ArcelorMittal's Mardyck Electrical Steel Line: Inside Europe's Biggest Non-Decarbonisation Bet

On 26 March 2026, a coil of non-grain-oriented electrical steel (NGOES) rolled off the end of a new line at ArcelorMittal's Mardyck works near Dunkirk. For ArcelorMittal France, this was more than a production milestone: it closed a loop that began with a €300 million announcement in March 2022, grew to a €500 million commitment, and now gives Europe a second integrated source of the specialty steel that sits at the heart of every electric motor, generator and transformer.

The project matters well beyond Dunkirk. Non-grain-oriented electrical steel is the silicon-alloyed flat product used in the rotating laminations of EV traction motors, industrial drives, wind turbine generators and — increasingly — the backup generators that data centres depend on. Europe has limited domestic capacity for it, import dependence is high, and the product's magnetic properties (core loss, permeability, thickness tolerance) are notoriously difficult to make consistently at scale. ArcelorMittal's new Mardyck unit is the largest single European greenfield addition to that supply base in years.

What has actually been built

Mardyck is not a single line but a small complex of five interrelated lines, delivered in two phases:

• Phase 1 (three lines): a preparation line, an annealing-and-coating line, and a slitting line.
• Phase 2 (two lines): an annealing-and-pickling line and a reversing rolling mill.

The first marketable coil in March 2026 came off the Phase 1 lines. The full five-line configuration is what gives the site the ability to take hot-rolled or cold-rolled substrate and turn it into finished, coated, slit electrical steel laminations stock — a genuinely integrated flow rather than a finishing operation on externally sourced material.

The Mardyck site itself is an extension of ArcelorMittal's Grande-Synthe/Dunkerque steelworks complex (in the commune of Grande-Synthe, Nord). Historically Mardyck has specialised in cold rolling — galvanised automotive steel and thick pickled steel. The electrical steel unit sits alongside that existing business, sharing site services and logistics.

The capacity question — four different numbers

One of the most confusing aspects of Mardyck is how many different production figures circulate. They are not all wrong; they describe different things at different points in time. But the inconsistency is real and worth untangling, because it bears on how much the project actually changes European supply.

The key distinction: the 295,000 t/yr figure is the company-wide European total across two sites, not Mardyck's own capacity. Saint-Chély-d'Apcher, in Lozère, is the older electrical steel plant and produces iCARe grades, largely for automotive. Mardyck's specific contribution is closer to the 155,000–170,000 t/yr range in the most recent disclosures. The 200,000 t/yr figure that still appears on some ArcelorMittal pages appears to be either an early design number or a rounded reference.

My assessment: the most current and auditable figure is the 155,000 t/yr in the 2025 20-F, with 125,000 t/yr earmarked for automotive applications. The downward revision from 170,000 t reflects a product-mix decision rather than a physical constraint — ArcelorMittal explicitly says it reassessed the portfolio. Anyone modelling European NGOES supply should use the 20-F number and treat the 200,000 t figure as superseded.

Who is involved — and who is conspicuously absent

The most notable absence in the public record is the equipment supplier. ArcelorMittal has not named the OEMs behind its annealing-and-coating line, pickling line, reversing mill or furnaces. This is unusual for a project of this scale; in most European flat-steel investments of comparable size, the technology supplier is named in the announcement.

The closest thing to an identified engineering partner is Artelia Group, the French engineering and project-management firm. Artelia's project page states it handled preliminary and detailed design, coordination and supervision of construction, and design of fluid networks and steel structures supporting furnaces, a shot-blasting cell, the pickling unit and the rolling mill. Artelia says it has worked with ArcelorMittal France for over 30 years at the Dunkirk-area sites. This is engineering and project management, not lump-sum turnkey EPC — ArcelorMittal appears to have procured the lines itself, with Artelia as engineering manager.

Companies capable of supplying the specific technology exist — Tenova (annealing-and-coating lines for silicon steel) and Primetals Technologies (continuous annealing lines) are the obvious candidates — but no public source confirms either as a Mardyck supplier. Readers should treat any such attribution as speculation.

