CBAM Trilogue Opens 20 October: 180 vs 400 vs 457 Products, Next Round Set for 30 November

The Analysis

A broker in Ningbo sent me a message at twenty past nine on 21 October with a screenshot of a client's question: does my fastener shipment to Rotterdam need a CBAM declaration in January 2028. I asked one thing back: what is the CN code and who is the importer of record. He came back with 7318.15 and a name. Then I asked what the mill's verified emissions figure was. He said the mill has never heard of CBAM. That is where this starts, and it is roughly where the European-bound steel business in this country is sitting on 25 October 2026.

What happened on 20 October is this. The European Parliament and the Council opened trilogue talks on extending the Carbon Border Adjustment Mechanism to downstream steel and aluminium goods, and closed the first round without agreement. The Commission's proposal covers 180 products. The Council wants 400. Parliament's mandate, adopted on 15 September by 464 votes to 50 with 159 abstentions, goes to 457 and pulls in some chemicals and polymers. Four items were peeled off to technical work: a de minimis threshold for aluminium, metal scrap, online platforms, and default values. The next meeting is 30 November, and the intended start date is 1 January 2028.

The headlines will be about 180 versus 400 versus 457. Let me say up front what I make of those three numbers: they are the least useful information in the file. Useful later, useless today, because none of them tells you which CN code you are. What decides your paperwork this week is the fourth item on the technical list, and I will come to it.

Let me draw the document chain, since that is the part I am paid for. On an ordinary import I work through five stations in order: classification, the declaration elements, inspection, duty, release. CBAM adds a sixth station that sits between classification and declaration, and it does not behave like the other five. Classification gives you the CN code, and the code decides whether the consignment is in scope at all. Only after that do you need the emissions figure, and the emissions figure does not come from the shipment, it comes from the plant that made the goods, verified, with an installation identifier attached to it.

Then it goes into the declaration as an element sitting next to value and origin, the declarant buys and surrenders certificates against a carbon price that moves every week, and only then does release happen. Every one of those six stations is a place where a box can sit, and five of them belong to somebody else. The one that is yours is the second one, and it is the only one that needs a year of lead time.

Whether the fifth and sixth stations run smoothly is not the same at every entry point, and this is the bit I always have to say out loud: it depends on the port. Rotterdam, Antwerp and Hamburg have been handling these declarations since the definitive regime started and their brokers have the workflow. A smaller port that sees two steel entries a month may not have a declarant who has ever filed one, and a consignment that clears in a day at Rotterdam can sit while somebody reads the guidance for the first time. If your customer routes through an unusual port, ask them now rather than in January 2028 whether their customs agent has actually filed one. The answer is often no, and knowing it before you price the business is worth more than knowing it after.

Start where I always start, with the code, then the duty treatment. That order does not change for this. The current CBAM covers the basic materials: iron and steel at the mill end, aluminium at the ingot end, plus cement, fertilisers, hydrogen and electricity. What is on the table now is the downstream layer, the things made out of those materials. If you ship fasteners under 7318, or drawn wire and wire products out of 7213 and 7227, or aluminium sections under 7604, you are the population this extension is aimed at. Not confirmed, not published, not final. But if your goods are steel or aluminium in a finished or semi-finished shape rather than in a molten or mill shape, you should be looking at your codes now rather than in December.

Take the code check concretely, because that is where most people stall. Say you ship six lines into the EU: bolts and nuts under 7318.15 and 7318.16, wire rod and alloy bar in coil under 7213 and 7227, aluminium profiles under 7604.21, and miscellaneous aluminium articles under 7616.99. Build the three columns on an explicit working assumption, because no code-level list has been published: treat the two fastener lines as sitting in all three columns, treat the two wire lines as appearing only once the count reaches 400, and treat the two aluminium lines as appearing only at 457, where Parliament's mandate also brings in polymers and chemicals.

