SITELIVE 
What happened in Québec
On 13 November 2025 ELYSIS — the joint venture between Alcoa and Rio Tinto, supported by the governments of Canada and Québec — started up a 450 kiloamp inert anode cell at the end of an existing potline at Rio Tinto's Alma smelter. It is the first implementation of inert anode technology at commercial cell size, and the cell was built for industrial demonstration: testing continues on it to gather data for future commercial deployment. Separately, Rio Tinto holds the first ELYSIS smelter technology licence and is building a ten-pot, 100 kA demonstration plant at its Arvida site — a US$285 million (CAN$375 million) joint venture with Investissement Québec, sized at up to 2,500 tonnes a year of commercial-quality aluminium, with first production targeted for 2027.
The chemistry, and why it is only part of the problem
Conventional Hall-Héroult smelting consumes its carbon anodes: the anode reacts with oxygen stripped from alumina and leaves as CO₂, so the emission is inherent to the process rather than to the fuel. An inert anode does not react — it releases oxygen instead — eliminating direct greenhouse gases from electrolysis. But those direct emissions are the minority of the total. Australia's Department of Industry, Science and Resources puts scope 2, the purchased electricity, at around 85% of the emissions from aluminium smelting. Inert anodes address the remainder of a problem whose bulk sits in the grid.
The number that actually moves
The International Aluminium Institute puts the global average intensity of primary aluminium at 15.1 tonnes CO₂e per tonne in 2022, down from 16.1 tonnes in 2018 — cradle-to-gate, covering mining, refining, anode production, electrolysis and casting. Hydro guarantees its low-carbon primary range at a maximum of 4 kg CO₂e per kg of aluminium, about a quarter of the global average, achieved with hydropower and high-efficiency electrolysis. Its CIRCAL product, made with at least 75% post-consumer scrap, averages 1.9 kg CO₂e per kg, is traceable per batch from the scrap and is verified by DNV. Same metal, same alloys, roughly an order of magnitude between the extremes — the difference is where the electrons came from and how much of the metal was already above ground.
Recycling remains the biggest lever available today
The IAI's figures are blunt: 186 gigajoules of primary energy per tonne for primary aluminium from mine to cast house, against 8.3 GJ per tonne for recycled metal, a saving of about 95%. Construction is well placed to exploit that, because extruded products — window frames, façade sections, louvres, balustrades — are long-lived, alloy-consistent and recoverable in identifiable streams at demolition. The constraint is not technology but collection, sorting and alloy contamination. High post-consumer content is a supply-chain achievement rather than a metallurgical one, which is why the products that carry it are branded, third-party verified and priced accordingly.
Australia's very expensive year
Australia has four primary smelters, and two have now been underwritten with public money. In March 2026 the Queensland and Commonwealth governments each committed A$1 billion over ten years to 2040 to keep Rio Tinto's Boyne smelter at Gladstone operating beyond the end of its current power contract in 2029, alongside Rio Tinto underwriting around A$7.5 billion of new generation and transmission; it has contracted more than 2.8 GW of renewables and over 600 MW of storage in Queensland since January 2024. In December 2025 the Commonwealth announced a plan with the NSW government and Tomago — Australia's largest single electricity user — to negotiate a long-term, fixed-price power purchase agreement before its energy contract expires in 2028, with Tomago contributing at least A$1 billion in capital and major maintenance over the next decade. Behind both sits the A$2 billion Green Aluminium Production Credit announced in January 2025: payable per tonne from 2028–29 for up to ten years, with rates set facility by facility and tied to cutting scope 2 emissions.
The honest limits
Inert anodes are not arriving at scale this decade. The Arvida demonstration plant's 2,500 tonnes a year is a rounding error beside a smelter producing hundreds of thousands of tonnes, and the 450 kA cell at Alma is a test article at the end of a conventional potline, not a production line. The Australian money is buying continuity of supply and time to build renewable generation, not clean metal in the short term, and the December 2025 Tomago announcement was explicitly a plan to work through with the state and the company, not a signed contract. Low-carbon primary metal also remains a constrained product that carries a premium, largely committed to automotive and consumer brands with public commitments — which is why façade contractors rarely see it in a standard quotation.
What specifiers should ask for
If carbon is scored in your tender, corporate averages are worthless. Ask for a product-specific EPD covering scopes 1 to 3 for the metal you are actually buying, the post-consumer scrap percentage, and the name of the third party that verified it. Ask which smelter and which grid: hydropower in Norway, Québec or Tasmania is a different product from coal-firmed metal, even under the same supplier's letterhead. Then price the trade honestly — verified low-carbon aluminium usually means a premium, a longer lead time and a narrower range of alloys and finishes. A project that discovers this at shop-drawing stage pays for it twice.
What to watch next
Three markers will tell you whether this is moving faster than the press releases suggest. First production at Arvida in 2027 would show the licensed technology works outside ELYSIS's own research centre. The final design of the Green Aluminium Production Credit — the credit rates and what counts as green — will determine whether Australian metal can be sold as a low-carbon product rather than merely a subsidised one. And in the meantime, watch scrap: recycled content is the only lever that delivers a step change in embodied carbon on a project timeline rather than a smelter timeline.