Commodity price volatility: energy, metals and ags?

Commodity prices are distributed lognormally, so the average price will tend to be higher than the median price.

Commodity price volatility tends to be lognormally distributed, based on the data from twelve commodities, over the past 50-years. Means are 20% higher than medians. Skew factors average +1.5x. Standard errors average 50%, while more volatile prices have more upside skew.


This data-file contains data plotting the statistical distributions of volatility for twelve major commodities, ranging across energy commodities such as oil, gas and coal; industrial metals such as iron ore, copper and aluminium; precious metals such as gold and silver; and agricultural commodities such as sugar, soybeans and palm oil.

Commodity price volatility tends to be lognormally distributed, based on starting with the charts shown below, then smoothing all of these statistical distributions together, for the title chart shown above. This statistical distribution is intuitive, as prices are effectively uncapped to the upside during commodity shocks, but they are effectively capped to the downside, as commodities cannot sustainedly trade below zero.

A fascinating finding is that when commodities are more volatile overall (e.g., as indexed by standard error) then this is 75% correlated with skew in the commodity, or in other words, the mean tends to run further above the median. In other words, this is another indicator that commodities illustrate more upside volatility than downside volatility.

The positive skew (mean to median ratio) and standard error of commodity prices. These measures turn out to be 75% positively correlated, so rising volatility drives the average price further from the median.

If base case forecasts are thought of as the median price levels of commodities (e.g., $65/bbl for oil over the past 50-years, $8/mcf for global gas, etc), then the data imply that mean average prices will tend to run 1.1 – 1.5x above median expectations.

The final two tabs of the data-file model the lognormal volatility of commodities, illustrating how the value of commodity marketing and trading is likely to rise during the energy transition, as volatility grows on an absolute basis, but also inter-regional volatility is growing due to the ascent of renewables such as wind, solar and hydro.

Fantastic underlying data that helped to build this data-file came from the World Bank pink sheets, which we recommend to anyone looking for free monthly or annual commodity price data.

For more recent and more detailed pricing across a wider range of commodities, please see our commodity price database, for time series that go further back in time to the 1800s, please see our database of very long-term commodity prices, while we also have analysis into the performance of commodities during conflicts and performance of commodities during recessions.

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.

Commodity prices
Steel Price by year by steel grade in $ per ton

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).

Commodity prices
Base metal prices by year and over time for zinc, aluminium, copper, and nickel in $ per ton

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 prices
Battery material prices over time $ per ton for lithium, cobalt, manganese, nickel, LiPF6 and lithium carbonate in $ per ton

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.

Commodity prices
Industrial Acids and Caustic Soda Prices over time. NaH, H2O2, HCl, H2SO4 Sulfuric Acid, HNO3 Nitric Acid, H3PO4 Phosphoric Acid, HCN and HF in $ per 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.

Commodity prices
Polymer prices by year LDPE HDPE PET EVA Polyurethanes Paints and Adhesives in $ per ton
Commodity prices
Olefins and Aromatics Prices over time

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).

Commodity prices
Silica price, silicon price and polysilicon price in $ per ton

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.

Levelized cost of electricity: stress-testing LCOE?

Levelized cost of electricity of different electricity sources, in cents per kWh and their true CO2 intensity, in kg per kWh.

This data-file summarizes the levelized cost of electricity (LCOE), across 35 different generation sources, covering 20 different data-fields for each source. Costs of generating electricity can vary from 2-200 c/kWh. There is more variability within categories than between them. All the numbers can readily be stress-tested in the data-file.


Levelized cost of electricity (LCOE) breaks down the costs of adding new electricity generation, across capex, capital, tax, fuel, O&M, CO2 and T&D, distilled down in c/kWh terms, or $/MWH terms.

We have constructed over 200 economic models calculating the specific levelized costs of onshore windoffshore wind, solar, hydro, nucleargas powercoal powerbiomassRNG, diesel gensets, geothermal, hydrogen, fuel cells, power transmission, batteries, thermal storage, redox flow, pumped hydro, compressed air, flywheels, CCS and nature-based CO2 removals.

