Field Notes_008_Landed cost 2026: CBAM, tariffs and the end of the cheap import

"The cheap import did not die of one cause. It was repriced, line by line, while the invoice still looked familiar."

Core point — Since 1 January 2026, the EU's CBAM (Carbon Border Adjustment Mechanism) is in its definitive phase: importers of steel, aluminium, cement, fertilisers, electricity and hydrogen file verified annual declarations and surrender certificates priced off the EU ETS. The 2026 cash cost is small by design; the trajectory is not. Combined with the tariff stack and structurally repriced freight, the number that decides which products and customers you can afford is no longer the purchase price. It is landed cost, computed honestly, SKU by SKU. Landed cost is a design variable, not an accounting output. Boards that treat it as bookkeeping will discover their margin was repriced by third parties, without warning, in three separate currencies: carbon, duty and time.

Field Notes — A permanent reference for the CEO whose purchase price still looks fine, and whose margin quietly does not.

For thirty years, the landed cost model of a European distributor was a short piece of arithmetic. Purchase price, plus freight, plus a duty line that rarely moved, divided by units. The arithmetic was so stable that most companies stopped doing it. The purchase price became the decision; everything after it became overhead allocation.

That world is gone. It did not end with a bang but with three quiet line items: a carbon price at the border, a tariff environment that changes faster than contracts do, and freight that stopped behaving like a constant. Field Notes #003 argued that the tariff shock belonged in every board's stress test. Eighteen months later, the shock is not a scenario anymore. It files paperwork.

Why "the customs broker handles it" is not a strategy

Walk into a Belgian mid-market importer today and ask who owns CBAM. The answer is almost always a compliance function: the customs broker, the freight forwarder, an external consultant with a reporting tool. The declarations get filed. The company is compliant. And the board concludes the topic is handled.

It is handled the way a fire alarm handles a fire. Compliance answers the question "did we report correctly?" It never answers the questions that decide money: which SKUs stop making sense at the 2028 carbon factor, which supplier's embedded emissions are double their competitor's, which customer contracts silently absorb the cost because nobody wrote a pass-through clause. A carbon cost that is reported but not priced into decisions is not managed. It is merely documented.

This is the same failure mode as the resilience decks of 2024: a compliance artefact where a decision instrument should be. The work begins where the filing ends.

What actually changed on 1 January 2026?

Four things, each verifiable, none of them temporary.

First, CBAM became real law with real invoices attached. The transitional phase (2023 to 2025) was reporting only. Since 1 January 2026, the definitive regime applies to imports of iron and steel, aluminium, cement, fertilisers, electricity and hydrogen. Importers above the 50-tonne annual threshold need authorised CBAM declarant status, verified emissions data, and certificates priced off EU ETS auction averages. The first annual declaration and certificate surrender falls due on 30 September 2027. Miss the surrender and the penalty runs at €100 per tonne of CO2.

Second, the cost curve is a ramp, deliberately. In 2026 the certificate obligation covers only a small slice of embedded emissions, aligned with the phase-out of free allowances under the EU ETS, which runs from 2026 to 2034. Roughly half of embedded emissions become payable around 2030 and the full amount by 2034. This is the detail most boards get wrong in both directions. The optimists see the small 2026 bill and file the topic away. The pessimists multiply full ETS prices by full tonnage and panic. Both miss the operational point: the bill grows every single year on a published schedule, which means the redesign work has a deadline you already know.

Third, the scope is widening, not settling. In June 2026 the Council agreed to extend CBAM to downstream goods and to reinforce anti-circumvention safeguards. Chemicals and polymers are under study for 2027 and 2028. If your product catalogue touches processed steel or aluminium, the question is not whether more of it enters scope, but when.

Fourth, the tariff environment stopped being an administrative variable. Between US baseline tariffs, EU trade defence measures and the anti-circumvention rules that follow them, the cost of crossing a border is now a strategic number that can move faster than your customer contracts allow you to reprice. Field Notes #003 called this the tariff shock. It is now simply the operating environment.

What does CBAM actually cost an importer? A worked example

Numbers, because a risk without a number is theater. Take a representative, deliberately simplified case: a Belgian distributor importing 10,000 tonnes of steel products per year from outside the EU. Embedded emissions for ordinary carbon steel run in the neighbourhood of two tonnes of CO2 per tonne of product; assume 1.9. That is 19,000 tonnes of embedded CO2 per year. At an assumed ETS price of €80 per tonne, the fully phased-in exposure is roughly €1.5 million per year.

