Can Solar Costs Keep Shrinking?

Thank you John Arnold, Casey Handmer, and Austin Vernon for having a look at this article. You can check out Austin’s two articles that inspired mine here and here.

I’ve shared this graph before: 

The price of solar photovoltaic (PV) panels has been shrinking around 12% per year for decades. That decline is because of this:

We’ve been installing more and more solar panels. They’re now the lion’s share of new electricity capacity in countries like the US:

The more solar panels we¹ produce, the cheaper they get. If the trend continues for a few years, the promise is that electricity will be too cheap to meter. Indeed, solar is already the cheapest source of electricity.²

But hold on: According to that graph, the cost of solar electricity increased between 2021 and 2023—by about 30-100%—like all sources of energy. Here’s a zoom in on solar and wind:

Note that the Y axis is truncated at $25. Note that according to this, solar electricity costs have grown over the last 5 years

Part of it was due to supply chains breaking during COVID and the subsequent inflation, but is that all? Or were there other reasons? Can we count on solar electricity shrinking further, or are we facing a future where the promise of cheap energy remains unfulfilled?

That would be catastrophic, because our energy consumption has stagnated now for decades.

This is known as the Henry Adams Curve, and it describes the long-term trend of about 7% annual growth in total energy usage. We diverged from it in around 1970. Source. 

And there is no such thing as a rich country that consumes little energy:

Think about it. If we had stayed on the Adams curve, we would be consuming 2-5x more energy than we do today. For the US, that means that GDP per capita today would not be the current amount of $65k, but $100k-$200k. It is a catastrophe that we don’t have more energy: We should be much richer.

Energy stagnation makes us poor. If we want to be richer, we must produce more energy, and the best way to achieve that is if energy costs nothing.

So what might prevent the cost of solar energy from shrinking further? 

Solar panels matter less now.

The following graph looks at the costs of installing solar panels in a residence, comparing the cost of the hardware (“unit costs”) in teal, and in red the “soft costs”, which traditionally include permitting, inspections, interconnection fees, transmission lines, sales tax, operations, and the profits of the installer.

Reader John Arnold brought this to my attention. I believe it’s for the US.

Today, if you want solar electricity in your home, the cost of the actual solar panels is minimal. Over 60% of the costs are soft! It’s harder to shrink those and all the other costs outside of the panels. For example, as reader John Arnold highlights, land and labor are inflationary. Labor and engineering costs grew by 23% and 22% in 2024. Interest rates have also raised costs significantly, as well as connecting to the grid and the fact that so many other solar plants are producing electricity at the same time, congesting the electric network. Other sources of cost increases include interconnection delays, permitting hurdles, lack of transmission, best sites already developed, tariffs on solar panels…

If we want to know whether solar electricity costs will continue shrinking in the future, we need to break down its costs and figure out to what extent each one can shrink.

Breaking Down the Cost of Solar Energy

First, we need to narrow our focus. There are three main types of solar installations: residential (your home), commercial (businesses), and utility-level (companies whose main purpose is producing and selling electricity). 

A big reason for the 2023 jump in utility-level installations is Biden’s Inflation Reduction Act. There was a solar panel tariff waiver that expired in June 2024, and utilities must have installed their new capacity within 6 months of buying them. That’s why the 2024 volume of new utility solar is similar to 2023’s, but might decrease somewhat in 2025 before growing again, thanks to the natural evolution of the industry. Source.

As you can see, the vast majority of installed capacity is utility-level, and that will probably be even more true in the future, now that solar electricity is so cheap and there’s big money to be made with it. So we need to pay attention to utility-level solar costs.

OK, so what’s the cost of a utility-level solar farm, broken down by types of costs?

Source. The proportions are similar in this other site.

Conveniently, today every watt of solar capacity in a farm costs about $1, so we can simplify and use percentages of costs (like 39% for PV modules) interchangeably with straight up cents (like $0.39).

Solar Panels

The PV (photovoltaic) module is still 39% of the cost (about $0.39 per watt). That will go down fast—probably faster than the historic 12% annual drop because the investments are so much higher now. The world average is already around $0.25/watt. In some places in the US, solar panels already cost $0.20/watt. Overseas, it can go as low as $0.10-$0.12/watt. 

Is there a floor for these costs?

Solar manufacturers are investing hundreds of billions in expanded capacity in an all out war for market share against a background of panel price drops of 15-25% per year. There is an extreme economic forcing function towards rapid improvement and ultimately convergence with the Platonically ideal solar panel - some 20 um thick layer of silicon supported by a 100 um thick layer of plastic rolled off a spool - or some other tech that's thinner and cheaper than paper.—Casey Handmer

Thinner and cheaper than paper. Think about that!

Solar panels are basically sand atoms reconfigured with intelligence. Sand is everywhere. It’s cheap. We can bring the cost of PV panels down radically. What will the world be like when panels cost $0.05/watt? $0.01/watt?