The goal in this data-file is to allow for easy comparisons between different power generation options, across 20 different dimensions. We have written that we hate levelized cost, because it is often portrayed as though one energy source will emerge “to rule them all”, whereas there is more variability within each category than between them (see below).

The simple chart below shows how our levelized cost estimates change if we make simple changes in this comparison file: flexing risk-free rates between 1-5%, flexing fuel costs +/- 50%, flexing capex costs by +/- 50%, or changing the distances needed for AC power transmission, CCS pipelines, or other variables.

Levelized cost of electricity of different electricity sources, in cents per kWh-e. Coal, gas, and solar are some of the cheapest but there is a lot of variability within each category. Note that this is on a partial electricity basis, not total.

This kind of stress-testing is really the main point of the data-file, asking questions like: how do levelized costs change with WACCs? Or how do levelized costs change with higher gas prices? How do levelized costs change with capex deflation? What are the best options for lowering CO2 intensities of grids (chart below) without inflating total costs? The data-file answers these questions across several dimensions…

Levelized cost of total electricity of different electricity sources, in cents per kWh, versus their true net CO2 intensity, in kg per kwh-e. Coal is cheap yet polluting while green hydrogen is the opposite. Different gas options tend to be the best.

Capex costs are broken down for each category and are defined as the total installed capex, in $/kWe, which can then be divided by the total number of lifetime operating hours, yielding a number in c/kWh. Usually, the capex estimates in our underlying models draw from both top-down surveys of past projects and bottom-up build-ups.

Levelized cost of total electricity of different electricity sources, in cents per kWh, versus their capex cost, in $ per kWe. The best are coal and gas sources.

Capital costs can be described as the after-tax income that needs to be earned on top of recovering the capex to derive a passable IRR (usually 7-12%), after whatever build-time is incurred prior to start-up (usually 2-6 years). We have taken this requisite after-tax income level from our individual underlying models, where it captures nuances such as time value of money, decline curves, and volatility.

Tax costs come on top of after-tax income. For simplicity, our models assume a 25% corporate tax rate across the board, but not tax breaks or changeable policies. Thus, we can think about the numbers in our data-file as being true economic costs.  

Fuel costs cover the costs of buying gas for a gas plant, coal for a coal plant, hydrogen for a hydrogen plant, etc. By contrast, fuel costs are often zero for renewables. Again, these can readily be flexed in the model, which is especially important for gas value chains, amidst high dispersion in global gas prices.

O&M costs cover operations and maintenance; and are generally going to be lowest for large and simple systems.

T&D costs cover transmission and distribution, to move power to the load center. An advantage for on-site generation is that power can be used directly, whereas the average offshore wind farm in the North Sea needs to be transmitted 20km back to shore, then onwards. AC transmission costs 1.5c/kWh/100km at large scale. For CCS value chains, we also include $3/ton/100km for CO2 transport in the T&D line.

CO2 costs cover the cost for offsetting or disposing of gross CO2 emissions: either via nature-based CO2 removals, using high-quality reforestation at $50/ton; or for CO2 geological disposal in subsurface reservoirs with a base case cost of $15/ton. Otherwise, for CCS value chains, additional costs are reflected in higher up-front capex, higher fueling costs due to energy penalties, and higher maintenance costs.

Other dimensions are also compared for all of the generation sources in our database: TRLs, logistical risks, development times, efficiency factors, CO2 intensity (kg/kWh), typical load factors (%) and land intensity (acres per MW) (see below).

Levelized cost of total electricity of different electricity sources, in cents per kWh, versus their direct land intensity, in acres per MWe. This method only counts land used by the power plants themselves.

Solar insolation: by latitude, season, date, time and tilt?

Solar insolation varies from 600-2,500 kWh/m2/year at different locations on Earth, depending on their latitude, altitude, cloudiness, panel tilt and azimuth. This means the economics of solar can also vary by a factor of 4x. Seasonality is a key challenge at higher latitudes. Active strategies are emerging for orienting solar modules.


