CBAM by the Numbers

The headline is concentration with improving reporting, not proof of readiness. Physical mass is concentrated in iron and steel, but the Commission's estimated-emissions picture gives aluminium a much larger share than its mass share. Actual-value use rose sharply, particularly for large entries, while electricity remained overwhelmingly default-based. These are important planning signals, but none establishes verified company emissions or definitive-regime performance.

The core evidence is the Commission's cleaned extract from the CBAM Transitional Registry. It covers quarterly reports from Q4 2023 through Q2 2025 and carries a data cut-off of 31 August 2025. The Commission describes an “entry” as an import for the purpose of its charts. The world overview reports approximately 124,000 reports, 1.1 million entries, 32,000 distinct declarants and 41,000 distinct importers for non-electricity goods.C-001 C-002

Those populations are related but not interchangeable. One importer can generate many entries and reports; the declarant and importer may be different actors; rounded chart counts cannot be reconstructed into a precise conversion funnel. The dataset is evidence about reporting activity, not a census of every customs movement and not a measurement of how many firms will ultimately cross the definitive-period threshold.

The reporting workload was also concentrated across national competent authorities: Germany received about 18,000 reports, Italy 16,000, Poland 15,000, Spain 8,000 and France 7,000. Those are rounded report-routing counts—not Member-State import mass, error rates or readiness scores.C-026

This article therefore does not recycle every number from The State of CBAM 2026. The annual report owns the dated strategic synthesis. This report owns the statistical question: what was measured, with which denominator, and what inference survives the source limitations?

The 156 million tonnes divide into four displayed mass sectors: 108 million tonnes of iron and steel, 23 million tonnes of fertilisers, 18 million tonnes of cement and 7 million tonnes of aluminium. Hydrogen rounds to 0% in the world mass chart and does not appear as a separately legible mass total. Electricity is a separate 96 million MWh dataset.C-003 C-012

The comparison changes the interpretation. Iron and steel dominate both measures, but their 69% mass share becomes 60% of the default-based emissions estimate. Aluminium moves in the other direction: only 5% of mass but 24% of estimated emissions. Cement is 11% of mass and 5% of the estimate. These are not measured carbon intensities; they are the arithmetic consequence of reported quantities and the transitional default values used by the Commission.C-004

A second Commission total is useful only if its shorter period stays attached. For calendar-year 2024, the application report estimates 167 million tCO₂e: 102.5 million for iron and steel, 38.4 million for aluminium, 18 million for fertilisers and 8 million for cement. The displayed components add to 166.9 million because they are rounded. This annual estimate is not an alternative precision for the seven-quarter 343 million figure; it is a different time window.C-023

The full sector ranking changes again when defaults are applied

The Commission prints a separate estimated-emissions total and ten leading origins for each mass sector. Aluminium's 81M tCO₂e is nearly twice fertilisers and five times cement despite being the smallest sector by reported tonnes. Keep this table in the violet evidence class: every value is a rounded, default-based estimate—not a verified emissions inventory.C-029

Sector ownership remains with the dedicated CBAM sector pages. Use the iron and steel, fertilisers, cement and aluminium pages for goods-specific scope. The numbers here compare scale; they do not replace CN-code classification.

Limiting the story to Ukraine and Türkiye would hide most of the dataset. The Commission's mass chart identifies 20 origins ranging from 21.18 million tonnes for Ukraine to 1.64 million tonnes for South Africa. The list includes major steel, fertiliser, cement and aluminium origins across Europe's neighbourhood, Asia, the Americas and Africa.C-005

The order changes when the Commission applies transitional default values. China leads the estimated-emissions chart at 43 million tCO₂e, followed by Türkiye at 42 million and Russia at 39 million. India and Ukraine are each displayed at 19 million. Mozambique and the United Arab Emirates enter the emissions top 20 because aluminium's default-based emissions weight is larger than its physical mass share.C-006

Annex IV supplies detailed dashboards for the five largest origins by total tonnes: Ukraine, Türkiye, Russia, Canada and China. A sixth card reconstructs Egypt only from values explicitly printed across the world and four sector charts. Egypt is the stronger sixth example than India because it appears in every displayed mass sector and ranks sixth by both reported mass and default-based estimated emissions. Missing Egypt actor counts and actual-value percentages remain visibly unpublished rather than inferred.C-015 C-028

Ukraine's reported mass is overwhelmingly iron and steel. Türkiye is almost evenly split between iron and steel and cement by mass. Russia combines iron and steel with fertilisers. Canada is almost entirely iron and steel in this chart. China is dominated by iron and steel but has visible aluminium and fertiliser shares. Those differences explain why one reporting or supplier-engagement playbook cannot be assumed to fit every origin.

