Predicting Growth in the Battery Industry

Second-guessing the IEA for fun and profit

In a global economy that ran on “clean” energy — that is, renewable, non-carbon-emitting sources like solar, wind, hydroelectric, and nuclear — we’d need a lot more batteries than we have today.

We’d need stationary batteries to store solar and wind energy (whose production fluctuates with the weather and time of day) and we’d need portable batteries to power electric vehicles.

How much more battery capacity would we need, exactly?

Institutions like the International Energy Agency (IEA) answer these kinds of forecasting questions, but they are reliably way off in their estimates, vastly underestimating the growth of solar power.

So can we do better in a blog post?

We can try!

One reason it’s not a completely insane thing to do is that I’m going to be making much simpler and rougher assumptions than the IEA would, things like “assume this multi-decade exponential curve continues for another decade.”

If we assume that solar and wind power keep growing as they have been, demand for electric vehicles keeps growing as it has been, etc — no hypotheticals about “what if” policy changes are enacted or new technologies discovered, just straight extrapolation — what does the world look like in 2030?

In particular, how big will the battery industry be?

Let’s break this down into sub-questions.

How much energy and electricity does the world use? How fast is it growing?

As of 2019, the global electricity consumption was 23,845 terawatt-hours.  

Since 1965, global energy consumption has been growing roughly linearly, at slightly over 2000 terawatt-hours per year; and global electricity consumption since 1985 has been growing at roughly 430 terawatt-hours per year.

The IEA’s 2019 world energy outlook report predicts that electricity consumption will grow 2.1% until 2040 under the business-as-usual “Stated Policies Scenario,” which would put electricity consumption at about 34,000 terawatt-hours in 2030 and 37,000 terawatt-hours in 2040.  

Under the IEA’s greener “Sustainable Development Scenario”, which includes both increased use of renewables and decreased global energy consumption overall, predicts a slightly lower 32,000 terawatt-hours in 2030 and 35,000 terawatt-hours in 2040.

If we naively extrapolated long-term trends, instead of using the IEA’s estimate, we would expect global energy consumption to be roughly 180,000 terawatt-hours in 2030, and global electricity consumption to be roughly 30,000 terawatt-hours in 2030 -- somewhat lower than the IEA’s predictions.

What is the current breakdown of energy sources? How is it shifting?

Global energy consumption in 2018 was only 4.7% derived from “renewable energy”, and 15.5% from “renewables” plus nuclear and hydroelectric, with the bulk coming instead from coal, oil, and natural gas.

Global electricity consumption in 2018 was 11.3% solar + wind + other renewables, and 37.7% “renewables” plus nuclear & hydroelectric.

BP forecasts that by 2050, global energy consumption will shift to 22% “renewables” and 33.5% “renewables” plus nuclear and hydroelectric.

Under the IEA’s Stated Policies Scenario, global electricity production is projected to be 28.6% “renewables” and 50% “renewables” plus nuclear & hydroelectric by 2040.

What is the trend in solar power generation? What happens if that’s extrapolated into the future?

Between 2010 and 2020, global solar power has grown from 30.52 terawatt-hours per year to 844.39 terawatt-hours per year.

Here’s how global solar power looks on a log scale:

Since 1990, this has been roughly linear, though it was a bit slower in the 90’s and a bit faster in the 2000’s. 

Modeling solar power as an exponential curve, it’s roughly growing at 29% per year since 1990.

If you naively extrapolated that into the future, that would mean 10,775 terawatt-hours would come from solar power in 2030 -- more than triple what the IEA predicts in the Stated Policies Scenario.

This means that, in 2030, we’d expect roughly a third of global electricity and 6% of global power to come from solar.

What is the trend in wind power generation? What happens if that’s extrapolated into the future?

In 2020, wind power produced 1590.19 terawatt-hours, a bit higher than the world’s production of solar power in the same year.

Again, now let’s look at the log scale:

Wind power has been growing roughly exponentially since 1989, at an average rate of roughly 22% per year, somewhat behind solar’s growth rate.

If we naively assume this will continue until 2030, that would predict 7804 terawatt-hours produced from wind power in 2030.

Together, if we extrapolated current exponential trends eight years in the future, wind and solar would be over half of all electricity generated in 2030.

How much battery capacity exists today?

IRENA, the International Renewable Energy Agency, estimates there are 4.67 terawatt-hours of energy storage available globally.

