Batteries light the way to renewable energy but the IEA is missing it

Stationary batteries are getting ridiculously cheap. They will make it relatively easy to transition to 100% solar and wind. They will also make the grid more resilient and eliminate blackouts and voltage fluctuations.

Most people currently managing the grid fail to wrap their heads around this. They can’t imagine the accompanying change from a centrally managed top down grid to the decentralized and bottom-up grid that will result from renewables plus batteries. A vivid example given in this post is the underestimation of batteries by the IEA.

Introduction and IEA solar predictions

My entire professional life I’ve made my living predicting technological innovations and then getting to work on them. First as a self employed project leader in the Telecom sector (PCs, TCP/IP, WWW, mobile phones, smart phones) and since my sabbatical fifteen years ago it’s been about renewable energy and electric vehicles at the Eindhoven University of Technology (where I started a research initiative) and with my simulation company Zenmo.

This sabbatical was actually the result of a discovery I made regarding the underestimation by the IEA of the growth in solar panels. That made me want to dive in (even though I ended up writing a book on electric vehicles to start with). The picture below visualizes the discovery that changed my career gave me some notoriety on twitter: the thick black line shows the actual growth of solar and the horizontal colored lines show the assumptions about the future of the IEA.

What made the picture pop is that I translated the numbers of the IEA to what that meant for the solar factories popping up everywhere. I still have the suspicion that the people at the IEA were not accustomed yet to the fact that a solar panel lasts 25 years instead of being burned and lost in an instant like coal, oil and gas. Or maybe it is just the perennial inability of people to understand exponential growth.

The chart shows that the IEA predictions implied that the solar sector would stop growing suddenly after the year in which they made their prediction. And they predicted that year after year after year. Now I’m going to show you the same problem in their first report on batteries.

(Note: the 2023 production of solar panels is estimated at 447 GW. That is ‘off the chart’ times three. And the predictions of the IEA still can’t catch up. Although I also agree with this defense.)

IEA battery predictions

I started with the new 2024 IEA battery report. Although I will complain about the predictions of future growth, it is an excellent resource of information on what is happening on the battery market. If they got the predictions right it would be almost perfect.

Let’s start with the actual historic developments vs the predictions of the IEA.

The graphs shows the historic developments (the dots) and the predictions of the IEA in the “business as usual” scenario STEPS and in the “let’s do everything we can to reach Net Zero Emissions” scenario NZE. The graph on the left has a regular linear vertical axis. There it doesn’t look so strange.

But on the right I used a logarithmic vertical axis and there you can see something strange is going on. The historic developments form an almost perfect straight line (except for a little problem during the COVID crisis). The straight on a logarithmic chart indicates exponential growth. I fitted an exponential growth curve to the historical data and I got a yearly growth 0f 62% that fitted the data almost perfectly. The IEA predicts a sharp break that is impossible to fit. What gives?

My first thought was: the battery market is simply getting saturated. I mean, that is what the chart suggests. But is it?

How much batteries do we need?

I’m only looking at stationary batteries here. (The market for electric vehicles is roughtly 10x as big but I will ignore that for now.) I will also make very rough and simple to follow calculations.

In the NZE scenario the IEA expects a final energy demand of roughly 100 000 TWh per year in 2050. Assuming we get all that from renewable energy this would mean an average production of 10 TW (=terawatt). That’s roughly 1 kW per person by the way.

Scenarios for renewable energy systems often assume around 5 hours of batteries. So that would be 50 TWh (or 5 kWh per person). (A nature article from 2019 by my friend Christian Breyer arrived at 48 TWh so we are definitely in the ballpark here.)

Assuming we replace the batteries every 25 years that means we will end up installing 2 TWh per year after 2050.

So what I did was fit an S-curve (aka logistic function) to the historical data that would arrive at 2 TWh per year after 2050. That way we simulate the usual development of a market that grows exponentially at first but then saturates as it reached full market potential.

I got a curve with a 59% growth rate and midpoint in 2033 that fits the data with 99.8% precision (R²). That’s an almost bloody perfect fit to the historical data that arrives at an endpoint in line with the research. I’ll take it.

Now let’s compare that combination of past developments and future needs to the IEA predictions.

On the left you see the S-curve I determined based on past exponential growth and future demand in a renewable future. It’s racing towards 1000 GWh per year around 2030 and then slows down to reach 2000 GWh per year around 2040.