1,353 W/m2 of solar energy arrives at the top of the Earth’s atmosphere, based on the Planck Equation, equivalent to almost 12,000 kWh/m2/year. Amazingly, solar is changing the world, even though only c2-3% of this energy is ultimately getting harnessed today.

((The location of the losses in the chart above is also a reason for exploring solar in space, then beaming the power back to Earth)).

50% of all solar energy is inaccessible due to night time (chart below). Another 20-40% is inaccessible as it is absorbed by the atmosphere and clouds (depending on location). And of the insolation that does reach a solar module, only c20-25% is currently converted into useful electricity, in today best HJT modules.

Calculating the insolation ultimately available for solar modules depends on the mass of atmosphere that is traversed by incoming sunshine, which varies hour-by-hour, with the elevation of the sun in the sky (i.e., vertical height) and its azimuth (i.e., compass point bearing).

Calculating these numbers is quite complex, because the Earth is 23º declinated on its axis. Hence the sun’s elevation and azimuth vary hour-by-hour, day-by-day and by location. Nevertheless, the charts below plot elevation and azimuth at a 45-degree latitude, based on 8,760 calculations throughout the year (24 hours per day x 365 days). The latitude can be varied in the data-file, which also contains hour-by-hour granularity.

Insolation at ground level can thus be calculated, based on the mass of air that has been traversed (chart below left). However, fixed solar modules are not always pointed directly at the sun. This can sacrifice 30-60% of the maximum available insolation, simply due to misalignment (chart below right), which is also calculated hour by hour in the data-file.

For fixed modules, losses can be minimized by matching the tilt of the panels to the latitude at which they are situated (chart below). The losses can be reduced even further with solar trackers, which rotate the panels to follow the sun, although this does also add cost.

It is usually best to orient solar modules directly South (in the North Hemisphere). But efficiency may be sacrificed for economics! West-facing panels generate one-third less energy than South-facing ones. But the generation profile is 2-4 hours later, to smooth out the duck curve.

Insolation available to solar modules can realistically vary from 700 – 2,400 kWh/m2/year, depending on latitude and cloudiness. These numbers can be stress-tested in the data-file.

Depending on latitude, generation will also be 0-80% lower in the winter versus the summer. This is visible in the charts above, as high latitudes have short days in the winter, while even when the sun is up, it is only sitting at a low angle in the sky. This seasonality is extremely challenging to back up economically using batteries.

The full data-file allows you to calculate solar insolation, and resultant solar generation, hour-by-hour and then on a fully annualized basis; by stress-testing latitude, elevation, module tile, module azimuth, cloud cover, tracking efficiency and module efficiency. This is helpful for informing our solar economic models. The numbers match our findings from assessing real-world solar volatility. A fantastic resource that helped us with the equations is pveducation.org.

Gas power: does low utilization entail spare capacity?

The US has >400GW of large gas-fired power plants running at 40% average annual utilization. Could they help power new loads, e.g., 60GW of AI data-centers by 2030? This 5-page note shows why low utilization does not entail spare capacity, and in turn, estimates true gas power spare capacity available for loads such as data-centers.


How much gas power spare capacity exists within the US power grid, and could this help to power the rise of AI or the rise of EVs, without having to construct new power generation?

To answer this question, we have aggregated EIA power market data across 1,850 active US gas-fired power generation facilities.

This 5-page note summarizes our key conclusions on the first page, followed by three pages of follow-up charts.

The note covers the generation capacity growth we are forecasting for AI and other new loads; the average utilization rates of gas generation by plant size (in MW) and by state; why low annual utilization cannot simply be translated into spare capacity; and our estimates for how much true spare capacity really exists within the US’s current fleet of gas turbines.

As a general rule of thumb, a typical US gas power generation facility runs at 40% annual utilization, which translates into 60% peak monthly utilization, 80% peak daily utilization and 100% peak hourly utilization.

This research note is available for TSE written subscription clients, while the underlying data behind our assessment of gas power spare capacity are linked below for TSE full subscription clients.

