Industry Data for the Energy Transition

Commodity prices: metals, materials and chemicals?

Annual commodity prices are tabulated in this database for 70 material commodities, as a useful reference file; covering steel prices, other metal prices, chemicals prices, polymer prices, with data going back to 2012, all compared in $/ton. 2022 was a record year for commodities. We have updated the data-file for 2023 data in March-2024.

Material commodity prices flow into the costs of producing substantively everything consumed by human civilization, and increasingly consumed as part of the energy transition. Hence this database of annual commodity prices is intended as a useful reference file. Note it only covers metals, materials and chemicals. Energy commodities and agricultural commodities are covered in other TSE data-files.

Source and methodology. The underlying source for this commodity price database is the UN’s Comtrade. This useful resource covers trade between all UN member countries, across thousands of categories, in both value terms ($) and mass terms (kg). Dividing values (in $) by masses (in kg) yields an effective price (in $/kg or $/ton). We have then aggregated, cleaned and averaged the data for 70 materials commodities.

The median commodity in the data-file costs $2,500/ton on an unweighted basis. Although this ranges from $20/ton for aggregates to $75M per ton for palladium metal.

2022 was a record year for material commodity prices. The average material commodity priced 25% above its 10-year average and 40 of the 70 commodities in the database made 10-year highs.

Steel prices reached ten-year highs in 2022, averaging $2,000/ton across the different steel grades that are assessed in the data-file. This matters as 2GTpa of steel form one of the most important underpinnings in all global construction. Our steel research is aggregated here.

Base metal prices averaged 40% above their ten-year averages in 2022, as internationally traded prices rose sharply for nickel, rose modestly for aluminium and zinc, and remained high for copper (chart below).

Battery metals and materials prices rose most explosively in 2022, due to bottlenecks in lithium, cobalt, nickel and graphite. This is motivating a shift in battery chemistries, both for vehicles and for energy storage. It also means that the average battery material in our data-file was higher priced than the average Rare Earth metal in the data-file (which is unusual, but not the first time).

Commodity chemicals all rose in 2022 across every category tracked in our chart below. These chemicals matter as intermediates. On average, sodium hydroxide prices reached $665/ton in 2022, sulphuric acid prices reached $140/ton and nitric acid prices reached $440/ton.

500MTpa of global plastics and polymers demand is covered in our plastics demand database. Both finished polymer prices (first chart) and underlying olefins and aromatics (as produced by naphtha crackers, second chart) prices rose sharply in 2022. Our recent research has wondered whether terms of trade are likely to become particularly constructive for polyurethanes.

Silicon prices matter as they feed in to the costs of solar, and traded silicon prices also reached ten year highs in 2022, before correcting sharply in 2023. Silica prices surpassed $70/ton, silicon metal prices reached $4,000/ton and polysilicon prices surpassed $30/kg (charts below).

The full database captures 70 globally traded materials commodities and their annual prices over time in $/ton, year by year, from 2012-2022. These are: Acrylonitrile prices, Adhesives prices, Aggregates prices, Aluminium prices, Ammonia prices, Battery Graphite prices, Benzene prices, Butadiene prices, Carbon Fiber prices, Cement prices, Cobalt prices, Cobalt Oxide prices, Cold Rolled Steel prices, Concrete prices, Copper prices, Copper Wire prices, Cumene prices, Electric Motor and Generator prices, Electrical Transformer prices, Epoxide prices, Ethanol prices, Ethylene prices, Ethylene Oxide prices, EVA prices, Formaldehyde prices, Glass Fiber prices, Gold prices, Graphite Anode prices, Graphite paste prices, HCl prices, HDPE prices, HF prices, Hot Rolled Steel prices, Hydrogen Peroxide prices, Integrated Circuit prices, LDPE prices, LiPF6 prices, Lithium Carbonate prices, Lithium Metal prices, Manganese prices, Manganese Oxide prices, Methanol prices, NaCN prices, Nickel prices, Nitric Acid prices, Paint prices, Palladium prices, PET prices, Phosphoric Acid prices, Platinum prices, Polyethylene prices, Polysilicon prices, Polyurethane prices, Propylene prices, Propylene Oxide prices, PTFE prices, Rare Earth Magnet prices, Scandium & Yttrium prices, Silica prices, Silicon Metal prices, Silver prices, Sodium Hydroxide prices, Stainless Steel prices, Steel Alloy prices, Sulfuric Acid prices, Toluene prices, Tubular Steel prices, Urea prices, Vehicle prices, Xylene prices, Zinc prices.