96% of this is pumped hydro storage, in which two reservoirs of water at different heights store potential energy. The potential energy can be released by allowing water to flow from the higher reservoir to the lower one. 

But, of course, pumped hydro storage is not usable for mobile energy storage needs (like electric cars). And it is difficult to find suitable sites, so it may not be usable for all solar-based grids.

For a lot of applications of variable renewable energy (like solar and wind), we’ll need batteries.

Official projections of growth in battery capacity

IRENA estimates that the total amount of stationary battery-based energy storage worldwide would have to grow by a factor of at least 17 by 2030 if the fraction of electricity generated by variable renewables (solar and wind) is to double by 2030.

The IEA’s Net Zero Scenario, setting a goal of zero carbon emissions by 2050, requires 585 Gwh of stationary battery capacity worldwide by 2030, a 35x increase relative to 2020’s 17 Gwh.

Bloomberg NEF has a much higher estimate, predicting 30x increases in the total stationary battery-based energy storage worldwide by 2030, in the baseline scenario. 

That’s just stationary battery capacity. What about electric vehicle batteries?

As of 2020, electric vehicle batteries constituted 110 GWh worldwide.

The IEA predicts that by 2030, in the Stated Policies Scenario, that will rise to 1700 GWh, or a 15x increase. To meet the Sustainable Development Scenario by 2030, EV battery storage will need to rise to 3200 GWh, or a 29x increase. And to achieve Net Zero by 2050, EV battery storage will need to rise to 6500 GWh, or a 59x increase, by 2030, and further rise to 14,000 GWh, or a 127x increase, by 2050.

How much battery capacity is needed to complement solar grid power?

A 2019 paper from the Trancik Lab at MIT found that the amount of energy storage necessary depended greatly on the type of power plant (baseload or peaker), and the location (Arizona, Iowa, Massachusetts, or Texas).

A baseload solar power plant (which, as far as I know, does not yet exist in the US) needs to be combined with enough energy storage that together they produce a consistent power that can always meet the minimum (or baseload) demand. A peaker power plant only generates power when there is unusually high demand.

Baseload solar plants in sunnier areas (Arizona/Texas) need about 6 kilowatts of storage per kilowatt of output; baseload solar plants in less sunny areas (Massachusetts/Iowa) need 3 kilowatts of storage per kilowatt of output. Peaker plants need about one kilowatt of storage per kilowatt of output.

Both types of plants need, roughly, 40 hours of storage in sunnier areas and 80 hours of storage  in less-sunny areas.

Peaker plants are rarer than baseload plants -- roughly 1 peaker plant for every 10 baseload plants in the US, for instance.

Extrapolating, very roughly, for every terawatt-hour per year generated in solar energy, we’d need about 25 gigawatt-hours of energy storage, if solar energy is deployed in a mix of 90% baseload, 10% peaker.

Using our extrapolated-exponential prediction of 10,775 terawatt-hours/year generated globally from solar in 2030, this comes to roughly 27 terawatt-hours of energy storage required.

At the other extreme, if all solar power goes into peaker plants, we’ll need about 4.6 gigawatt-hours of energy storage for every terawatt-hour per year, which comes to a predicted 5.5 terawatt-hours of energy storage required in 2030. 

Now, these are overestimates. Not all the energy storage in a solar grid needs to be met by batteries -- some unknown proportion could be met by pumped hydro storage, or other storage mechanisms. 

If stationary power storage remains 96% pumped hydro and only 4% batteries, that comes to a demand of 220-1080 Gwh from battery power in 2030, depending on the peaker/baseload ratio in new solar power plants.  

The upper end of this range -- which, to be clear, is based on extrapolation of current trends, aka a “business as usual” projection -- is still higher than the IEA’s ultra-ambitious Net Zero Scenario targets for stationary battery capacity. 

How much battery capacity is needed to complement wind grid power?

Again referring to the Trancik paper, we find that baseload wind plants need 1-2 kilowatts of energy storage per kilowatt of output, depending on plant location Peaker wind plants need 1 kilowatt of energy storage per kilowatt of output.

Baseload wind plants need 80-180 hours of battery storage, depending on plant location. Peaker wind plants need 30-70 hours of battery storage, again, depending on plant location.

This means, for every terawatt-hour generated in wind power, baseload wind plants need 9-27 gigawatt-hours of energy storage, and peaker wind plants need 3-8 gigawatt-hours of energy storage.

