On the spec sheet, a high-nickel NMC-811 pack is a fixed thing. On a carbon footprint, it isn’t. In our own modelling the same chemistry ranges from 70 to 138 kg CO₂e per kWh of storage depending on where the nickel, cobalt, lithium and graphite were refined, and on which grid. Our baseline pack comes in around 82. The cell assembly step everyone photographs, the gigafactory, barely moves: we model it at a near-static ~25 kg CO₂e. Almost all the variation lives upstream, in places the cell manufacturer doesn’t control and often can’t see.
For years that variation was invisible to the buyer. From 2026 and 2027 it stops being invisible. Under the EU Battery Regulation, that number has to be calculated to a prescribed method, verified by an independent notified body, and published in a Battery Passport that an OEM’s procurement team, and a customs officer, can read. The footprint moves from a sustainability talking point to a condition of selling into Europe.
This article is for the people who now have to produce that number and defend it: cell and pack manufacturers, cathode and precursor producers, and the upstream miners and refiners being pulled into the data request. It covers what the footprint actually is, where it comes from, why two identical batteries differ so much, the specific 2026–27 obligations turning this commercial, and what a number that survives verification actually requires.
What is the carbon footprint of a battery?
The carbon footprint of a battery is the total greenhouse gas emissions, in CO₂ equivalent, generated across its life cycle, from raw material extraction through refining, active-material and cell manufacturing, pack assembly, transport and end-of-life. Under the EU Battery Regulation it is expressed per unit of energy delivered: kilograms of CO₂e per kWh of total energy the battery provides over its service life.
That functional unit matters more than it looks. A footprint “per battery” is meaningless for comparison; a footprint per kWh-over-lifetime ties the embodied carbon to how much useful work the battery actually does, which is why durability and cycle life feed into the result. The regulation’s boundary is broad (cradle-to-grave or cradle-to-cradle) but it excludes the use phase, because how an EV is driven or an industrial battery is cycled introduces too much variability to compare fairly. The carbon you’re accountable for is the carbon built into the battery before it’s plugged in, plus what happens at end of life.
This is a different question from “are EVs better for the environment.” Over a vehicle’s life the answer to that is well established and not in dispute. The question the regulation forces is narrower and harder: what is the embodied footprint of this specific battery, built in this specific plant, from these specific suppliers, and can you prove it.
Where the emissions actually come from
Walk a lithium-ion supply chain stage by stage and the result is consistent across chemistries: the carbon concentrates upstream, in the materials, not in the cell line.
A battery’s life cycle runs from mining and processing, through active-material manufacturing (cathode and anode), to cell and pack assembly, distribution and end-of-life. When we disaggregate an NMC-811 model, the cathode and its raw materials dominate the climate-change result. The energy-intensive refining steps that turn ore into battery-grade inputs (nickel sulfate, cobalt sulfate, lithium hydroxide, and the graphite for the anode) carry the bulk of the embodied impact. Cell assembly and pack finishing are real but secondary; in our baseline they sit at roughly 25 kg CO₂e, a small slice of the 82 total.
This echoes what we see across the broader materials work: the biggest environmental impacts rarely sit in the mine itself. They appear in the refining and conversion steps, the high-temperature, high-energy processing that transforms a raw mineral into a precisely specified battery chemical. Lithium hydroxide produced by converting Australian spodumene in a coal-heavy Chinese facility is a very different carbon proposition from the same compound made elsewhere, even though the molecule is identical. The same is true of nickel: the processing route and its energy source can swing the embodied impact dramatically.
Cell manufacturing has its own footprint (dry rooms and electrode drying are energy-hungry, and natural gas and grid electricity both show up) but it is dwarfed by the embodied carbon arriving at the factory gate inside the cathode. Which leads to the uncomfortable conclusion for a downstream manufacturer: most of your battery’s footprint was decided by decisions you didn’t make, several tiers upstream.
Why two identical batteries can differ by ~2x
Here is the finding that reframes the whole exercise. When we model the same NMC-811 battery through different but entirely real supply chains, the footprint moves from 70 to 138 kg CO₂e/kWh without changing the chemistry, the capacity, or the spec.
