The charging network runs on data, not batteries

Since 24 November 2024 it has been unlawful to operate a public charge point in the United Kingdom without telling the world, continuously and for nothing, whether it is currently working.

That obligation sits in regulation 10 of the Public Charge Point Regulations 2023, and the wording is worth reading slowly. Reference data and availability data must be “made available to the public free of charge and in a machine readable format without any requirement to agree to terms and conditions regarding the use of that data”. Status must be “updated within 30 seconds of a change from one EVSE object status to another EVSE object status”. Location must be recorded before a member of the public uses the charger for the first time.

Set aside the vehicles for a moment. What the British state has actually built here is a mandatory, real-time, licence-free national feed describing the operating condition of tens of thousands of privately owned machines. There is no equivalent for petrol pumps, cash machines, lifts in railway stations or public lavatories. It exists for charge points because a driver who arrives at a broken one is stranded, and because nobody could work out a way to make that risk bearable other than forcing the network to publish its own faults.

The standard was named in the regulations, which is rarer than it sounds

Most open data obligations in British law describe a format in the abstract and leave the implementation to a working group. This one did not. The Department for Transport’s guidance on the regulations mandates the Open Charge Point Interface, and specifies the objects operators must expose down to the paragraph number of the OCPI 2.2.1 specification: the Location object, the EVSE object and the Connector object from paragraphs 8.3.1 to 8.3.3, plus the Tariff object from 11.3.1.

Naming a version of a third party specification inside a statutory instrument is a decision with consequences. It means an operator cannot satisfy the duty by publishing a spreadsheet it invented, and it means anyone building a route planner can write one integration rather than forty. It also means the regulation ages, because OCPI will move on and the guidance will have to move with it. That is a real maintenance liability, and it is a better problem to have than a market of forty incompatible feeds.

The commencement schedule tells you what the department thought was hardest. Pricing transparency applied immediately from 24 November 2023. Contactless payment, the 99% reliability standard, the helpline and open data all took a year. Payment roaming, which requires operators to accept each other’s customers, took two, and only landed on 24 November 2025.

Reliability is a number each operator calculates about itself

Regulation 7 requires an operator’s network of rapid charge points to be 99% reliable, on average, across each calendar year. Regulation 8 requires a report to the Secretary of State for each calendar year, with total charge points, network reliability as a percentage, and individual charge point reliability including location and time in each status. The first of those reports, covering 2025, was due by 14 January 2026.

The calculation is set out in the guidance and it has a term in it that decides everything:

Reliability % = (((M - mean time exempt) - mean downtime) ÷ (M - mean time exempt)) × 100

M is annual minutes. Mean time exempt is the period the operator is allowed to discount, for causes outside its control. Two operators with identical hardware and identical outages can report different reliability if they take different views of what counts as exempt, and nothing in the regulations makes those views comparable. The 99% figure is a compliance threshold, not a measurement of the experience of arriving at a charger.

Operators must also publish their compliance on their own websites, which is the part that quietly defeats the purpose. The information is public in the sense that it exists on forty-odd corporate websites in forty-odd formats, at forty-odd URLs that change. It is not public in the sense that anyone can compare it. There is no national reliability table. Enforcement sits with the Office for Product Safety and Standards, and the civil penalty for most breaches runs to £10,000, rising to £250,000 for obstructing enforcement.

A £10,000 maximum against a network worth hundreds of millions is not a deterrent. It is a filing fee. The real enforcement mechanism in this regime is the 30-second availability feed, because a network that lies about reliability in its annual report is contradicted, minute by minute, by data it is separately compelled to broadcast. Whether anybody is systematically archiving that feed to check is a different question, and the answer as far as this desk can establish is that no public body is.

The national statistic is bought, not collected

Here is the part that should trouble anyone who plans capital against these numbers. The government’s official count of the British charging network is assembled by a private company.

The Department for Transport’s statistics for 1 April 2026 record 119,080 public EV chargers in the United Kingdom, against 92,141 charging devices, with 27,372 chargers rated at 50kW and above. The data is collated by Zapmap, which the department’s methodology note says compiles operator data “with coverage of over 95% of the public charging network”. The series carries the “official statistics in development” label, which means it is being tested with users rather than accredited.

Two things follow. The first is definitional. Until 1 January 2026 the headline unit was the charging device, a physical unit that may hold several sockets. From that date the series switched to counting chargers, and the two numbers differ by around 27,000. Anyone comparing a 2025 press release with a 2026 one without checking which unit is in play will produce a growth rate that is mostly an accounting change. The second is structural. The state legislated to force this data into the open, then bought its own view of it back from an intermediary that aggregates the feeds. That is a reasonable procurement decision and it is also a single point of dependency for a national statistic.