Oscar Wilde noted that the cynic is the man who knows the price of everything, but the value of nothing. To avoid falling into this trap, we also have economic models for most of the commodities in this commodity price database.

We will continue adding to this commodity price database amidst our ongoing research. You may find our template useful for running Comtrade queries of your own. Or alternatively, if you are a TSE subscription client and we can help you to use this useful resource, then please do email us any time.

Duck curves: US power price duckiness over time?

In solar-heavy grids, power prices trough around mid-day, then ramp up rapidly as the sun sets. This price distribution over time is known as the duck curve. US power prices are getting 25-30% more ducky each year, based on some forms of measurement. Power prices are clearly linked to the instantaneous share of wind/solar in grids.

The famous duck curve shows how intra-day power prices are impacted by the rise of solar, rising gently in the morning, troughing in the middle of the day, then rising rapidly in the evenings after the sun has set. Apparently this looks like a duck. But is the duck curve getting more ducky over time, as solar gets built out?

This data-file aims to measure the duckiness of duck curves, over time, across the big five US grid regions: CAISO, ERCOT, MISO, PJM and SPP. On average, over the past 3-years, pricing ramps from c$40/MWH at mid-day to $65/MWH at 6-8pm, partly due to solar generation profiles, and partly due to other demand patterns.

3-year average wholesale marginal price for the Big-Five US grid regions.

The duckiness of the duck curve has risen over time, across these grid regions, as solar scaled up from 3% of US electricity in 2020 to 6% in 2023. In 2021, power pricing at 6-8pm was 30% higher than at 11am-1pm, in 2022 it was 45% higher, in 4Q23 it was +56%, and in 3Q24 it was +110% higher (chart below).

Duckiness of US power prices from 2021 to 3Q 2024. Measured as the increase from noon to evening power prices.

However, there is a vast amount of volatility in the data. Other cuts show a less clear increase in duckiness, as shown below, averaging across our big-five regions.

Wholesale average marginal power prices by quarter for the Big-Five US grid regions.

California makes for the most direct case study of duck curves, as utility-scale solar comprises 25% of its electricity mix in 3Q24, up from 15% in 3Q21. In the past, we have looked at individual nodes in California from CAISO, as compared apples-to-apples in individual months, which does appear to show rising duckiness.

California electricity price change between August 2021 and August 2023 — California IntraDay Wholesale Power Prices in 2023 and in 2021

But again, other cuts show a more volatile pattern for CAISO, with strong seasonal effects. Perhaps duckiness has also been muted by a large battery build-out, doubling every year, with batteries supplying an average of 3GW from 8-9pm in 3Q23 and an average of 6GW from 8-9pm in 3Q24 (chart below).

CAISO TTM grid share by generation source from 3Q 2021 to 3Q 2024

The most significant driver of power prices that we can find in the file is the call on non-wind and non-solar generation. Prices spike when renewables are not generating and markets must be balanced by ramping up gas peakers or disincentivizing demand.

Power prices depending on renewables grid share for ERCOT and CAISO.

As a simplified rule of thumb, average power prices rise (fall) $2-3/MWH for every 5% decrease (increase) in the share of renewables in the grid. In CAISO, when marginal prices fall below $10/MWH it is almost always associated with wind and solar supplying >80% of the grid, and when prices rise above $100/MWH, wind and solar are usually supplying <10%.