What that table buys you is not a prediction, it is a spending order. The two codes present in all three columns are where the emissions data work starts this year, because they survive every one of the three outcomes. The two that appear only at 400 or 457 are where you wait, because two thirds of the trilogue space could still leave them outside. I have watched people run it the other way round and burn a year's budget proving a number for a code that never entered the list.

Why now, and who is pushing. The mechanism is carbon leakage, and it is not complicated. If CBAM charges the steel but not the bolt made from the steel, the import simply moves one step down the value chain and the border charge never lands. Every month the downstream extension is not in force, a wire drawer or a fastener maker outside the EU has a price advantage over the same factory inside it, and the EU mills have been saying this loudly since the original regulation was agreed. That is the upstream pressure. The other half of the politics is in the vote: Parliament's mandate passed 464 to 50 with 159 abstentions. A hundred and fifty-nine abstentions on a mandate is not a mandate with a hard edge to it. It tells you the middle of the room has not committed, which is part of why the first trilogue closed without agreement and part of why the 457 figure is a negotiating position rather than a final list.

The spread between the three numbers is worth a moment, though not for the reason people think. One hundred and eighty against four hundred and fifty-seven is a factor of two and a half in scope. That is not a rounding difference, it is the difference between a mechanism that touches a defined set of metal products and one that reaches into composite articles. Chemicals and polymers appear only in Parliament's 457, and this matters more than it looks. Half the downstream metal products I have handled in twenty years are not pure metal: a fastener arrives with a nylon washer, a cable with a polymer jacket, a light housing with a plastic diffuser. If the final list includes polymers, the question of which part of a composite article creates the obligation becomes live, and that is a classification question long before it is an emissions question.

Now the part that is not in the headline. Four items went to technical work, and the one that decides your money is default values. The declaration logic of CBAM runs like this: if you can produce verified actual emissions for the installation that made the goods, you declare the actual figure; if you cannot, you declare on the EU's default values. How those defaults are set has now been handed to technical work, which means nobody knows the number on 25 October and nobody will know it on 30 November either. But you can price the fallback today, and that is the piece I want you to take away, because it is the one piece of this story that does not depend on the list.

Let me put my own assumptions on it, and I will label them as assumptions because they are. Take a Ningbo shipper of carbon steel fasteners, CN 7318.15 bolts, 1,200 tonnes a year into the EU, FOB average 1,180 dollars a tonne. Assume an EU carbon price of 85 euros a tonne of CO2. Assume the mill runs a blast furnace and basic oxygen route and the embedded emissions of the finished bolt come to 1.85 tonnes of CO2 per tonne of product, processing included. On the actual-data path that is 1.85 times 85, or 157 euros 25 a tonne.

On the default path, assume the EU default for the same product lands at 2.40 tonnes of CO2 per tonne: 2.40 times 85, or 204 euros a tonne. The gap is 46 euros 75 a tonne, which on 1,200 tonnes is 56,100 euros a year, roughly 61,700 dollars at 1 euro to 1.10 dollars. Against an FOB value of 1,180 dollars a tonne, 46 euros 75 is about 51 dollars 40, or 4.4 per cent of the goods.

Put that next to the cost of avoiding it. Third-party verification of installation data I will put at 8,000 euros a year per plant, and that is a realistic order of magnitude rather than a quote. You spend 8,000 to avoid 56,100. The payback is inside the first year on a single 1,200 tonne account. That is the whole argument for starting the data work now, and it is an argument that survives every one of the three list outcomes, because you are not doing it for a code, you are doing it for a number.

Run the same arithmetic at the other end of the volume range, because it tells you when not to spend. The gap I am working with is 46 euros 75 a tonne and the verification I am assuming costs 8,000 euros a year, so the break-even sits at 8,000 divided by 46.75, which is about 171 tonnes a year. Below roughly 170 tonnes into the EU in a year, an 8,000 euro verification does not pay for itself on this spread, and you should be planning on defaults rather than promising a figure you cannot produce. At 400 tonnes the gap is 18,700 euros against the 8,000, and it pays inside the first year. At 1,200 tonnes it pays several times over. This is the number I would carry into the conversation with the mill, because it turns should we do this into how many tonnes do you ship, and the mill knows that answer today.

