense in one market and sparse in another. Within a city, ports can cluster where land and grid connections are available rather than where unmet demand is strongest. A country ratio cannot decide between two sites or determine whether an existing operator has pricing power.

Supply growth has two opposing commercial effects. It can enable more charging and expand the managed or maintained asset base. It can also divide activity among more ports, increase competition for sites, or leave young locations below viable throughput. Model demand, capacity, and competitive additions together instead of treating every new point as proof that every operator's revenue will grow.

Utilisation, tariffs and site operating costs

The Alternative Fuels Data Center's operation and maintenance guidance identifies electricity, maintenance, networking, fee design, access structure, and utilization data as operating considerations. This is the bridge from physical availability to a commercial service, but “utilization” still needs a defined numerator and denominator.

Useful measures include sessions per available port-day, delivered kWh per available port-day, occupied minutes as a share of serviceable minutes, peak coincident load, successful session rate, and revenue per port. A power-based utilization percentage and a time-occupied percentage are not interchangeable. Record whether the denominator includes downtime and whether the observation covers a day, month, season, or full year.

Reliability and utilization should remain separate. A working site with low customer demand has low utilization. A frequently attempted site with failed sessions can show poor delivered utilization despite latent demand. A site closed for grid or equipment work has less available capacity. Combining all three conditions into one percentage can point management toward the wrong remedy.

Site maturity matters. A newly opened location may ramp as drivers learn it exists, vehicles accumulate, or nearby amenities develop. A mature corridor site can behave differently from a residential curbside portfolio. Compare like-for-like cohorts and show the distribution; a network-wide mean can conceal both congested sites and stranded assets.

A worked four-port site example

Consider a fictional public-charging site with four ports. Assume each port serves four paid sessions per day, each session delivers 35 kWh, and the site operates for 365 days. Annual delivered energy is:

4 ports × 4 sessions × 35 kWh × 365 days = 204,400 kWh

At a fictional realized customer price of EUR 0.40 per kWh, annual charging receipts are EUR 81,760. At a fictional electricity and network cost of EUR 0.20 per delivered kWh, that cost is EUR 40,880. The remaining EUR 40,880 is only a spread before maintenance, software, payment processing, site rent, staffing, insurance, taxes, depreciation, financing, downtime, charging losses, and other costs. It is not gross margin under a standard accounting definition and it is not operating profit.

Every input is invented solely to demonstrate the calculation. None is a benchmark for a real country, operator, charger type, tariff, or site. The example does not use the IEA sales rate or public-point total.

The sensitivity reveals the research priority. If sessions per port-day are two rather than four, annual delivered energy and receipts halve to 102,200 kWh and EUR 40,880, while many fixed costs do not halve. If sessions rise to six with all other assumptions unchanged, receipts increase to EUR 122,640, but queues, power sharing, site capacity, and service quality may need to be retested. Utilization and realized price can carry more of the result than the global vehicle-sales rate.

Do not multiply nameplate power by every hour of the year and call the result sales. That would calculate a theoretical ceiling, not expected energy. Use sessions or an explicitly defined load profile, then check that coincident demand is compatible with the site's grid connection and operating rules.

Separate operator, software, installation, and maintenance revenue

The Alternative Fuels Data Center's network definitions distinguish station operators from network service providers. It describes network software that can support access, fee setting, health monitoring, utilization data, station discovery, session initiation, and payment. One company may perform several roles, but a market model should still identify which function earns each amount.

For an operator, a simplified revenue bridge may use paid kWh, session fees, subscriptions, parking or occupancy charges, and roaming settlements. For a software provider, the bridge may use compatible managed ports, externally purchased software incidence, monthly fee, transaction share, contract term, and churn. For an installer, it may use funded new or upgrade projects, externally awarded work, project scope, net fee, and delivery capacity. For maintenance, it may use serviceable installed assets, coverage level, annual fee, failure work, and renewal.

Do not add the full value of all four models and call it one supplier's opportunity unless that supplier earns each revenue stream and internal transfers are removed. Exclude pass-through equipment and energy when the decision concerns net service revenue. If a prime contractor pays a subcontractor, decide whether the analysis measures the end-customer contract or supplier-layer revenues so the same project is not counted twice.

Obtainability adds a final filter. A growing installed base can be locked into another network, covered by warranties, maintained internally, incompatible with the product, or unavailable until contract renewal. A large theoretical pool may produce few procurement events in the forecast period. Track addressable assets, contract timing, channel access, win rate, implementation capacity, and ramp separately from total infrastructure.

From charging activity to a reachable contract

The European Commission's Alternative Fuels Infrastructure Regulation overview describes national fleet-based public power-output targets, including at least 1.3 kW for each battery electric light-duty vehicle and 0.8 kW for each plug-in hybrid light-duty vehicle. It also describes distance-based coverage and requirements concerning interoperability, payment, price transparency, and user information.

These rules are relevant to the amount, location, specification, and operation of infrastructure. They do not establish a supplier's revenue. A target may be met by privately financed sites, public support, utility investment, an incumbent's balance sheet, or a combination. It does not say when a specific project will be tendered, what the eligible service boundary is, which provider will win, or how much drivers will use the completed capacity.