We do think power grids are growing more volatile over time. This is yet another tracker, breaking down the hour-by-hour patterns and duckiness.

Decarbonize shipping: alternative fuel costs?

Cost of shipping as a percent of final product costs for different ship fuels and their CO2 intensities. Green fuels can be low-carbon, but expensive

This data-file screens the costs of alternative shipping fuels, such as LNG, blue methanol, blue ammonia, renewable diesel, green methanol, green ammonia, hydrogen and e-fuels versus marine diesel. Shipping costs rise between 10% to 3x, inflating the ultimate costs of products by 0.1-30%, for CO2 abatement costs of $130-1,000/ton. We still prefer CO2 removals.

Shipping consumes 5Mbpd of global oil demand, emits 1.5% of the world’s CO2, and adds c1% to the final cost of a typical shipped product, using $1.8/gallon marine diesel at 9.0 kg/gal CO2 intensity.

This data-file appraises the costs of alternative shipping fuels, drawing on models from our prior work into methanol, lower-carbon ammonia, renewable-diesel, green hydrogen, and electrofuels.

In each case, we have estimated the increased fuel costs of alternative shipping fuels versus marine diesel; plus the increased capex costs of ships that can handle these different fuels, increased maintenance costs and other increased operational costs. This draws on our models of container ships, bulk shipping, LNG tankers and other vessel types.

The title chart above shows a base case where shipping with marine diesel fuel adds 1% to the final price of a product that is transported between continents, and emits about 100kg of CO2 per ton of product that is shipped. Alternative shipping fuels add 0.1 – 3.3% to this baseline cost.

LNG is most competitive, adding just c10% to total shipping costs in LNG-fueled ships – possibly much less, or even deflating costs, when oil prices are higher, or LNG prices are lower. But LNG only lowers CO2 emissions by c10%. And even this is debatable, if a gas-fired marine engine suffers from methane slip.

Blue methanol, blue ammonia and renewable diesel are next most economical, but add 0.4 – 0.7% to the final costs of shipped products, while achieving 60-70% reductions in the CO2 intensity of shipping. This equates to a decarbonization cost of $135-260/ton.

Most costly are green methanol, green ammonia, green hydrogen and e-fuels, which add 2.3-3.3% to the final costs of shipped products, while achieving 80-90% reductions in the CO2 intensity of shipping. Thus, the decarbonization costs are an eye-watering $700-1,000/ton.

The numbers do vary markedly, however, based on the products being shipped, especially their mass, their costs and the shipping distance, which can all be stress-tested in the data-file.

For bulk products such as sugar, iron ore or grains, shipping using marine diesel can comprise as much as 5-10% of product prices, hence switching to the green fuels above can inflate end product costs by 20-30%.

Cost of shipping as a percent of final product costs for different ship fuels. Green fuels can be prohibitively expensive

Conversely, for light but high-value products, such as iPhones, shipping costs are basically irrelevant. You can use any fuel you like, and it will not even sway final product prices by 0.00%. Most other products are in between. Numbers can be stress-tested in the model.

The most economic options to decarbonize shipping are through larger and more efficient ships, using high-quality hydrocarbon fuels and coupling these ships with nature-based CO2 removals. Decarbonization must increasingly prove it can be competitive. We have also looked at carbon capture on ships.

US power generation under development over time?

An all-time record of 180GW of new power generation is currently under development in the US in 4Q24, enough to expand the US’s 1.3TW power grid by almost 15%. This data-file tracks US power generation under development, as a leading indicator for gas turbine, wind, solar and battery demand. Gas turbines and battery co-deployments are accelerating in 2024, while wind and solar initiations are slowing on grid bottlenecks?

This data-file captures the development pipeline of new US power capacity, based on 860M reports from the EIA, which cover all existing and proposed generating units of >1MW of greater. As a leading indicator for wind, solar, gas turbine and battery demand, we have aggregated the data in these c110 monthly reports, from 2015 to 2024, to track the pipeline over time, and how expectations have progressed.