What moves it is the production route and the energy behind it. Two variables dominate. First, where and how each material was refined: the geology and grade of the deposit, and the processing technology and its electricity source. Second, the grid intensity at the energy-intensive conversion and cell-manufacturing stages. A cathode precursor refined on a coal-heavy grid and a cathode precursor refined on a low-carbon grid produce materially different numbers from the same ore.
The practical implication mirrors what we find in every materials supply chain: the leverage is upstream, in refining and conversion, not in optimising the cell line or swapping a minor bill-of-materials component. A manufacturer trying to lower a pack’s footprint by tuning assembly is optimising the smallest part of the problem. The number that determines a performance class is set by supplier selection and the energy mix of the refining steps, which is exactly why the regulation demands data from across the supply chain, not just from the gigafactory.
It’s also why a generic, industry-average figure is dangerous to rely on. If your battery happens to sit on the low-impact route, an average undersells you and you lose a competitive edge you’ve actually earned. If it sits on the high-impact route, an average flatters you, until a notified body asks for the underlying data and the real number turns out to be 138, not 82.
The 2026–27 shift: the footprint becomes verified, classed, and gated
For years a battery’s carbon footprint was a voluntary sustainability claim. The EU Battery Regulation (Regulation (EU) 2023/1542) turns it into a sequence of binding obligations, and the timeline has been moving, so the direction matters more than any single date.
The mechanism is staged and deliberate. First, manufacturers must calculate and declare the carbon footprint for each battery model, per manufacturing plant. Then batteries get sorted into carbon footprint performance classes, an A-to-G-style label for embodied carbon, benchmarked against the rest of the market. Finally, the Commission sets maximum thresholds: a line above which a battery cannot be placed on the EU market at all. Declaration, then classification, then exclusion. The same logic France already applies to solar modules, scaled to the entire EU battery market.
On timing, the honest position as of mid-2026 is that the framework is real but the precise application dates are tied to secondary legislation that has slipped. The draft delegated act setting the EV calculation methodology was published in April 2024; its adoption was delayed, and declaration deadlines are pegged to the act’s entry into force plus 12–18 months rather than a fixed calendar date. The JRC published its calculation rules for industrial batteries in April 2025 and a first technical report for light-means-of-transport batteries in June 2025. Mandatory labelling with performance class lands around August 2026; maximum thresholds are expected around mid-2027; and the Digital Battery Passport becomes mandatory in February 2027 for EV, LMT and industrial batteries above 2 kWh, accessible via a QR code carrying the footprint, recycled-content and durability data.
Three details change how seriously a manufacturer should take the methodology, not just the deadline:
The number is verified, not self-asserted. Declarations must be checked by a notified body, with on-site assessment where required. A footprint assembled from spend-based estimates or convenient averages is not built to survive that. The phrase doing the rounds in the industry, that self-declarations won’t survive a third-party audit, is accurate.
Offsets can’t help you. Carbon offsets may be disclosed separately but cannot reduce the declared figure. The only way to a lower number is a genuinely lower-carbon supply chain.
The footprint is built on the PEF method, the EU’s Product Environmental Footprint framework and the battery PEFCR, underpinned by ISO 14040/14044 and ISO 14067, the same methodological backbone Minviro uses, and which we’ve been engaged with through the regulation’s development.
Read it together and the message to a battery maker is commercial before it is environmental. The regulation sets the deadline. The driver is access: your OEM customers’ procurement teams are already asking for verified footprint data because their compliance depends on it, and a number you can’t substantiate is a number that costs you the contract, or the EU market.
Where credible footprints are won or lost: electricity modelling
If you take one technical point into supplier conversations, make it this one. It’s where the rules are still contested, and where two honest modellers can land far apart.
How you account for the electricity used in refining and cell manufacturing can swing the result more than almost any other choice. The draft methodology has leaned toward using national average grid mixes or directly-connected generation, rather than fully crediting power purchase agreements (PPAs) for renewable electricity. That distinction is not academic: a manufacturer that has invested in a renewable PPA may find that investment doesn’t translate into a lower declared footprint under the prescribed method, while the choice of which grid factor applies to a Chinese refining step can move the cathode’s contribution substantially.