Bill of materials: electronic devices and data-centers?

Electronic devices are changing the world, from portable electronics to AI data centers. Hence what materials are used in electronic devices, as percentage of mass, and in kg/kW terms? This data-file tabulates the bill of materials, for different devices, across different studies.


This data-file captures the bill of materials for electronic devices, such as cell phones, tablets, laptops, hard discs, solid state-drives, printed circuit boards, servers in data-centers, power supply units, adapters, copper cables and fiber optic cables.

Five materials make up c85% of the mass of typical electronic devices: advanced polymers (c20%), steel (c20%), glass (18%), aluminium (12%) and copper (12%). However, the exact numbers vary by product, as shown in the chart above.

Steel is the joint largest material exposure for electronic devices, although this is unsurprising, as steel is the most-used structural material on the planet, and in digital devices as well, it is used for the chassis/enclosure of data-center racks and other components, in switchgears, fans, heat sinks, etc.

Advanced polymers are the single most important material, both by mass and by specialization. HDPE and PVC are often used for electrical insulation in wires, cables and power supply units. PCBs are c35% epoxy resin. Polycarbonates are used in hard drives and optical disc drives. Solid state drives use specialty polymers, such as liquid crystal polymers.

Copper use from the rise of AI is more debatable. For example, several older studies suggest copper use in AI data-centers can range from 30-60 tons/MW. But on the other hand, these older studies may not fully reflect the scale-up of computing density per rack, which could reduce copper use to 10 tons/MW, albeit this would still tighten global copper balances by around 1% per year through 2030.

The ability to thrift out bulk material intensity factors by raising computing performance density, using advanced materials and manufacturing techniques is highly reminiscent of the same trend in new energies (raising solar efficiency, raising battery voltages). This creates opportunities in vapor deposition equipment, advanced polymers, and ultra-high purity materials including tantalum, silver, gold, tin, et al.

Finally, the vast range of advanced materials used in electronic devices and data-center components is shown by the vast number of materials in the data-file: ABS, Al2O3, Aluminium, Barium, Barium Titanate, Benzoic acid polymer, Brass, Calcium Oxide, Carbon, Cardboard, Chromium, Copper, Cromium, Dioxygen, Epoxy Resin, Ethylene Vinyl Acetate, Fan, Ferrous, Fibrous Glass Wool, Glass, Glass Fiber, Gold, HDPE, HVA-2, Iron, Iron Oxide, LCP Polymer, Lead, Li-ion batteries, Magnesium silicate, Magnesium, Magnets, Manganese, Neodymium, Nickel, Palladium, Paper, PCB, Pegoterate, Phenol polymer, Pigment Black 28, Polybutyl Terephthalate, Polycarbonate, Polycarbonate Acrylonitrile, Polycarbonates, Polyimides, Polymers, Polyurethanes, Proprietary, PVC, Silica, Silicon, Silver, Sodium Oxide, Solder, Steel, Styrofoam, Synthetic Rubber, Tantalum, Tin, Titanium, Vinyl Silicone Oil, Zinc.

Electromagnetic energy: Planck, Shockley-Queisser, power beaming?

Electromagnetic radiation is a form of energy, exemplified by light, infrared, ultraviolet, microwaves and radio waves. What is the energy content of light? How much of it can be captured in a solar module? And what implications? We answer these questions by modelling the Planck Equation and Shockley-Queisser limit from first principles.


Electromagnetic radiation is the synchronized, energy-carrying oscillation of electric and magnetic fields, which moves through a vacuum at the speed of light, which is 300,000 km per second.

Most familiar is visible light, with wavelengths of 400 nm (violet) to 700 nm (red), equating to frequencies of 430 (red) to 750 THz (violet).

At the center of the solar system, our sun happens to emit c40% of its energy in the visible spectrum, 50% as infra-red and c10% as ultraviolet, and very little else (e.g., X-rays, gamma rays at high frequency; microwaves and radio waves at high wavelength). But this is not a coincidence…

Planck’s Law: Spectral radiance as a function of temperature?