Over the past decade, an average of 4 GW of new power generation projects have been added to the queue each month. 1 GW of previously proposed projects have been abandoned each month. And another 2.2 GW of projects have been completed each month. Hence, the overall project queue has grown 0.8 GW larger per month, rising from 90 GW in September-2015 to 183 GW in September-2024 (charts above).

New power generation capacity projects initiated in the US from Q3 2015 to Q3 2014.

Fears over power grid bottlenecks and rising interconnection times are strongly supported by the data. 75% of the increase in the overall pipeline size is in projects that have not yet commenced construction and are thus effectively sitting in a queue.

Development times have not changed materially, although they have always been quite variable. In the data-file, we have tracked 1,715 projects from the time they were first proposed, to the time when they were completed. Their average construction time was 0.8 years, with an average delay of 0.4 years versus initial estimates. These are shorter than the development times for other energy infrastructure.

Looking across the life-cycle of projects that entered the 860M reports during the planning stage (rather than later), the average development time was 2.1 years, including 0.9 years of planning, 0.3 years of permitting, 0.8 years of construction, and an average delay of 0.6 years versus initial estimates. Larger projects tend to take longer.

Development times vs planned capacity of gas, solar, and wind power projects in the US

Delays in constructing power generation facilities are also heavily skewed, as 10% of the projects comprise 50% of the delays.

Distribution of delays in start-up times for gas, solar, wind, battery, and other power projects versus their originally planned timelines

In 2024, renewables momentum has slowed, gas has re-accelerated, but grid-scale battery activity is accelerating fastest and now making an all-time peak, based on tracking new projects being added to the EIA’s 860M filings. Numbers are in the data-file for TSE clients.

Gas, solar, wind, and battery power projects initiated in the US from 2016 to 2024. Each data point is for the trailing twelve months

Again this supports the notion that bottlenecked power grids are hindering the ramp of wind and solar, while we specifically see battery co-deployments as a route to expedite bottlenecked projects. The re-acceleration of natural gas projects also supports our outlook on US natural gas and our outlook on gas turbines. We will continue updating this data-file over time.

Solar plus batteries: the case for co-deployment?

US natural gas: the stuff of dreams?

EV incentives: vehicle taxes by country?

Taxes on new ICE vehicle purchases in different countries

Vehicle taxes by country are tabulated in this data-file, based on vehicles’ pre-tax prices, tailpipe emissions, weight, engine size and power. They range from <10% of the cost of the underlying vehicle in the US, through to 150% in Norway, and can also be well above 100% in other Northern European countries such as Netherlands, Denmark and France.

Super-high taxes on ICEs have been successful in promoting EV adoption, especially in Northern Europe, but how palatable is this option more broadly, especially in countries with large domestic auto industries?

In one of the most entertaining energy-themed advertisements of all time, Will Ferrell laments Norway’s lead over the United States in electric vehicle ownership. 90% of Norway’s new vehicle purchases were BEVs/PHEVs in 2023, and electric vehicles make up 26% of the fleet. What is not in the advertisement is the tax policy that has propelled EVs to such high adoption in many Northern European economies.

This data-file quantifies vehicle taxes by country, which turns out to be a complex calculation, with sliding-scale formulae linked to vehicles’ tailpipe CO2 emissions (Norway, Netherlands, France, Denmark, UK, Germany), weight (Norway, France, Australia, Japan), value (all geographies, but especially Denmark) and engine power (Italy, France, Japan, Germany).

ICE vehicle taxes by country are plotted above, for a vehicle with a $25k pre-tax vehicle purchase price, 1.8 ton gross weight, 30mpg fuel economy, and 2.0L engine with 180hp of engine power. You can stress-test all of these variables in the data-file, and the tax consequences flow through the file. Typical vehicle parameters are available here.

In the average country globally, taxes add c35% onto the pre-tax purchase price of an ICE vehicle. However, the range is wide, varying from <10% in many US States, to >100% in France, Denmark, Netherlands and of course Norway, which reaches 150%. Yes, a vehicle with these parameters costs 1.5x more in taxes in Norway than the vehicle itself.