On the owner's side, several named individuals anchor the project:

• Gaëlle Le Papillon, head of the electrical steels production department, who took charge of the start-up in January 2024. She leads a team of more than 140 people and was named a finalist in the 2026 Trophées des Femmes de l'Industrie. Her background is process engineering at Maizières R&D and cold-rolling/galvanising process support at Mardyck — a typical ArcelorMittal internal career path for a line-startup leader.
• Hugues Baudin, Electrical Steels Program Director, named in 2022-era material.
• Bruno Ribo (CEO, ArcelorMittal France), Cristina de Lucas (CMO Electrical Steels, ArcelorMittal Europe – Flat Products) and Eric Vasseur (Technical & Service Director, leading a new multi-disciplinary customer satisfaction team for electrical steels) are the public-facing voices.

On the public-finance side, the French State contributed €25 million to the project through the France 2030 programme, against a total investment of around €500 million. The project was initially announced at just over €300 million in March 2022 and was later described consistently as €500 million — the largest ArcelorMittal investment in Europe in ten years outside decarbonisation. The ramp-up in headline value reflects both scope growth (five lines rather than an initially narrower scope) and cost inflation.

Employment numbers also shifted. Early announcements cited "more than 100 direct jobs" for the original design. A 2022 public-consultation record referenced a broader plan of 350 hires at Mardyck, of which 65 were linked to electrical steels. At peak development, ArcelorMittal Europe says "up to 400 people and 300 suppliers" were involved in construction and preliminary tests. The electrical steels department as it stands today is reported at over 140 people.

Why it matters commercially

Three forces make Mardyck strategically significant.

1. European electrical steel import dependence. Non-grain-oriented electrical steel is a specialist product where Europe has historically leaned on imports, particularly for the higher-end grades used in EV traction motors. A new integrated European source reduces that dependency and, more importantly, gives European OEMs a domestic supply chain for a product where logistics and lead times matter almost as much as price.

2. The data-centre and AI demand pull. ArcelorMittal has made the data-centre angle explicit, with the electrical steels team publishing a feature on "data center and AI boom" demand. The connection is real: data-centre backup power relies on diesel and increasingly gas generator sets, each of which needs a generator with NGOES laminations. Combined with the EV motor demand and renewable generation, this creates a demand stack that is less cyclical than automotive alone.

3. The decarbonisation adjacency. Mardyck is not itself a decarbonisation project — ArcelorMittal is careful to describe it as its largest European investment in a decade outside decarbonisation. But the two are linked strategically. Electrical steel feeds the electrification that underpins Europe's climate targets, and ArcelorMittal simultaneously confirmed a new electric arc furnace at Dunkerque. The Mardyck electrical steel unit and the Dunkirk EAF are, in effect, two halves of the same strategic repositioning of ArcelorMittal's French flat-products business: higher-value specialty products, lower-carbon primary steel.

What remains uncertain

A few things are genuinely unresolved in the public record and worth flagging:

• Equipment OEMs remain unnamed. This is a material gap. Without knowing the annealing-and-coating line supplier, it is hard to judge the line's technical specifications, its ability to produce the thinnest (0.2 mm) automotive grades at high yield, or its ramp-up curve.
• Procurement timelines are not public. ArcelorMittal has not disclosed when it placed the equipment orders or how the procurement was structured.
• The 170,000 → 155,000 t/yr downward revision deserves more explanation than the 20-F provides. The stated reason is a portfolio reassessment, but whether this reflects a demand-side decision, a technical constraint on achievable mix, or a capital-efficiency choice is not clear.
• The staged Phase 1/Phase 2 build raises the question of what is fully operational today versus still ramping. The first coil came off the Phase 1 three-line configuration in March 2026. Whether the Phase 2 annealing-and-pickling line and reversing mill are fully commissioned is not clearly stated in the most recent public communications.

What to watch

Three indicators will show whether Mardyck delivers on its strategic promise:

1. Automotive qualification wins. With 125,000 t/yr earmarked for automotive, named OEM qualifications — particularly for EV traction motors — would confirm the line is hitting the required magnetic and dimensional specs.
2. Ramp-up speed. Whether Mardyck reaches its rated capacity in the 12–24 months following first coil, or shows the prolonged teething typical of electrical steel lines, will indicate how well the technology integration has gone.
3. Capacity restatement. If ArcelorMittal revises capacity again in its 2026 annual report, the direction of travel — up or down — will say a lot about both demand and operational performance.

For European steel professionals, Mardyck is a case study in how the flat-products industry is being restructured: away from commodity carbon steel, toward the specialty electrical steels that the energy transition and digital economy require, and built with a mix of private capital, state support and long-cycle engineering discipline. The first coil is a beginning, not an arrival.