Keep policy stages distinct. A target is not an allocated budget; an announced programme is not an award; an award is not an operational site; an operational site is not a utilized site; and a utilized site is not necessarily profitable. Evidence can move through those stages, but it should not be counted at all of them.

Policy can also change costs and service design. Payment hardware, price disclosure, interoperability, data, accessibility, uptime, cybersecurity, or reporting requirements can create implementation and software work while increasing capital or operating expense. Model the obligation and the potential revenue on their own lines instead of treating compliance as a free demand accelerator.

Collect local evidence that links vehicles to a buying event

A defensible research programme closes the bridge in a deliberate order:

1. Fix the commercial boundary. Name the service, buyer, supplier layer, geography, vehicle classes, charging channels, revenue unit, and forecast period. State pass-through and tax treatment.
2. Build the active fleet. Reconcile registrations, sales cohorts, used-vehicle movements, retirements, and relevant through-traffic. Separate battery electric, plug-in hybrid, passenger, commercial, and fleet groups as required.
3. Measure charging access and behaviour. Sample home, street, workplace, depot, destination, and corridor access. Gather distance, energy, session, dwell, timing, and willingness-to-pay evidence for the target segments.
4. Audit existing and planned supply. Verify locations, simultaneously usable ports, connector compatibility, power, grid status, opening date, accessibility, pricing, uptime, and competing projects. Deduplicate data sources.
5. Obtain operating distributions. Use cohort-level sessions, kWh, occupied time, successful-session rate, downtime, seasonality, realized receipts, and ramp. Protect customer privacy and distinguish observed data from management targets.
6. Model site and network economics. Record equipment, installation, grid, rent, electricity, demand charges, networking, maintenance, payment, staffing, financing, tax, and replacement on a consistent basis. Stress the inputs that carry the result.
7. Verify procurement and capture. Identify budget owners, tender or renewal dates, internal delivery, approved vendors, contract values, compatible assets, sales access, win assumptions, and delivery capacity.
8. Reconcile demand and supply. Compare vehicle-derived energy with site-derived throughput and operator evidence. Explain gaps rather than selecting the most favourable estimate.

Sampling must fit the decision. Surveys of drivers who already use one network can overstate public-charging reliance. Operator data can exclude failed attempts or competitors. App inventories can lag openings and closures. Registration data may not show where vehicles travel. Use multiple sources, publish the observation period and denominator, and retain uncertainty by segment.

Update volatile inputs on their own cadence. Sales and registrations may update monthly or quarterly; tariffs and prices can change faster; site openings, availability, and contracts may be event-driven. A model should not acquire a new publication date merely because unchanged source data were rebuilt.

Use decision rules tied to the service model

For a proposed public site, proceed only when the addressable catchment, expected ramp, utilization distribution, price, grid connection, site control, competition, and downside economics support the investment threshold. A national sales increase cannot rescue a site whose traffic, dwell, power, or access is wrong.

For an operator entering a region, require a credible portfolio of sites and contracted access, not merely a large vehicle forecast. Test whether the network can achieve coverage, reliability, customer acquisition, roaming, and operating scale before capital is committed.

For a software or managed-service provider, require compatible assets, an external-buying path, decision dates, supported fees, retention, and implementation capacity. Ports already controlled by an incumbent should not be treated as immediately contestable.

For installers and maintainers, distinguish funded projects from policy targets and announced pipelines. Reconcile implied job counts with qualified procurement, labour, equipment, permitting, grid lead times, and the provider's delivery capacity.

Run a bounded pilot when demand is plausible but site-level behaviour is uncertain. Define the minimum sessions, energy, successful-session rate, contribution, and learning period in advance. Stop or rescope when the case relies on multiplying global electric-car sales by an assumed charging spend, uses connectors as simultaneous ports, treats nameplate power as annual energy, or combines ecosystem spending with supplier revenue.

What the evidence supports within the 2025 evidence edition

The evidence available in 2025 shows substantial expansion of electric-car sales and public charging infrastructure during 2024. It also supports a clear distinction between the annual sales flow, accumulated vehicle fleet, station locations, ports, connectors, rated power, delivered energy and service-provider roles. It does not establish completed full-year 2025 outcomes.

The evidence reviewed here does not establish annual charging-service revenue, site utilization, profitability, an installer's project pipeline, a software provider's addressable ports, or any company's obtainable share. Those outputs require local fleet and travel data, channel behaviour, observed operating distributions, realized prices and costs, procurement timing, competition, and delivery capacity.

Use the automotive research hub to frame the market context, document the missing evidence with the research scope worksheet, or scope a custom study when a vehicle-sales headline cannot answer the site, network, or supplier decision.

Choose the catchment before the forecast: operating vehicles, trip patterns, access to home or depot charging, and competing sites. Then test paid energy and contribution under the site's actual commercial terms. An installer should forecast commissioning projects; a network-software provider should forecast contracted assets. Each business can use the same fleet evidence while arriving at a different opportunity.