The second technical item, metal scrap, is the same argument pointed the other way. An integrated mill using iron ore sits somewhere around 1.8 to 2.0 tonnes of CO2 per tonne of steel. An electric arc furnace running on scrap can sit around 0.6 to 0.7. Whether scrap counts, and how it counts, decides whether the figure you declare is 1.85 or 0.65. Under the assumptions above, 0.65 tonnes at 85 euros is 55 euros 25 a tonne, which is 102 euros a tonne below the verified integrated route and 148 euros 75 below the default route. On 1,200 tonnes that is 178,500 euros a year. Metal scrap is the second line in a four-item list in every news story written about 20 October, and it is the only one of the four that can take seven-tenths of this cost off your book. It has no vote attached to it and no headline, which is exactly why it is being settled by people you will never meet.

The de minimis threshold for aluminium cuts the other way, and it is the one that could make the whole thing disappear for smaller shippers. If the threshold is set on the importer's annual volume or on consignment value, then 200 tonnes a year spread across three importers of record may put every one of them below it, and nobody declares. The threshold is not set. So you cannot assume you are in and you cannot assume you are out. The same goes for the fourth item, online platforms: if the obligation lands on the marketplace rather than on the importer, then parcels going out B2C through a platform may have a declarant who is not your customer at all, and if the platform ends up not carrying it, you have a consignment with nobody's name on the declaration. Who owns that is not settled either.

Let me put a second set of assumptions on the aluminium side, because a threshold is easier to reason about with a real split in front of you. Take a Foshan extruder shipping 480 tonnes a year of 7604.21 profiles to four EU customers, at 190, 140, 90 and 60 tonnes. Assume verified emissions of 1.30 tonnes of CO2 per tonne for the extrusion and a default of 1.90 tonnes, both my assumptions and neither of them published: the spread is 0.60 tonnes, which at 85 euros is 51 euros a tonne, or 24,480 euros across 480 tonnes.

Now put a threshold against it. If the line is drawn at 100 tonnes per importer per year, the 60 tonne and 90 tonne customers fall out, 150 of the 480 tonnes carries nothing, and the exposure drops to 16,830 euros. If it is drawn at 250 tonnes per importer, all four sit under it, the exposure is zero, and the 8,000 euro verification fee is money you should never have spent. The same 480 tonnes produces 24,480 euros, 16,830 euros, or nothing at all, depending on a figure nobody has published.

Which is why the answer I want before 30 November is not where the threshold lands but which entity gets measured against it: if your customer buys through three legal entities, the test may run per entity rather than against your book, and until you know that you cannot tell your own customer what his bill will be. I am not suggesting anybody restructure to slip under a line. I am saying the line is measured at a point you have not identified yet.

On timing, let me kill the number everybody gets wrong. A start date of 1 January 2028 sounds like fourteen months of slack, and there is not fourteen months of slack. Under the CBAM approach as it stands, what gets declared is verified actual emissions, and actual emissions come out of a production year at the installation. Read that as the year before the import and you have the problem I want you to see: for goods arriving in January 2028, the figure you need is likely to be the plant's 2027 data. From 25 October 2026 to 1 January 2027 there are 67 days.

If 2027 data is going to be declaraable, it has to be recorded to the EU methodology from 1 January 2027, not from whenever the trilogue finishes. Ocean transit from Ningbo to Rotterdam is about 35 days, so cargo loaded in the last third of November 2027 lands in early January 2028 and is in the first batch. Work backwards from that and you get ten months between the day the data recording has to begin and the day the first affected box sails.

One structural point about who files, because it changes who pays. The declaration obligation under CBAM sits on the EU side, on the authorised declarant or the importer, not on the Chinese exporter. You do not file. What you do is supply the number, and the number is yours. If you cannot supply it, your importer declares on default values, pays the higher amount, and then comes back to you at the next price review with a very specific request. The paperwork is not in your name and the cost still ends up in your price. That is why this is on my desk and not only on a Brussels lawyer's.

