Amazon’s Ohio data centre to use Prysmian cables made with Rio Tinto ELYSIS aluminium

Prysmian S.p.A. (Euronext Milan: PRY), Rio Tinto Limited (ASX: RIO) and Amazon.com, Inc. (Nasdaq: AMZN) are putting low-carbon aluminium into an Ohio data centre, moving the emissions contest around artificial intelligence infrastructure beyond electricity consumption and into the physical materials that carry power through the site. Prysmian said it had partnered with Rio Tinto to supply electrical cables manufactured with aluminium produced through ELYSIS inert-anode technology, which removes direct greenhouse gas emissions from the smelting stage and releases oxygen instead. The companies did not disclose the quantity of cable, the value of the order, the precise Amazon facility or the delivery timetable, leaving the immediate financial contribution unclear. The significance is therefore less about one contract and more about whether a technically different form of primary aluminium can become a repeatable specification in the rapidly expanding data-centre market, according to the initial report.

The language around the project requires care. ELYSIS is described as eliminating direct greenhouse gas emissions from aluminium smelting, not as making the completed cable, data centre or aluminium supply chain entirely emissions-free. Mining bauxite, refining alumina, generating electricity, transporting metal and manufacturing cable all remain part of the lifecycle footprint. That distinction matters because corporate buyers are under pressure to make defensible environmental claims, while investors increasingly separate genuine process innovation from broad labels such as green or zero-carbon. The project nevertheless addresses one of the hardest industrial emissions sources in the aluminium chain, the use of consumable carbon anodes in the conventional Hall-Héroult smelting process.

What exactly are Prysmian and Rio Tinto supplying to Amazon’s Ohio data centre?

The commercial object is an electrical cable made by Prysmian using Rio Tinto aluminium produced with ELYSIS technology. In a data centre, such cable forms part of the less visible but essential power-distribution architecture linking incoming electricity, transformers, switchgear, backup systems and computing halls. The announcement does not specify the cable design, conductor volume or whether the material will be used across an entire facility or in a defined application. Those omissions prevent a calculation of revenue or avoided emissions, but they do not diminish the value of the installation as a live industrial reference for all three companies.

For Prysmian, the order fits a strategy built around electrification, grid reinforcement and digital infrastructure. The Italian group has been expanding aggressively in North America, where it has bought Encore Wire, agreed to acquire Atkore and signed large supply arrangements tied to data-centre connectivity. Its proposed $3.8 billion Atkore acquisition would deepen its position in electrical products used by utilities, construction and data centres, while North America already represents roughly 40% of group revenue, according to a Reuters account of the transaction. A cable order that embeds a differentiated aluminium input gives Prysmian a product-level response to customers that increasingly evaluate both capacity and carbon intensity.

For Rio Tinto, the arrangement converts a long-running technology programme into a customer-facing application. The miner is not simply selling a commodity grade and leaving the downstream story to manufacturers. It is helping place process-specific aluminium in equipment that a high-profile buyer can identify inside a real asset. That creates a potential bridge between Rio Tinto’s investment in lower-carbon smelting and the willingness of end users to specify, track and possibly pay more for metal with a different emissions profile. The commercial test is whether that willingness survives beyond demonstration-scale contracts and becomes part of normal procurement.

Why does ELYSIS aluminium matter for data-centre construction emissions?

Data-centre sustainability has traditionally been discussed through operational electricity, water use and renewable-power contracts. The scale of the current artificial intelligence build-out is making embodied emissions more important because developers are constructing campuses faster, installing denser power systems and consuming large volumes of steel, concrete, copper and aluminium before a server is switched on. Purchased materials sit in a customer’s value chain and can be material to its Scope 3 inventory. The Greenhouse Gas Protocol’s value-chain framework is designed to capture emissions outside a company’s own facilities, including purchased goods, which is why procurement choices can matter even when a data centre later operates on lower-carbon electricity.

Aluminium presents a particularly large opportunity and a difficult accounting challenge. It is light, conductive, durable and widely recyclable, but primary production is electricity-intensive and conventional smelting also creates process emissions when carbon anodes are consumed. A 2025 industry analysis estimated that aluminium production accounted for about 1.12 billion tonnes of carbon-dioxide equivalent in 2023, or roughly 2% of global greenhouse gas emissions. The same analysis noted that recycled aluminium uses only a fraction of the energy required for primary production, while inert-anode systems target the direct chemical emissions that remain at the smelter. ELYSIS therefore tackles a different part of the problem from renewable power or recycled content, and the strongest procurement approach may eventually combine all three.

The Ohio project also offers Amazon a way to show that data-centre decarbonisation is becoming an engineering specification rather than only a portfolio-level energy claim. That does not prove a meaningful reduction by itself because no baseline, tonnage or lifecycle assessment has been published. It does show that Amazon is willing to let a supplier and materials producer identify the smelting route behind a component. If the company begins requiring similar traceability across larger categories of cable, busbar or electrical equipment, suppliers would receive a clearer demand signal for investment in lower-emission capacity.

How does the project extend Rio Tinto and Amazon’s critical-materials strategy?

The aluminium cable arrangement follows a separate two-year agreement under which Rio Tinto is supplying copper produced through its Nuton leaching programme for components used in Amazon Web Services data centres. Under that earlier deal, manufacturers in Amazon’s supply chain were to use copper from Gunnison Copper’s Arizona mine, although volumes and financial terms were not disclosed. Rio Tinto’s Nuton process uses naturally occurring bacteria in a leaching system intended to recover copper from ore that might otherwise be difficult to process. The January agreement established that Amazon was prepared to look upstream into how essential conductive metals are produced, not merely who fabricates the final component.