In 2026, with only the first sliver of the phase-in payable, the cash cost is a rounding error: a few tens of thousands of euros. By around 2030, at half the phase-in, the same import flow carries a carbon line around €700,000. By 2034 the full €1.5 million. Now put that against the P&L of the company doing the importing: a distributor running a 4 percent EBIT margin on that flow. The carbon line alone, unmanaged and unpriced, consumes a third or more of the margin on those SKUs by the end of the decade.

Every number above is an assumption you should replace with your own. That is precisely the point. The company that has run this arithmetic on its own top twenty import SKUs owns a decision instrument. The company that has not is carrying a repricing event on its balance sheet and calling it overhead.

Landed cost is a design variable

Strip the machinery away and the whole discussion reduces to one operational principle: landed cost is a design variable of your sourcing network, not an accounting output of it.

True landed cost in 2026 has at least seven lines: purchase price, freight (at repriced Red Sea-era rates, not 2019 rates), conventional duty, trade-defence tariffs, the carbon line on its published ramp, the compliance cost of verification and declarant status, and the working capital cost of the route, because a longer, cheaper route is an inventory decision wearing a freight disguise. Each line is set by a different actor on a different clock. Only the first one is on your purchase order.

The implication is uncomfortable, and it is the same one Field Notes #003 drew for resilience: most import portfolios were optimised for a cost structure that no longer exists, and then patched. A patched optimisation is not a redesign. It is a postponed one. The difference here is that CBAM publishes the schedule of your future costs in advance. Very few forces in business do you that favour.

The five moves, in order

What follows is the operational sequence a leadership team can run in one quarter. Numbered, because sequence matters.

1. Rebuild landed cost at SKU level for the top twenty import lines by margin contribution. All seven lines, current rates, no averages across the catalogue. Averages are where margin leaks hide. One page per SKU, same discipline as the flow maps in Field Notes #003.

2. Add the carbon line at three points on the ramp: 2026, 2030, 2034. Use verified supplier emissions where you have them and default values where you do not, and flag every SKU where the default is doing the work. A supplier who cannot document embedded emissions below the default value is charging you the difference.

3. Stress the tariff line the way #003 stressed the routes. One question per SKU: if a 25 percent measure lands on this category overnight, what is pre-decided? If the answer is "we would convene a task force", the test has already failed.

4. Re-run the nearshoring arithmetic with the new lines in. An EU or EFTA supplier quote that lost on purchase price in 2023 may win on landed cost by 2028: no CBAM line, no trade-defence exposure, shorter working capital cycle, ETS costs already inside the price. The spread narrows every year of the phase-in. Somewhere on that ramp sits your switching point; the work is knowing the year, not the sentiment.

5. Reprice the customer side before the ramp does it for you. Every import-exposed contract needs three clauses: a carbon pass-through indexed to a published reference, a tariff reopener above a defined threshold, and indexation language a buyer can verify. Selling on fixed prices while your input costs sit on a legislated escalator is not commercial courage. It is a short position on EU climate policy, held on behalf of your shareholders, without a mandate.

What to do Monday morning

Not a transformation program. Three actions, one owner each, four weeks.

Ask finance for the true landed cost of your top five import SKUs, all seven lines. Expect the first version to be wrong; the argument about why is where the learning is. Ask procurement which of your suppliers can already deliver verified embedded-emissions data, because that list is your future sourcing shortlist, whoever is on it. Ask sales which customer contracts contain a carbon or tariff pass-through clause. The honest answer is usually "almost none", and that single fact, stated aloud in a board meeting, does more for repricing discipline than any consultant's deck.

The five numbers for the board dashboard

Monthly, one page, reviewed at every meeting:

1. (percentage of COGS in CBAM scope, current and under the agreed downstream extension.

2. Carbon cost per top-twenty SKU at 2026, 2030 and 2034 phase-in rates.

3. Tariff exposure as a percentage of COGS, by category.

4. Landed-cost drift year on year versus realised price increases, the single cleanest measure of whether you are absorbing or passing through.

5. Percentage of import-exposed revenue covered by pass-through clauses.

A board that cannot answer these five questions about the company it governs is not governing the import margin. It is hoping. And hope, as Field Notes #003 put it, is not a design property.

The distributors that treat landed cost as a design variable will spend the phase-in years quietly buying market share from those that treat it as bookkeeping. The schedule is published. The arithmetic is one quarter of work. What remains is the decision to do it before the ramp does.