Overall actual-value use rose from 8% of entries in Q4 2023 to 53% in Q2 2025. The break is sharp: 11% in Q2 2024 becomes 46% in Q3 2024, the quarter when the transitional rules tightened the route to defaults. For entries above 1,000 tonnes, the same progression reaches 90% in Q3 2024 and 93% by Q2 2025.C-007 C-008

The sector paths are not identical. Cement already showed 44% actual-value use in Q4 2023 and reached 84% by Q2 2025. Fertilisers rose from 21% to 77%. Iron and steel moved from 8% to 53%, while aluminium moved from 10% to 54%. Large entries in every sector reached at least 93% by the final quarter shown.C-009

That is evidence of a reporting-method shift, not evidence that 53%, 77%, 84% or 54% of emissions were verified. The Commission labels the entries non-verified. Teams deciding between methods should use the dedicated default values versus actual emissions guide and the verification workflow guide rather than treating the historical chart as a definitive-regime rule.

The adopted mass threshold is an annual cumulative 50 tonnes across the four mass-based sectors; electricity and hydrogen are excluded from that exemption. The Commission applied a retrospective >50-tonne simulation to its transitional data. In the rounded dashboard counts, reports fall from 124,000 to 54,000, entries from 1.1 million to 0.7 million, declarants from 32,000 to 12,000 and importers from 41,000 to 12,000. Yet the displayed reported mass remains 156 million tonnes.C-010

The default-based estimated-emissions total changes only from a displayed 343 million to 341 million tCO₂e.C-011 The simulation therefore illustrates concentration: many smaller actors and entries are removed while nearly all displayed mass and estimated emissions remain. It does not tell us how many 2026 importers actually remained below the legal threshold, because the transitional report population and definitive annual aggregation are not the same measurement.

Those two rounded Commission policy estimates are the cleanest summary of the simplification's intended concentration effect. They remain estimates, not a measured 2026 census.C-030

The retained population also has its own simulated actual-value series: 8%, 9%, 11%, 53%, 54%, 61% and 60% across the seven quarters. It should be compared with the all-entry series as a retrospective composition effect, not reported as observed definitive-regime performance.C-027

To test a current annual import profile, use the deterministic 50-tonne threshold checker and read the threshold guide. Do not apply the world simulation as a firm-specific exemption conclusion.

Electricity cannot be folded into the 156 million tonnes. Annex IV reports 96M MWh across the same seven-quarter span, 658 reports, 1,761 entries, 160 distinct importers and 131 distinct declarants. The top five displayed origins are United Kingdom 63M MWh; Serbia 15M MWh; North Macedonia 5M MWh; Bosnia and Herzegovina 4M MWh; Montenegro 3M MWh.C-012

Rounding note: The seven rounded quarterly labels add to 97M MWh, while the Commission displays a 96M MWh total. The apparent 1M difference is a source-rounding effect, not an alternative total.

The Commission also displays 86M tCO₂ and immediately warns that the figure overestimates real embedded emissions because the calculation uses CO₂ factors reflecting electricity generated from fossil sources.C-013 That warning is not a footnote to discard—it changes the permissible claim. The number can describe the Commission's displayed estimate, but it should not be presented as actual electricity emissions.

Default values represented roughly 96%–98% of electricity entries in every quarter shown.C-014 That makes electricity the clearest example of why an entry-method percentage is not a verified-emissions percentage. The dedicated electricity default-values guide owns the operational rules.

The Transitional Registry charts end in Q2 2025. A separate Commission launch publication reported that, by 7 January 2026, more than 12,000 economic operators had submitted authorisation applications, more than 4,100 had obtained authorised-declarant status before and immediately after launch, and 10,483 customs declarations containing CBAM goods had been validated automatically through integrated customs systems.C-019

The earlier implementation record adds another bounded operational count: 573 Operators Portal access requests processed between December 2024 and August 2025. The Commission describes guidance improvements, but does not publish an exact rejection count or rate. The number therefore belongs in an operations chronology, not in a market-readiness percentage.C-024

The report also places the changing population in scale: about 200,000 importers in 2024, an estimated 18,000 expected to remain in scope in 2026, 5,902 authorisation applicants at the report snapshot and 78 registered third-country operators. These are observed, estimated and operational populations with different dates. They are not a funnel and should not be divided into a readiness percentage.C-025

These are deployment signals, not a conversion funnel. The authorisation counts do not share one exact timestamp, and customs validation does not prove embedded-emissions completeness, verification or later certificate settlement. The right operational interpretation is narrower: the definitive regime had live applicants, authorised actors and real-time customs processing in its first week.