You also see that the IEA basically throws in the towel before the race to provide the world with enough batteries has even begun. This is not a believable prediction of business as usual (STEPS) or an an all out zero emission scenario (NZE).

I’m sorry, but I cannot come up with a satisfactory explanation for what the IEA predicted here. (Other than scepticism of change.) This is so much below what is needed and below where the historical trend is pointing that it’s basically a data crime.

Will these batteries brake the bank?

We don’t know how the IEA got to its predictions because it’s models are a secret, but in the past the lack of uptake of solar has often been blamed on unrealistic price assumptions.

Might high prices be the problem here? Let’s find out.

First let’s determine a starting price per battery. In 2024 the price of LFP batteries is around $65/kWh on the wholesale pack level (see picture) and since sodium batteries are already being installed and are supposed to be cheaper I’ll conservatively assume they cost the same.

To determine how much they will cost in the future we have to calculate the learning curve. We know that batteries on average become 25% cheaper for every doubling of cumulative production volume (see my previous substack post for details). So how much will production of sodium batteries increase?

I start with the assumption that currently there is about 10 GWh of new sodium batteries. The first large factory is being build in 2024 and is planned to eventually produce 30 GWh per year but that is still one or two years off. So 10 GWh is a rough estimate but that can’t be helped.

I assume we stop learning in 2034 when we have roughly 10 000 GWh of stationary batteries (cumulative, not per year) and I assume that half of them (5 000 GWh) are sodium batteries. Going from 10 to 5 000 is over 9 doublings. Subtracing 25% nine times from a price of $65/kWh would result in a price of $6.5/kWh.

That seems insane but in the previous substack post I already described how sodium batteries can have a ridiculously low material costs of around $1/kWh if you buy the raw materials directly from the mine. So raw materials won’t be the issue.

Let’s round up to $10/kWh.

We want to end up with 50 TWh which is 50 billion kWh at $10/kWh or $500 billion total. That sounds like a lot but we spend over 5 trillion a year on energy so over the period to 2050 this is less than 0.5% of our global energy bill.

Replacements every 25 years would increase the price per kWh with… 0.02 cents. You read that right. So even if my calculations are 5x too optimistic we are talking about massive amounts of batteries costing less than 0.1 cent per kWh.

One kWh battery per kWp of solar panels

An interesting heuristic I came up with is that we should install 1 kWh of batteries for every kilowattpeak (kWp) of solar panels. On average solar panels will output around 20% of their rated peak power worldwide (see table). A good rule of thumb is that you want a minimum of 5 hours of storage in the future renewable grid. Five times 20% is 100%, hence 1 kWh of batteries for 1 kWp of solar panels.

I think buying batteries with your solar installation will become as normal as buying an inverter. And not that expensive either. Solar panels cost at least $100 per kWp wholesale and an entire system can easily cost $1000 per kWp. Adding a matching battery of $10 or even $100 so you can get more out of your system seems only prudent.

A brand new electricity grid

I’m becoming convinced that we are moving towards an abundance of cheap batteries. You will be able to buy them as standard 4 kWh (25 kg) blocks the size of a sack of sand that are easy to place by one person and can be connected up to any desired size. Grid operators will use 20 foot containers with 5 MWh per container that can be stacked if room is an issue.

First these batteries will make it easy to get 100% of energy from cheap solar and wind.

(Caveat: we still have to combine it with seasonal storage using e.g. hydrogen or other electro-fuels for the last 5% or so, because batteries will always be expensive for storing seasonal fluctuations. But since that’s only 5% of our energy flows it will not break the bank.)

But these batteries will do so much more. They will also make our grids more resilient by making blackouts impossible and by largely eliminating fluctuations in current and voltage. Just the grid reinforcements that we avoid this way will pay for the batteries.

I envision solar homes everywhere that could run on batteries without a grid connection for a few hours (and get more electricity using their electric car every other day when the outage lasts longer). Solar buildings combine to form solar neighborhoods that combine into solar countries. With every step grid costs are minimized and resilience is maximized.

I’ve made a profession out of being one of the first to see new technologies coming but I must be honest: although it seems logical to me and it is where the data points me, I still have to get used to this seismic shift. Fortunately it is a very wholesome seismic shift. So I hope I’m right.