This is the kind of detail that separates a footprint built for a press release from one built for a notified body. It’s also why “we used an LCA tool” is not the same as “we have a defensible number.” The methodology choices, system boundary, electricity modelling, primary versus representative secondary data, functional unit, are the number.
What a number that survives verification actually requires
Most published guidance on the regulation stops at the timeline: here are the dates, get ready. Useful, but it doesn’t tell a manufacturer what “ready” means. From doing this work under ISO critical review, a few principles separate a compliant declaration from a fragile one.
Reach up the supply chain, because that’s where the carbon is. A model that starts at cell assembly misses the cathode and refining stages that actually decide the result. The regulation requires data from across the value chain precisely because the footprint is concentrated there. That means collaborative data collection with suppliers, and, where primary data isn’t yet available, representative secondary data, not a blended average. A lithium hydroxide dataset that reflects the actual conversion route and location, not an industry mean, is the difference between a number that holds and one that doesn’t.
Use the prescribed method, and declare every assumption. PEF, the battery PEFCR, EF 3.1, ISO 14040/44/67, and an explicit, transparent treatment of electricity, allocation, and the functional unit. A certifier checks the method before the result.
Carry the uncertainty, and use it. A single point estimate invites the question “compared to what?” A professional declaration shows the sensitivity range, the 70-to-138 spread that supply-chain choices create, because that range is where the decarbonisation opportunities are. Hotspot analysis tells you that shifting one refining step to a cleaner grid moves the class; a single number tells you nothing actionable.
Treat the declaration as a design tool, not a compliance chore. Once the cathode and refining hotspots are quantified, you can sequence the next reductions (recycled content, where the regulation sets recycled-content targets for cobalt, lithium and nickel from the early 2030s; cleaner conversion energy; supplier selection) against a target and a threshold you’re trying to beat. The manufacturers who treat the footprint as a roadmap rather than a label are the ones who’ll move down the performance classes before the thresholds force them to.
Modelling this in XYCLE: the 70–138 kg CO₂e/kWh spread is a supplier-and-grid problem, and that’s exactly the layer XYCLE is built around. The XYCLE Battery Materials Database covers roughly 70% of global cathode and anode production with region- and route-specific data, so a coal-grid Chinese nickel sulfate and a low-carbon alternative are distinct, EF/ILCD-reviewed datasets rather than a single industry mean. Supplier-Centric Data Management lets you bring primary LCI in via spreadsheet, supplier exchange or integration and trace it with source labelling for the notified-body audit; the EF Data Quality engine scores and aggregates DQRs across the system automatically, which is the difference between a declaration that survives verification and one that gets sent back. When you want to test whether shifting one refining step to a cleaner grid moves the performance class, Scenario Analysis answers it before you commit to a supplier.
A battery is a real climate asset over its life, and also a product whose birth carbon is concentrated in a handful of energy-intensive refining steps in a handful of carbon-heavy locations. That paradox is being turned into a market filter, complete with classes and thresholds and a passport. The producers who can quantify their position credibly, and act on the upstream hotspots that actually move the number, keep their market access as the filter tightens. The ones relying on averages and offsets find out the hard way that the regulation was never really about the cell.
Go deeper
The methodology behind compliant carbon footprint declarations, system boundaries, data types, performance classes and thresholds is set out in our guide, Carbon Footprint Compliance: EU Battery Regulations Guide. [→ Read the guide]
The full NMC-811 modelling behind the 70–138 kg CO₂e/kWh range is in our white paper on the environmental impact of battery supply chains. [→ Read the white paper]
If you’re preparing a carbon footprint declaration for the Battery Passport, or your OEM customer has asked for verified footprint data, and you want a number built to survive notified-body review, [book a meeting with our LCA team →]. We also provide [independent critical review of existing LCAs →] and can show you how supplier data and the Battery Materials Database come together in [XYCLE →].