The Commission later published Q1 and Q2 2026 certificate prices of €75.36 and €75.28 per certificate/tCO₂e.C-020 Those prices are observed official values, but they do not convert the historical 343 million tCO₂e estimate into a bill. The estimate covers an earlier transitional period, uses defaults and does not apply firm-specific free-allocation adjustments, carbon prices paid abroad or definitive methods. Use the certificate-price tracker and cost calculator only with their stated assumptions.

The application report adds a separate 2024 trade-value and GDP lens. Ukraine had the largest printed CBAM export value among the selected neighbourhood countries at €3.3bn, equal to 1.8% of GDP. Mozambique's €1.2bn was smaller in absolute terms but equal to 5.5% of GDP. Egypt and Morocco sit between those examples. This is precisely why tonnes, euros and GDP shares need separate tables.C-031

The official 2035 model is a scenario, not observed evidence

The Commission's JRC-GEM-E3 work reports small GDP effects against a no-CBAM baseline and larger-looking emissions changes when expressed in MtCO₂e. These results are useful for scale, but cannot be described as measured outcomes or guaranteed forecasts.C-032

The screening method behind the roughly 120 substances used a 0.845 carbon-leakage indicator threshold, at least 25ktCO₂e of EU ETS emissions and more than 1Mt annual EU production. These criteria explain an assessment process; they do not mean all shortlisted substances entered adopted CBAM scope.C-037

These boundary figures answer different questions. The free-allocation estimate is assumption-dependent; indirect and transport percentages are broad assessment context; the cement share is an average; and the screening thresholds are not a declaration calculation. They belong here only with those qualifications attached.

The Commission identified recurring transitional data problems: decimal separators, confusion between kilograms and tonnes, incorrect CN codes and inconsistent country codes.C-018 These are not abstract statistical caveats. Each can break a practical data chain: a unit error can change quantity by three orders of magnitude; a CN-code error can change sector treatment; a country-code error can misroute a default; and an identity mismatch can prevent installation evidence from joining to an import line.

The Commission removed unrealistic high-end outliers and warns that minor fluctuations remain possible; potential underreporting was not corrected.C-017 Therefore the charts are stronger for order-of-magnitude, composition and reporting-pattern questions than for exact reconciliation to customs totals. They do not establish that “most firms fail” or provide a current error rate.C-022

1. Prioritise without prejudging. Use sector and origin concentration to decide which supplier relationships deserve early evidence work, not to assign company readiness.
2. Keep dimensions separate. Maintain tonnes, MWh, tCO₂e, entry counts, firm counts and euro values in separate fields with explicit periods.
3. Challenge defaults with evidence, not optimism. The historical rise in reported actual values shows a method shift. It does not waive the evidence and verification requirements attached to actual values.
4. Preserve the join keys. CN code, country, installation, supplier, reporting period, quantity unit and method state should survive from customs line to supplier request and review file.
5. Use the legal threshold on your own annual profile. The Commission simulation is context; your cumulative import record is the decision input.
6. Record what is unknown. A missing installation, precursor, verifier or carbon-price document is a gap—not a value to infer from a country average.

The practical next step is to build a source-linked preparation record. Start with the goods checker, test the annual threshold, request supplier evidence with the supplier template, and organise open gaps in the reporting evidence checklist.

This article uses COM(2025) 783 and its Annex IV as the primary statistical spine. The source PDFs were downloaded from the EU Publications Office, checksum-frozen and extracted locally. Chart values that were not available in the PDF text layer were transcribed from rendered source pages at displayed precision. No interpolated chart segment is published as an observed value; where a small segment lacked a printed label, the country card says “Not printed.”

The mass and estimated-emissions rankings reproduce the Commission's displayed top-20 values. No hidden precision is invented. The sector comparison uses the Commission's displayed shares and preserves the 101% rounded emissions sum rather than normalising it to 100%. The 50-tonne figures remain labelled a simulation. Electricity remains in MWh and its displayed emissions warning travels with the number.

Internal calculations are limited to display widths. They do not create new quantitative claims. The article does not divide 343 million tCO₂e by 156 million tonnes because the sector mix, default factors and periods do not support a general CBAM intensity. It does not multiply the estimate by current certificate prices because the populations, legal methods and timing do not match.

Only official EU material is used for the quantitative findings. Search Console data informed vocabulary and confirmed live demand around sectors, defaults, reporting, thresholds and certificate prices, but it did not override canonical ownership or source quality.

Update triggers: a new Commission Registry statistical release; revised Annex IV figures; a methodology correction; new observed definitive-regime statistics; or new quarterly prices. Research cut-off: 10 August 2026.