Planck’s Law quantifies the electromagnetic energy that will be radiated from a body of heat, across different electromagnetic frequencies, according to its temperature, the speed of light, Boltzmann’s constant (in J/ºK) and Planck’s constant (in J/Hz).

In the chart below, we have run Planck’s equation for radiating bodies at different temperatures from 3,000-8,000ºK, including the sun, whose surface is 5,772ºK. Then we have translated the units into kW per m2 of surface area and per nm of wavelength.

Hence by integration, the ‘area under the curve’ shows the total quantity of electromagnetic radiation per m2. If the surface of the sun were just 10% hotter, then it would emit c50% more electromagnetic radiation and 55% more visible light!

Charts like this also explain why the filament of an incandescent light bulb, super-heated to 2000-3000ºC is only going to release 2-10% of its energy as light. Most of the electromagnetic radiation is in the infra-red range here. And this is the reason for preferring LED lighting as a more efficient alternative. LEDs can reach 60-90% efficiency.

Planck’s Law and Solar Efficiency?

Planck’s Law also matters for the maximum efficiency of a solar module, and can be used to derive the famous Shockley-Queisser limit from first principles, which says that a single-junction solar cell can never be more than c30-33% efficient at harnessing the energy in sunlight.

Semiconductor material has a bandgap, which is the amount of energy needed to promote a single electron from its valence band into its conduction band: a higher energy state, from which electricity can be drawn out of a solar cell. For silicon, the bandgap is 1.1 eV.

What provides the energy is photons in light. The energy per photon can be calculated according to its wavelength. This involves multiplying Planck’s constant by the Speed of Light, dividing by the wavelength, and then converting from Joules to electronVolts. For a radiating body at 5,772ºK, the statistical distribution of photons and their energies is below.

So what bandgap semiconductor is best? If the bandgap is too high (e.g., 4eV), then most of the photons in light will not contain sufficient energy to promote valence band electrons into the conduction band, so they cannot be harnessed. Conversely, if the bandgap is too low (e.g., 0.5eV), then most of the energy in photons will be absorbed as heat not electricity (e.g., a photon with 2.0 eV would transfer 0.5eV into electron promotion, but the remaining 1.5 eV simply heats up the cell).

The mathematical answer is that a bandgap just above 1.3 eV maximizes the percent of incoming sunlight energy that can be transferred into promoting electrons within a solar cell from their valence bands to their conduction bands, at 43-44% (chart below).

If we run a sensitivity analysis on the bandgap, the next chart below shows that our 43-44% conversion limit holds for any semiconductors with a bandgap of 1.1-1.35eV, more of a plateau than a sharp peak.

The Shockley-Queisser limit is usually quoted at 30-34%, which is lower than the number above. In addition to the losses due to incomplete capture of photon energy, the maximum fill factor of a solar cell (balancing load, voltage and current) is around 77%, so only 77% x 44% = 34% of the incoming light energy could actually be harnessed as electrical energy. Moreover, in their original 1961 paper, Shockley and Queisser assumed an 87% efficiency limit for impedance matching relative to the 77%, which is why the number they originally quoted was around 30%.

Another issue is that the solar energy arriving at a given point on Earth has been depleted in certain wavelengths, as they are absorbed by the atmosphere. 1,362W/m2 of sunlight reaches the top of the Earth’s atmosphere. While on a clear day, only around 1,000W/m2 makes it to sea level at the equator. We know the atmosphere absorbs specific infrared wavelengths as heat, because this is the entire reason for worrying about the radiative forcing of CO2 or radiative forcing of methane.

Hence for an ultra-precise calculation of maximum solar efficiency, we should not take the Planck curve, but read-out the solar spectrum reaching a particular point on Earth, which will itself vary with weather!!

Multi-junction solar is inevitable?

The biggest limitation on the efficiency of single-junction solar cells is that they only contain a single junction. This follows from the discussion above. But what if we combine two semiconductors, with two bandgaps into a ‘tandem cell’. The top layer has a bandgap of 1.9eV (e.g. perovskite) and the second has a bandgap of 1.1eV (e.g., silicon). The same analysis now shows how the maximum efficiency can reach 44%.

