In a rapidly decarbonising world, higher carbon prices will raise power costs most for assets on fossil-heavy grids, reshaping the cost curve. Power-intensive aluminium's cost floor observes the sharpest rise in the medium term, while zinc and copper floors climb later.
Our central case for global greenhouse gas emissions shows a trajectory in line with a 2.5°C–3.0°C temperature pathway above pre-industrial levels, at least over the next decade. Using our central case as benchmark, we present a transition scenario through 2035 that combines both sharper carbon price trajectories and rapidly growing shares of renewables in electricity generation required to meet climate goals.
We use these paths to assess how higher carbon prices – passed through to the cost of purchased electricity (indirect, Scope 2-related emissions) – affect metal cash costs and, in turn, the price floor set by marginal producers on the cost curve. This is an important channel because electricity is a large part of cash costs for many metals. Aluminium and copper are the most power-intensive on a per tonne basis, using on average 7.3 and 5.9 MWh per tonne, respectively. Power makes up almost 29% of aluminium cash costs, ~13% for copper, ~11% for zinc and almost 9% for lithium. A sharp rise in carbon prices can, therefore, push their cash costs up more than for most other metals.
Steel is the least exposed through purchased electricity, though exposure depends on the production route. Electric arc furnace (EAF) production uses about five times as much power per tonne as the blast furnace–basic oxygen furnace (BF-BOF) route (0.51 against 0.10 MWh/t), and power makes up 9.5% of its cash costs, against just 1.9% for BF-BOF. Steelmakers under EAF have lower direct emissions, but more of their costs depend on grid power.
Carbon prices and grid mix reveal an uneven cost burden
As well as major differences in exposure across commodities, there are major differences in country-level exposures. We group producing countries in four blocs, based on each country's expected carbon price uplift to 2035 in the transition scenario against their current fossil share of power generation:
- High carbon price uplift, fossil-heavy grid: Australia stands out as the most exposed. It combines a fossil-heavy grid (about 60%) with the largest expected uplift at $240 /t CO2. South Africa and Mexico face a similar double exposure, with uplifts of about $130–140 /t and fossil shares above 75%.
- High carbon price uplift, renewable-energy grid: Latin America looks less exposed than its carbon price uplift alone would suggest. Uplifts for Brazil, Colombia, Peru, Chile and Argentina are well above $100 /t. However, hydropower dominates much of the region's power mix, which keeps fossil shares at or below about 50% and will limit how much of the carbon price reaches electricity costs.
- Low carbon uplift, renewable-energy grid: Canada and Europe are the least exposed on both counts. Both have fossil shares of about 20–30% because of large nuclear and wind fleets, and fast-growing solar. Both also face relatively modest uplifts, of $80 /t for Canada and close to $10 /t for Europe, because we expect them to decarbonise relatively quickly in the central case.
- Low carbon uplift, fossil-heavy grid: Several of the largest fossil-heavy producers – including China, India, Indonesia, Kazakhstan and the Middle East – face uplifts below $60 /t by 2035. Much of their power cost advantage is, therefore, likely to remain, at least over the next decade.
Our in-house power price model estimates industrial electricity prices by country from generation costs, the projected power mix and the extra costs of supplying industrial users. Carbon prices raise coal- and gas-fired generation costs in line with their emissions intensity. Coal and gas set the price reference until wind and solar pass 40% of supply. After that, the model gradually shifts weight to the output-weighted average cost of building and running power plants, while fossil fuels still influence prices when renewable output is low. We then add network, policy, balancing and reliability costs, minus assumed industrial reliefs.
Same power price shock transmits differently across commodities
The ten countries that together cover most global output of these commodities illustrate the uneven distribution of the power price shock. Mexico sees the largest rise in power prices. By 2035, its real power price is $51 /MWh (60%) higher in the transition scenario than in the central case. Australia comes next, with a rise of about $43 /MWh (37%) to about $158 /MWh, the highest level of the ten. India, the US and Russia follow, with increases of about $16–25 /MWh. In Congo, China and South America, power prices barely change, as renewable energies have already become the main source of electricity generation by 2035.
The US shows how the energy transition can raise power prices without a higher carbon price. It has no national carbon price, only a few state and regional schemes, and we assume this remains the case in both scenarios. Therefore, cheap domestic gas remains competitive with renewables, and gas-fired plants keep setting the power price. Around 2035, renewables grow large enough that prices start to follow the average cost of the whole generation mix instead. In the absence of a national carbon price, renewables are more expensive than gas in the US, so power prices rise sharply from then on.