Tax exemptions for electric vehicles are offered in almost all of these countries. Norway, for example, has exempted new vehicles from both VAT and other purchase taxes. In Denmark and the Netherlands, EVs receive large deductions from vehicle purchase taxes. In many countries, EVs also receive direct fiscal incentives.

Decelerating EV sales growth has been a theme that has worried us in our 2024 research. One factor that could re-accelerate EV sales growth is the ratcheting up of taxes on ICE vehicles. But on the other hand, it is interesting to note that the countries that have implemented large vehicle taxes tend not to have a large domestic auto industry. Whereas for obvious reasons, there may be opposition to inflating the costs of new vehicle purchases by 2x from leading vehicle makers in their home markets.

Mapping vehicle value chains?

Electric vehicles: the road to cost parity?

Electric vehicles: saturation point?

Solar+battery co-deployments: output profiles?

Output of a solar+battery co-deployment power plant on a typical summer day.

Solar+battery co-deployments allow a large and volatile solar asset to produce a moderate-sized and non-volatile power output, during c40-50% of all the hours throughout a typical calendar year. This smooth output is easier to integrate with power grids, including with a smaller grid connection. The battery will realistically cycle 100-300 times per year, depending on its size.

The output from a standalone solar installation is notoriously volatile, varying +/- 5% every 5-minutes on average, plus sudden power spikes and drops, and achieving an annual utilization factor of just 20%.

But how does co-deploying solar+batteries lower the volatility? This data-file uses real-world data, from an Australian solar asset, measured at 5-minute intervals, and then applies simple rules about when to flow power into and out of the batteries, to maximize the delivery of 100MW, smooth, non-volatile power.

The solar+battery output also includes synthetic inertia and frequency regulation, which helps rather than hinders overall grid stability.

The title chart above shows how the output profile of our solar+battery system might behave on a summer’s day, with the net asset providing 100MW to the grid for 24-hours. Excess solar is shunted to the battery throughout the day. Then the battery is gradually discharged to zero after sunset. This model works well in the summer.

However solar generation is highly seasonal, and on a winter’s day, this exact same battery does help to keep output stable at 100MW, but it only achieves 20% of a full charge-discharge cycle, as there is simply not enough solar generation to fill the battery. The bigger the battery, the less likely it gets full in the summer, and the less utilized it is in the winter.

Output of a solar+battery co-deployment power plant on a typical winter day.

This can be stress-tested in the data-file. We can also calculate the number of charge-discharge cycles that different batteries achieve, if they are charged exclusively with solar generation. Some decision-makers assume daily charging-discharging when modeling the economics of batteries, but this is shown to be much too optimistic (below).

Number of charge-discharge cycles achieved by a battery per year depending on the ration of battery capacity to co-deployed solar capacity

Overall, remarkably, solar+battery co-deployment model means that a 275MW solar installation + a 275MW battery can dispatch 95% of its generated output through a mere 100MW grid connection. This is why co-deploying renewables+batteries can help to surmount power grid bottlenecks. And in turn, this is why we think battery co-deployment is accelerating.

How much solar power is dispatched (ie utilized) for a 275MW solar project depending on the size of its grid connection and capacity of co-deployed batteries.

If a battery is run purely for solar smoothing, with 1MW of battery capacity per MW of solar, then the battery will tend to achieve 180 charge-discharge cycles throughout the year, and it will allow a 275MW solar asset to output precisely 100MW to the grid in c50% of the time throughout an entire year (but still producing no power about 40% of the time).

The production profiles vary month by month. The results vary with battery sizing and charging-discharging rules. These sizings and rules can be stress-tested in the data-file, to assess how different-sized batteries result in different dispatch rates and charge-discharge cycle counts.

Gas distribution: loss rates, leakage, unaccounted gas?