Cells with multiple semiconductors are already being commercialized. For example, we wrote last year about heterojunction solar (HJT) and this year about the push towards perovskite tandems in solar patents from LONGi. It feels like the ultimate goal will be multi-junction cells that capture along the entire solar spectrum (chart below). It will simply take improvements in semiconductor manufacturing.

solar efficiency in record-breaking multi-junction cells

Power Beaming and Other

Elsewhere in the electromagnetic spectrum, this data-file also contains workings into the energy efficiency of microwave energy, transmitting it through space and converting it back to useful electricity via rectennas.

All of the numbers and calculations go back to first principles, in case you are looking to model the Planck Equation, Shockley-Queisser limit, multi-junction solar efficiency, lighting efficiency, or other calculations of electromagnetic radiation energy.

Generac: power generation products?

Cost per kW of Generac product suite as a function of the generators' capacities. Different fuel types are in different colours.

Generac is a US-specialist in residential- and commercial-scale power generation solutions, founded in 1959, headquartered in Wisconsin, with 8,800 employees and $7bn of market cap. What outlook amidst power grid bottlenecks? To answer this question, we have tabulated data on 250 Generac products.


Generac‘s $4bn pa of sales, in 2023, were >50% residential, >35% commercial, 10% other. 80% was domestic within the US, and 20% was international. 12% is attributed to ‘energy technologies’ which includes storage, solar MLPE, EV charging, smart thermostats, electrification, etc. But what about the product mix? How is it exposed to power grid bottlenecks? Or indeed a broader US construction boom?

In this data-file, we have tabulated details on 250 Generac Products. 70% are generators, and another c10% are transfer switches to connect generators to loads. Our data show the breakdown of units by size, fuel, prices and other technical parameters.

Residential solutions comprise 55% of Generac’s revenues, as 6% of US homes now have standby generators, but a smaller share of the SKUs. Our data show the average size (in kW), list prices (in $) and costs (in $/kW), but we also think low efficiency in some of these residential generation units is not entirely helpful for decarbonization aspirations.

Generac’s larger industrial generators range from 100kW to 2MW in size. Generac’s diesel-fired units cost $500/kW and are c30% efficient, while its larger gas-fired units cost $700/kW and are c25% efficient. These units tend to generate 10-40 kW/m3 of space, and discharge exhaust at 700-800ºC, so they could find application amidst grid bottlenecks?

In our base case model for a diesel genset, we assume a 16-20c/kWh LCOE is needed for a $700/kW installation of a 10MW unit with 42% efficiency buying wholesale diesel. Generac units would appear to have higher costs for reasons in the data-file.

In our base case model for a CCGT, we need a 6.5c/kWh LCOE at an $850/kW installation of a 300MW unit with 55% efficiency buying $4/mcf wholesale natural gas. Again, for the Generac units, we get to a higher LCOE, yes this is the area we think could be most economically justified by growing power grid bottlenecks, especially at industrial facilities that can harness the 700-800ºC waste heat.

Outside of the generation business, remaining SKUs comprise transfer switches to connect generators to loads, pressure washers, light towers for construction, pumps, batteries and mobile heaters, hence there may be heavy exposure to construction and infrastructure projects.

Grid connection sizes: residential, commercial and industrial?

Typical Sizes of Grid Connections of different residential, commercial and industrial facilities in kW. The largest connections are needed for green hydrogen, aluminium and data-centers.

What are the typical size of grid connections at different residential, commercial and industrial facilities? This data-file derives aggregate estimates, from the 10kW grid connections of smaller homes to the GW-scale grid connections of large data-centers, proposed green hydrogen projects and aluminium plants. Also included are our notes on each category, data into 5GW of US micro-grids and assessments of the possible winners-and-losers from growing power grid bottlenecks.