Conservatively, if most wind plants are peaker plants, that’s roughly 23-70 gigawatt-hours of energy storage needed for wind power by 2030.

If, as before, we assume that no more than 4% of this stationary energy storage takes the form of batteries, that’s 1-3 gigawatt-hours of battery storage needed in 2030.

In other words, battery storage needs for wind power are practically a rounding error compared to battery storage needs for solar power.

How many electric cars are sold per year? How fast are sales growing?

In 2020, there were 6,850,330 battery electric vehicles in use worldwide. This number has been growing exponentially at an average of 55% per year over the past half decade.

Electric vehicle sales as a fraction of all new car sales have also been growing at over 50% per year:

If we naively continued this trend through 2030, there would be nearly 550 million battery electric vehicles worldwide by 2030, reaching roughly a third of all vehicles in use worldwide.  By 2030, all new light vehicle sales would be electric.

Morgan Stanley’s projections are more conservative, expecting that only a quarter of new vehicle sales will be electric by 2030.

Under a more conservative assumption where electric vehicle sales grow only at 20% per year, in line with Morgan Stanley’s projections, we’d see something more like 40 million electric vehicles in use globally by 2030, an order of magnitude less than we’d get from naively extrapolating from the 2015-2020 data.

The IEA’s projections are intermediate, predicting 89 million battery electric vehicles worldwide in 2030 under the Stated Policies Scenario and 166 million battery electric vehicles under the Sustainable Development Scenario.

How much EV battery capacity will be required by 2030?

Electric vehicles today contain battery packs ranging from 28.9 kilowatt-hours to 200 kilowatt-hours.  The larger the battery capacity, the more driving range is possible; compact cars designed for city driving have less battery capacity than SUVs. 

The most popular electric car of 2021, the Tesla Model Y, has a battery capacity of 75 kWh.

Using 75 kWh as a “typical” electric car’s battery capacity, and assuming it remains unchanged, this means world EV battery capacity in 2030 will range from 3-41 terawatt-hours, depending on how fast EV sales grow.

The IEA’s projections for global EV battery capacity are below this range, 1.5 terawatt-hours in 2030 in the Stated Policies Scenario and 3 terawatt-hours by 2030 in the Sustainable Development Scenario. 

Personally, I think that the lower estimates of projected EV sales are more credible. Naively extrapolating exponential trends based on 5 years of data is sketchy (compared to extrapolating trends from the 30+ years of data we have for solar power generation). 

Using assumptions of EV sales growth more in line with the IEA’s and Morgan Stanley’s projections, I’d give a range of 3-7 terawatt-hours of global EV battery capacity in 2030. 

This means that global EV battery capacity will have to grow 27x-64x over this decade to meet EV demand.

What is the current size of the battery market? What is its growth rate?

The global lithium-ion battery market was $40.5B in 2020, and is projected to grow by about 14% per year by Research and Markets.

The growth rate of lithium-ion battery sales over the past few years is a bit higher, more like 20%.

McKinsey ‘s forecast matches the recent trend, predicting the lithium-ion battery will grow over 20% per year until 2030.

In 2020, $23B of the lithium-ion battery market was in stationary energy storage, while $22B was in electric vehicle batteries.  

More advanced battery chemistries have much smaller market sizes; the solid-state battery market was $84M in 2021, while the lithium-sulfur battery market was under $700M in 2019.

Lithium-ion battery revenues are growing quite a bit slower than my predictions for battery capacity demand, which will be around 30% annually for stationary energy storage and 40-50% annually for electric vehicles.  

This can probably be explained by the falling cost of lithium-ion batteries, which have dropped by more than 97% since 1991.

If battery energy storage demand grows at 40% per year but energy storage costs fall by 13% per year, then battery revenues only grow at around 21% per year. Which is pretty much what we see in the recent past.

For comparison’s sake, a sector growing revenues roughly 20% a year is doing great, comparable to the leading tech giants like Google and Apple. 

Conclusions

I’m estimating 3-7 terawatt-hours of electric-vehicle battery capacity will exist in 2030, and 0.2-1.1 terawatt-hours of stationary utility-scale battery capacity in 2030.

That basically means 30-50% yearly growth in global battery capacity over the next decade.

“Battery” here is understood broadly; none of my assumptions require them to be lithium-ion batteries. They could be other battery chemistries, or thermal energy storage, or compressed air, or anything else.

Even with exponential drops in the cost of lithium ion batteries, that still puts the battery industry as one of the fastest growing markets anywhere.