Cash cost impacts depend on both the power price change and power intensity for that commodity. Copper is at the top, with cash costs rising by about $407 /t (8.8% of total cost) in Mexico and $322 /t (4.8%) in Australia. Aluminium cash costs rise by about $195 /t (11%) in Australia and $151 /t (6.3%) in the US, but by only about $37 /t (1.5%) in China, the largest producer. Australian lithium rises by about $189 /t (6.3%) and Mexican zinc by about $222 /t (6.1%). Steel is barely affected – the rise is below $35 /t (6.1%) on the EAF route and below $7 /t (1.8%) on the BF-BOF route. In Chile and Congo, which are both major copper producers, the cost change is close to zero.
Marginal tonnes drive the shifting price floors
The largest country-level cost increases do not necessarily cause the largest shifts in price floors. What matters is how exposed the marginal cost band is, which we define as the production-weighted average cost of assets in the 88th–92nd percentile. As power prices change, assets move into or out of this band, which changes which countries set the floor. Where one country dominates, as China does in aluminium smelting, its power price trajectory remains the key driver. The floor is a reference point, not a hard lower bound or a forecast – if prices stay below marginal cash costs, supply cuts can follow until the market rebalances.
China’s coal-heavy grid sets the marginal price, making power prices sensitive to the higher carbon price in the transition scenario. The real power price premium over the central case reaches $20.5 /MWh in 2030, driving aluminium’s floor uplift to 7.1%. Pressure eases from 2031, when cheaper wind and solar surpass 40% of generation, triggering the switch to a lower power price base through the weighted-average pricing scheme. Chinese smelters dominate the marginal band, where power accounts for ~35% of cash costs, so both the initial shock and subsequent relief feed through strongly. The price floor uplift then settles around 2.6–2.7% through 2035.
Lithium refineries face the same Chinese power shock, but their cost structure limits the impact. In 2032, both face a shock of about $7.8 /MWh in real terms, yet aluminium’s price floor rises 2.8% and lithium’s only 0.4%. This is because power represents only around 6% of lithium cash costs, compared with 35% for aluminium. Lithium also enjoys the same post-2030 relief as the Chinese power mix changes, but the uplift edges up to 0.8% by 2035 as Argentine tonnes enter the marginal band with a widening power premium.
Zinc’s uplift tracks a changing mix of marginal mines. It jumps from 0.7% in 2029 to 2.8% in 2030 as Eritrean tonnes enter the band, relying heavily on fossil fuels for power generation. The uplift peaks at 3.2% in 2031, partly because the , but also because Eritrean tonnes then leave, China’s power premium narrows and Mexican tonnes provide only partial support, pulling the uplift down to 1.6% in 2033. It rebounds to 2.7% in 2035, led by Bolivian tonnes with Mexico and South Africa adding support. Mexico’s premium keeps widening despite renewables exceeding 40%, because gas-fired electricity remains cheaper there.
Copper’s uplift also reflects a changing marginal mix, but follows a steadier upward path than zinc’s, reaching 2.4% in 2035. Its marginal tonnes span several power markets, so no single threshold crossing drives the trend. European and Chinese mines influence the early trajectory, while Australian and Latin American tonnes help shape the result through 2032. From 2033 to 2035, asset reordering materially lifts the floor. The final-year increase also reflects the US pricing switch – wind and solar pass the generation share threshold in 2035, and the real power premium jumps from roughly zero in 2034 to $21 /MWh in 2035.
EAF steel’s floor uplift falls to just 0.07% in 2030, when US plants dominate its marginal band and face almost no power-price premium. For BF-BOF steel, China’s narrowing premium helps keep the uplift at minimal levels almost until the end of our horizon. Both rise in 2035 as US switches to average-pricing, which adds a $21/MWh power premium. US plants then account for roughly two-thirds of EAF marginal tonnes and just over half of BF-BOF’s, but the floor shifts differ sharply – 1.9% for EAF versus 0.26% for BF-BOF. Purchased power represents about 8% of EAF marginal cash costs, compared with roughly 2% for BF-BOF. EAF’s response is cushioned as some heavily exposed plants move out of the marginal band, but for BF-BOF it is low power exposure, rather than asset substitution, that explains the muted response.
Power prices are not the only way decarbonisation affects costs, but they are a crucial one
Our analysis isolates the power-price channel, holding grades, by-product credits, other costs and production volumes at the central case levels rather than capturing the full transition impact. Assets with long-term fixed-price power contracts or their own generation may also face less of the modelled change, depending on their contractual terms and fuel exposure. Fossil fuel costs may themselves change – in either direction - in response to a more rapid energy transition.
Within those limits, the analysis shows how changing power prices feed into marginal production costs, and why the timing, location and scale of that pressure differ across commodities.
© CRU. All rights reserved. This content may not be copied, reproduced, republished, distributed, or otherwise used, in whole or in part, without CRU’s prior written permission. Please refer to CRU's Terms and Conditions.