A typical home in the developed world currently has a 10-15kW maximum power capacity. Exceeding this load may cause its circuit breaker to trip. Hence some homes may need upgraded grid connections to add electric vehicle chargers or heat pumps.

Office buildings that have crossed our screens typically require 50-500kW grid connections, rising to MW-scale connections for larger office buildings with 15,000m2 of space or more.

Other facilities with which we are all familiar include Walmart Super-Centers (1.3MW average grid connection), medium-sized hospitals (5MW), London tube underground stations (6MW) and airports (c10MW). Although again, capacity varies with size.

One excellent source for these numbers is looking at the sizes of around 1,000 US microgrids, with over 5GW of capacity, of which around 50% were constructed in the last decade, powered by CHPs (50%), gas turbines (17%), diesel generators (10%), solar (12%), wind (5%) and hydro (6%), and supported by 5 GWh of battery storage.

Number of microgrids in the US and the total capacity of microgrids constructed per year.

Light manufacturing and food-processing facilities will also tend to have an average grid connection of around 1MW, across c50,000 such facilities around the United States, which are aggregated in our database of electricity consumption by sector.

For larger facilities, we turn to our own economic models, to quantify the typical grid sizes. As usual, facilities with larger capacities will have larger power grid connections.

Size of grid connections will range from 10-30MW for 200kTpa chlor-alkali plants, 1MTpa cement plants, 1MTpa CCS compressors or 100,000 vehicle per year auto plants.

Finally we come to the true monsters, with grid connections above 100MW, such as larger data-centers, aluminium plants and proposed green hydrogen facilities, some reaching 2-3GW in scale.

Size of grid connection and growth trajectory determine whether industrial facilities will realistically need to generate their own power in increasingly bottlenecked power grids.

Renewables plus batteries: co-deployments over time?

More and more renewables plus batteries projects are being developed as grids face bottlenecks? On average, projects in 2022-24 supplemented each MW of renewables capacity with 0.5MW of battery capacity, which in turn offered 3.5 hours of energy storage per MW of battery capacity, for 1.7 MWH of energy storage per MW of renewables.


Co-deployments of renewables and batteries are tracked in this data-file, tabulating the details of over 100 projects that combined a grid-scale battery with their construction of wind and/or solar assets. The average of these projects in 2022-24 added 0.5MW of battery capacity per MW of renewables, with 3.5 hours of energy storage, for 1.7 MWH of energy storage per MW of renewables.

These numbers have all approximately doubled versus a decade ago, when the co-development of renewables plus batteries was a rarity, and tended to occur at smaller scale. This suggests that rising interconnection costs and risks of curtailment are motivating greater deployment of batteries.

A dozen recent renewables plus battery projects are very large in size, ranging from 100-1,000MW of battery storage capacity, almost all being developed in 2020 or thereafter (chart below). For example, the 875MW Edwards & Sanborn solar project in Kern County, California is co-located with 971MW of BESS units from LGChem, Samsung and BYD.

Conversely, the largest batteries from pre-2017 are c30-50MW in size, and many of the technical papers over this timeframe are consciously considering different battery chemistries — lead-acid, sodium-sulphide — rather than today’s projects that are predominantly LFP lithium ion.

The duration of these grid-scale batteries has also increased from 2.6 hours prior to 2020 to 3.5 hours after 2020, with the upper decile projects hacing 5-6 hours of storage (chart below).

It is fine to co-develop renewables with batteries, but it is also more costly. A utility-scale solar project might cost $1,000/kW. A grid-scale battery might cost $1,500/kW. Hence combining 0.5MW of batteries per MW of solar might cost $1,750/kW in total, re-inflating levelized costs of solar by around 50-75%, but still possibly less costly than funding network upgrades.

Our long-term forecasts for power grid capex assume that 0.15MW of grid-scale batteries will be deployed per MW of renewables capacity, comprising a mixture of standalone renewables projects and renewables projects that are co-developed with batteries. And there could be upside?

Companies that stood out in deploying and supplying grid-scale batteries are noted in the data-file.

Copyright: Thunder Said Energy, 2019-2024.