What are the loss rates in gas distribution? 1-4% of all the gas that flows into downstream gas distribution networks may fail to be metered and monetized. Stated leakage rates are usually around 0.5%, but could be higher. This data-file aggregates data from Eurostat and the UK’s Joint Office of Gas Transporters.

1-4% of all of the natural gas that flows into downstream gas distribution networks may fail to be metered and monetized. This matters not just for avoiding methane leaks, but also should ideally be improved before integrating biogas or blending hydrogen, or for ensuring that similar issues do not lessen trust in CCS value chains.

In the US, the variation between gas inputs to the gas distribution network and gas that is ultimately metered by customers is designated as Lost or Unaccounted For (LAUF). LAUF gas is generally estimated at 1-4%, and one study quotes an average rate of about 2% across the nation. In the past, the American Gas Association has somewhat questionably stated LAUF is mainly a metering and accounting adjustment and [sic] “there is no correlation between LAUF and emissions”.

In the UK, the variation between gas inputs to the gas distribution network and gas that is ultimately metered by customers is split between Shrinkage (losses, mostly leaks) and other Unidentified Gas (UIG, e.g., due to theft, or pressure-temperature differences at metering sites). In gas year 2023/24, 0.6% of the gas flowed into the downstream distribution network was lost as shrinkage and 3.5% as Unidentified Gas, based on data from the UK’s Joint Office of Gas Transporters (see below).

Build-up of Unaccounted Gas in the UK. Most of it is allegedly due to theft, not leakage.

Stated loss rates in developed world distribution networks are usually around 0.5%, based on data from Eurostat, but higher LAUF/UIG rates have opened the door to gas skeptics alleging higher leakage.

Power generation: asset lives?

Asset lives of different power generation sources.

Power generation asset lives average c70-years for large hydro, 55-years for new nuclear, 45-years for coal, 33-years for gas, 20-25 years for wind/solar and 15-years for batteries. This flows through to LCOE models. However, each asset type follows a distribution of possible asset lives, as tabulated and contrasted in this data-file.

Asset lives of power generation infrastructure are tabulated in this data-file, covering both the design life and age at retirement, for coal, gas, wind, solar, batteries, nuclear and hydro.

Average lives are c70-years for large hydro, 55-years for nuclear, 45-years for coal, 33-years for gas, 20-25 years for wind/solar, 15-years for batteries. However, the numbers follow a distribution, as can be quantified based on data in the data-file.

Distributions of lifetimes for different power generation assets.

Thus, the capex of c$1,000/kW for wind, solar and batteries is not necessarily cheaper (per year) than $1,000-1,500/kW for gas or $3,000/kW for hydro. These very long-run costs/benefits are not well captured in LCOE models.

Low-carbon baseload: walking through fire?

Our personal perspective is that long-term infrastructure has huge hidden value within stable, developed world countries. Their public benefits continue long after their capex costs have been forgotten. Our favorite example is the Brooklyn Bridge, completed for $15M in 1883, yet still standing today.

Some power plants can also be replaced and re-fitted, piece by piece, like Theseus’s Ship. It might cost $650/kW to extend a nuclear plant’s life by a further 20-years (attractive for data-centers, and stoking the order books of nuclear contractors, such as Westinghouse, now owned by Cameco).

Likewise for new energies, there may be upside in the 2030s for module-makers, turbine-makers and battery materials and manufacturers, as existing assets need to replace failing components.

Vehicle emissions of CO, NOx and HCs?

There has been a remarkable reduction in the negative air quality impacts of combustion vehicles since 1970, as quantified in this data-file and over time. Vehicle emissions of CO, NOx and HCs have all fallen by 20-60x over the past 50-years, to 5 grams/mile, 0.2 grams/mile and 0.3 grams per mile, respectively.

This data-file quantifies vehicle emissions of CO, NOx and HCs across the active US fleet, using data reported by the BTS. The reported BTS data series go back to 1990, however, we have been able to take the data-series back to 1970, by interpolating between other data-sets, such as the total miles driven across the US since 1970.