The ratio nobody publishes

Put the two official series side by side and a figure appears that neither release states.

At the end of March 2026 there were 2,148,000 licensed zero emission vehicles in the United Kingdom, up 29.6% on a year earlier and 5.1% of the 42.4 million vehicles on the register, of which 1,864,000 were cars. Set that against 119,080 public chargers at 1 April 2026 and the country has roughly eighteen zero emission vehicles for every public charger.

That number is close to meaningless as a measure of adequacy and extremely useful as a measure of dispersion, because most of those vehicles charge at home most of the time and a small minority never can. The public network is not serving 2.1 million vehicles, or the 1,864,000 of them that are cars. It is serving the subset of journeys, and the subset of households, for which a private socket is not an option, and no published dataset tells you how large that subset is.

The regional spread is published and it is stark. Scotland has 228.5 chargers per 100,000 people. Northern Ireland has 59.4. Those are not the same country in charging terms, and the gap has nothing to do with vehicle technology and everything to do with who has been funding installation and for how long. The policy question that follows is uncomfortable: the areas with the fewest chargers are also the areas where the commercial case for the next one is weakest, which is the ordinary shape of infrastructure and the ordinary reason it does not build itself.

Your driveway contains a metering device

The other half of the regime is domestic, and almost nobody who owns a home charger knows it is there.

The Electric Vehicles (Smart Charge Points) Regulations 2021 require every charge point sold for private use to send and receive information over a communications network, and to respond to signals by increasing or decreasing the rate of electricity flowing through it, or by changing when it flows at all. Regulation 9 requires a measuring system recording electricity imported or exported in watt-hours or kilowatt-hours and the time for which it was importing or exporting, accurate to within 10%, viewable by the owner for each individual use, by month, and across a full twelve months.

Regulation 10 sets default charging hours outside peak, and defines peak as 8am to 11am and 4pm to 10pm on weekdays. The owner may accept, remove or modify the defaults at any time.

Regulation 11 is the interesting one. Every charge point must impose “a delay of up to 600 seconds, such delay to be of a random duration which is determined, to the nearest second” at each relevant charging moment, and the owner must be able to cancel it. That randomised delay exists for one reason: if a million charge points all resume simultaneously when a tariff flips at half past midnight, the resulting step change in demand is a grid event. The regulation solves a coordination problem in physics by injecting entropy into software.

This is what an electrified transport system actually consists of. Not batteries, which are a manufacturing problem largely solved elsewhere, but a distributed telemetry and control estate with a legal duty to measure itself and a statutory instruction to disagree about when to start.

The network operators were told to presume openness

The same logic reached the wires. Regulation 10(2) of the 2023 regulations requires charge point data to be provided on request to the Secretary of State and to any distribution network operator, transmission owner or electricity system operator supplying energy to those charge points. The network needs to know where the load is, and the law makes the operators tell it.

Ofgem went further with the licensees themselves. Its Data Best Practice Guidance binds gas distribution, gas and electricity transmission, electricity distribution and the national system operator through Special Licence Condition 9.5 and Condition C3, and sets eleven principles. The eleventh is the one that matters: treat all data assets, their metadata and the software scripts used to process them as presumed open. The default is publication, with a triage process for genuine sensitivities, rather than publication as an exception granted after a business case.

Three separate instruments, written by three different bodies, all arriving at the same conclusion for the same reason. Nobody can plan a charging network, a substation upgrade or a tariff without seeing what the other two are doing, and none of them would have shared it voluntarily.

What this changes for anyone buying

If you run a fleet, a car park, a retail estate or a local authority, the practical consequence is that you no longer have to take a supplier’s word about anything.

Reliability claims can be checked against the operator’s own published compliance statement, which it is obliged to put on its website. Location and availability can be pulled from the OCPI feed directly rather than through an app. Coverage claims can be tested against the department’s regional figures. None of this requires a data science team. It requires knowing the obligations exist, and the reason to write them down here is that most procurement documents in this market still do not mention them.

The broader lesson has very little to do with vehicles. A regulator wanted a market to behave better and, instead of specifying outcomes and inspecting them, it specified a data format and required the market to broadcast its own performance. That is a distinct regulatory technique, it is spreading, and its weak point is always the same: publication without aggregation produces transparency that nobody can use. The 30-second availability feed works because somebody built a national view of it. The annual reliability reports do not, because nobody has.

The policy story about vehicle electrification, and why the country that went furthest did it with tax rather than technology, is taken up in Norway’s lead was bought, not invented. The wider question of what British open data obligations actually deliver once the press release is over sits under data and analytics.