Walk one consignment through in January 2028 with no data behind it, so the failure mode is concrete rather than abstract. The box lands at Rotterdam, the declarant files the entry, and because there is no verified installation figure the consignment goes in on default values at 2.40 tonnes of CO2 per tonne. Certificates are bought at that week's price and the goods are released, so nothing stops at the port and nobody telephones you that day. What happens instead is that the cost becomes a real number on your customer's book, 204 euros a tonne against the 157 euros 25 it should have been, and it arrives at the next price review as a request for 46 euros 75 a tonne off your price. Nothing was rejected, nothing was fined, nothing made the news, and you are 56,100 euros a year worse off because a plant in Zhejiang never started a spreadsheet. That is what a documentation failure looks like under this mechanism: not a seizure, a margin.

Now let me name the conditions under which my own advice is wrong, because there are three and they are all live. The list could land at 180 products rather than 457, and if your code is not in it, two years of verification is money burned, which is why I would not commission full-range verification before the list settles. The aluminium threshold could be set high enough that small-volume exporters have no obligation at all, and the 8,000 euro verification fee never gets spent, but you only learn that when the threshold is published.

And 1 January 2028 could move, since this mechanism has already slipped more than once; if it slips, the pace can ease, but the data recording cannot stop, because you will not know which production year a new start date points at. Any one of those three turns the 56,100 euro figure into nothing. The number is not there to make you pay it; it is there to tell you whether the data work is worth starting.

The mistake I see most on the ground is treating CBAM as a form you fill in when the box arrives. It is not a form, it is the end of a chain: plant emissions data, the methodology, a verifier, an authorised declarant on the EU side. Break any link in that chain and there is no rescue at the port. Files get written beautifully and the floor is never like the file, a thing I have been saying for twenty years. This cargo can be saved, but it has to be saved before it is loaded.

Whether this reaches goods moving through third countries, or consignments under a processing trade manual, and where the exceptions in scope sit, is a question of how the rule applies and that is Ingrid Voss's ground, not mine. I will stay on the paperwork.

On 30 November I will be watching two things, not four. The product list matters, but it is what everyone else is watching too, and it is the one thing I can react to in December. What I want out of that meeting is any signal on default values and on the scrap treatment, because those two decide whether the 46 euros 75 a tonne gap is the number I plan for or a number I can halve. If the technical work on defaults produces something that looks like an average rather than a conservative percentile, my 2.40 assumption is too harsh and the urgency drops. If it produces something that punishes missing data harder, the urgency rises. Neither answer changes what I do before 1 January 2027, and that is the useful part: the action is identical under both outcomes.

The move I would warn against is the one that sounds most professional, which is appointing a consultant in October 2026 to build a full compliance programme across every product line. On a 1,200 tonne account, an 8,000 euro verification is justified by a 56,100 euro gap. A 40,000 euro programme covering codes that may never be on the list is not justified by anything. Spend on the 180-product overlap, spend on getting 2027 recorded from day one, and hold the rest of the budget until the list lands. I have watched companies buy compliance they did not need because the deadline looked close, when the deadline was fourteen months away.

So what I would actually do, and all of it before 30 November, because that meeting will clarify the list and will not generate a single gram of data. Pull every CN code you ship to the EU, mark the downstream steel and aluminium lines, and build a three-column table against 180, 400 and 457 so that on 1 December you can see in five minutes which of your codes moved. Ask your importer one written question: are you, or will you be, an authorised CBAM declarant, and will you accept our verified installation data. If the answer is no, price 2028 on default values and stop hoping. If you run an electric arc furnace, get your scrap inputs and their emissions treatment documented separately, starting this month, because that is where the 148 euro 75 a tonne lives. And hold off on full-range verification until the list lands.

↑ Back to the key points

— By Derek Xu