Vehicle emissions of CO. The average modern ICE vehicle emits 4 grams of carbon monoxide (CO) per mile, while comparable vehicles 50-years ago emitted 20x more CO.

Vehicle emissions of NOx. The average modern ICE vehicle emits 0.2 grams of NOx per mile, while comparable gasoline vehicles 50-years ago emitted 30x more NOx, and comparable diesel vehicles emitted 40x more.

Gasoline vehicle emissions of HCs. The average modern gasoline vehicle emits 0.3 grams of uncombusted hydrocarbons per mile, while a comparable vehicle 50-years ago emitted 35x more.

Diesel vehicle emissions of HCs. The average modern diesel vehicle emits 0.2 grams of uncombusted hydrocarbons per mile, while a comparable vehicle 50-years ago emitted 60x more.

The main reason for these reductions in air emissions has been improving engine technology and the use of PGMs within catalytic converters, as mandated by emissions standards.

Going platinum: PGMs in energy transition?

Continued improvements will come from electric vehicles which do not have any tailpipe emissions at all. Please see our broader vehicles research.

Global heat pump sales by country?

Heat pumps sold in different geographies from 2012 to 2023. 2023 was the first year in this database that sales declined.

Global heat pump sales by country are tabulated in this data-file, for 14 countries/regions. Developed world heat pump sales rose at an 11% CAGR over the decade since 2012, reaching 7M units sold in 2022, but then unexpectedly fell by -10% in 2023, including YoY declines in 7 out of the 14 countries we are tracking.

How are heat pumps defined? In the broadest sense of the term, a heat pump is any small-medium sized modular device that evaporates a refrigerant against a heat source (absorbing heat), then re-releases that heat elsewhere by compressing and re-condensing the refrigerant (releasing the heat). Strictly, therefore, all air conditions are heat pumps.

However, in this data-file, we are hoping to tabulate global heat pump sales by country, defining a heat pump as a system that is largely used to transfer heat into a space or a system, and competing with other forms of boilers and heating systems.

This exercise is relatively challenging, as some regions do not report heat pump sales at all, and others report heat pumps with different definitions. For example, Australia sold an enormous 1.4M heat pumps in 2023, including air conditioners; but of this total, only c160k were linked to hot water heating systems, and therefore we have estimated true heat pump sales somewhere closer to 300k.

Across the developed world, we think that heat pump sales reached 7M in 2023, or around 6 heat pumps sold per million inhabitants, then fell back to 6M heat pumps sold in 2024, with YoY declines in 7 out of the 14 countries/regions in our data-file (the US, Canada, Italy, Spain, Sweden, other Europe, and Japan). This is another data-point warranting caution over drawing S-curves in new energies.

Europe has some of the highest heat pump sales among regions in our database, with sales of around 10 units per year per thousand inhabitants in France, and 20-30 units per year per thousand inhabitants in Northern European countries such as Norway and Sweden.

The largest category is reversible air-air units (45%), reversible air-water (30%), while only c10% of the sales in Europe have been ground-source heat pumps, based on data from the European Heat Pump Association. Projections for European heat pumps are found in our European natural gas model.

Installed heat pump stock in Europe by category. Most common is air-air heat pumps, while ground-source is the smallest group.

US residential and commercial HVAC system deployments are available from AHRI. There was a surge across gas boilers, gas water heaters, air conditioners and heat pumps in 2021, linked to higher home construction. Heat pumps rose from 30% of cooling solutions to 40% from 2013 to 2023, and rose from 15% to 23% of heating solutions (ex electric heaters, where AHRI does not report data) (chart below).

US residential and commercial HVAC system sales from 2010 to 2023.

Heat pump sales are estimated in other regions based on public data sources. Heat pumps are clearly more efficient than combustion-based heating, as they can generally convert 1 kWh of electricity into 4 kWh-th of heat transfer. However, we think that high costs and challenging practicalities may still be a hurdle for heat pumps, and rather than following an S-curve, sales may, like EVs, follow a saturation curve.

Cookies?