Taken together, the copper and aluminium arrangements point to a broader sourcing model. Amazon gets early access to traceable material pathways and a narrative around supply security and emissions. Rio Tinto gets a direct relationship with one of the largest builders of digital infrastructure and a route to validate technologies with a demanding end user. Manufacturers such as Prysmian translate the raw material into qualified equipment that can actually be installed. The arrangement is strategically more interesting than a conventional sale because the demand signal runs from the data-centre owner through the cable maker to the smelter and mine.

That model may become more valuable as artificial intelligence increases demand for grid connections and electrical equipment. Copper remains the superior conductor in many uses, but aluminium can provide cost, weight and supply advantages in suitable applications. A customer that understands where each material works best can pursue both resilience and lower embodied emissions rather than treating the choice as a simple substitution contest. Rio Tinto’s presence in both metals gives it a natural interest in helping Amazon optimise the mix, while Prysmian brings the system-design and manufacturing expertise required to turn that mix into certified cable products.

What does the order mean commercially for Prysmian, Rio Tinto and Amazon?

The immediate earnings effect is likely to be modest because no contract value or volume was disclosed. Prysmian generated €17.03 billion of sales and €1.93 billion of adjusted earnings before interest, tax, depreciation and amortisation in 2024, while free cash flow exceeded €1 billion. Against that scale, a single facility order would need to be unusually large to move group results. The more relevant measure is whether the project helps Prysmian win repeat data-centre specifications, especially in North America, where the group is investing through acquisitions, manufacturing expansion and long-duration customer agreements. Prysmian’s own 2024 results showed that sustainable products and solutions were already becoming an important sales category.

Rio Tinto faces a similar equation. The tonnes involved are unlikely to transform an aluminium division that operates at global commodity scale, but successful use can create commercial evidence for ELYSIS and support a premium or preferred-supplier position as output expands. In May, Rio Tinto began commissioning a $1.5 billion low-carbon aluminium expansion at Complexe Arvida in Quebec, adding capacity that relies on hydropower and advanced smelting technology. That project is distinct from ELYSIS, but it illustrates the capital required to reduce aluminium intensity and the need for credible customer demand. A recognisable installation at Amazon helps connect industrial investment with a buyer rather than leaving low-carbon capacity as an abstract corporate target.

Amazon’s benefit is partly reputational, but it is also operational and strategic. Data-centre developers face criticism over electricity demand, grid congestion, water consumption and the local impact of rapid construction. Lower-emission material does not resolve those issues, yet it broadens the set of levers available to reduce the footprint of new capacity. It may also improve supply-chain visibility at a time when access to transformers, switchgear and conductive metals can constrain build schedules. The company gains the most if the Ohio deployment produces measurable performance and procurement data that can be used in later facilities.

How are investors likely to read the Prysmian, Rio Tinto and Amazon announcement?

Investors are likely to classify the announcement as strategically positive but financially unquantified. Rio Tinto’s New York-listed shares closed at $97.37 on 18 September, down about 0.7% for the session, while Amazon rose about 1% to $253.71. Those moves occurred in broad market trading and cannot reasonably be attributed to a cable order whose terms were not released. The lack of a visible transaction value limits any near-term earnings revision, while the participation of three established companies raises the credibility of the technology and the end-market use case.

Prysmian’s more important investor question is execution across a much larger North American expansion. The group is balancing organic growth with major acquisitions and financing demands, so shareholders will judge whether data-centre demand supports margins and returns rather than simply adding volume. The Ohio order supports the strategic thesis that cable makers can sell differentiated systems into the artificial intelligence infrastructure cycle. It does not remove the integration, leverage or project-execution risks attached to that wider strategy.

For Rio Tinto, the project adds a proof point to a portfolio that still depends heavily on commodity prices and large-scale mining operations. Technology-led differentiation can improve customer relationships, but investors will ultimately look for production scale, cost competitiveness and evidence that lower-carbon aluminium earns durable premiums. Amazon shareholders are even less likely to see the project as financially material in isolation. Their interest lies in whether procurement innovation helps the company build capacity faster, control lifecycle emissions and avoid future regulatory or customer pressure without materially increasing capital intensity.

Can inert-anode aluminium scale beyond a showcase data-centre project?

Scaling depends on industrial reliability, electricity supply, qualification standards, production cost and customer willingness to commit. Aluminium smelters operate continuously and at enormous electrical loads, so a new anode system must perform safely and predictably over long campaigns. Cable makers must also establish that metal produced through the route meets the mechanical, conductivity and manufacturing requirements of the intended product. A successful Ohio installation would answer only part of that question, but field use can shorten the distance between laboratory claims and routine procurement.

Traceability will be equally important. A buyer needs credible chain-of-custody information to distinguish ELYSIS metal from conventional units as it moves through casting, rod production, cable manufacturing and installation. Without that discipline, the emissions attribute risks becoming a book-and-claim exercise that cannot be connected clearly to a physical asset. The announcement implies coordination across the chain, but the companies have not yet published the accounting method, the product carbon footprint or the avoided-emissions estimate. Those are the details that would allow customers and investors to compare the product with alternatives.

The decisive signal will be repetition. If Prysmian incorporates the material into standard product families, Rio Tinto increases commercial ELYSIS volumes and Amazon specifies the cable across multiple campuses, the Ohio project could mark the beginning of a procurement category. If the arrangement remains a small showcase with no disclosed follow-on volumes, its strategic value will be more limited. The current announcement is therefore best read as a credible industrial trial with three capable participants, not as proof that data-centre construction has already solved its aluminium emissions.