And here’s the key insight of this article: For the longest time, PV panels were so expensive that no other cost mattered. All the focus was on improving the efficiency of the panels, other costs be damned. We could waste everything that was not the PV panels and the economics barely changed.

Cheap solar panels are a new paradigm

That is not the world we live in today. For the first time, solar farm designers must change their mindset completely. They must stop thinking about optimizing for panels, and instead start focusing on trimming the other costs. How will that change them?³

Building the Farm

To build a solar farm, you need all the stuff I put in brown on the graph above:

• Design and engineering to figure out how you’ll lay out your farm (~3% of costs)

• Civil works to adapt the site to the trucks, machinery, and solar panels (~8% of costs)

• Balance of System (BOS) for all the other material we need to build the solar farm (~17% of costs)

• Logistics to bring all the material to the farm (3% of costs)

• Direct Labor to install the PVs and the BOS costs (~12% of total costs). 

Together, these items make up about 43% of solar installation costs today—more than the panels! 

To begin a solar farm, you need design and engineering, to adapt the panels to the land.

Solar farm designer designing. Maybe costs will go down when they start using computers for their designs? Source of the image. Yes, I’m joking. 

Then you lay the ground for the site. For example, add gravel so that the trucks that bring the material don’t get stuck in mud or trenches (dug to hide the cables). That’s the civil work.

Civil costs (or preparing the ground for the solar farm) are unlikely to go down by much. Source.

Once your site is ready, you start installing the panels. Structural Balance of System (SBOS) refers to the cost of the infrastructure that will hold the panels. 

First you need to pound the pilings into the ground.

The source for all these images is the same, so just go watch the video if you want to see it in action.

Then, if you want your solar panels to track the Sun to maximize power, you’ll need trackers that pivot the panels through the day.⁴

These pilings and trackers will then hold the racks where panels sit. 

You can also see the direct labor in action throughout these pictures

Once you have the racks with the panels, you need to wire it all up. That’s what Electric Balance of System (EBOS) means: the cabling, switches, electronic control systems, battery management system etc.

Here are some cables collecting the energy from the panels, with a gentleman explaining how miraculous all of this is. Source.

Gotta pay for these boxes here, among other things. Source.

Will these costs go down?

On one side, these are hard costs that have been around for a long time and are difficult to shift. Labor cost usually increases, while steel, aluminium, wires, and the like are unlikely to go down.

But remember: They’ve never been optimized! Our paradigm was: “Do whatever you need to get the most out of every solar panel!” If we change that paradigm and now the panels are dirt cheap, how would we completely rethink the installation?

Here’s one example: Historically, solar panel makers have reduced the cost of solar energy by making their panels collect sunlight more efficiently. Now that installation is such a big share of costs, they have started redesigning their panels to reduce these BOS costs. One example of recent redesign is bigger panels: The bigger each panel is, the less rack and wiring it needs.

Or think about the entire structure: trackers to pivot the panels to follow the Sun, racks high up to allow room for that movement and to protect the panels from ground-level hazards, strong pilings deep in the ground to hold all this weight… 

But if your panels are cheap, you can basically dump them on the ground! No pilings, no racks, no trackers. Barely any SBOS! And since you don’t have heavy materials to transport, your logistics costs drop and you barely need any civil engineering: No need for big trucks to enter the farm, no need for infinite trenches to hide cables underground.

And this is not a pie in the sky. Some companies are doing it already, like Erthos:

The company doesn’t share much detail, but it appears that the panels are on some kind of rail system. Source: Erthos.

According to the company making these panels, it can reduce costs of CAPEX (capital expenditure; fixed upfront costs) by 20% thanks to:

1. Tighter spacing that reduces land usage by about 50%.

2. The simple layout that reduces site preparation, planning, and design expenses.

3. 70% less trenching and wiring.

4. No racks or trackers.

5. 50% reduction in construction time (reduces civil engineering costs).

6. Farms using modules with glass on both sides without the expense of beefier racking. Double-glass panels are more reliable and have longer lifetimes.

It also means lower operation costs: little mowing because plants barely grow underneath, easier to clean, less maintenance because of fewer moving parts like trackers.

Just to show that this is not a one-off empty promise, here is Jurchen Technology’s PEG system:

The panels are about 1 meter above ground, but crucially they have no moving parts and are laid on the basic steel rebar you can see. Here’s how it’s installed:

If you watch their video, you’ll see it just takes one person to install this rebar, and a much smaller machine, for a way faster installation than traditional panels.

No need for trenches, gravel paths, concrete, heavy machines. Since the panels are one meter high, workers don’t need to go up and down ladders.

The slight angle of the panels means that they catch a bit less light in peak hours. But who cares? That’s when all the solar farms are producing at maximum intensity. Instead, this angle catches slightly more energy in the morning and evening, which are peak demand times—electricity is more expensive and there’s less production of solar energy from competing farms.

Since it’s so easy to install, you don’t need skilled labor. Unskilled workers can easily be trained to do the job.

Maintenance is also quite low:

Panels so tightly packed means little light, and hence not much vegetation below. The little vegetation that does grow can be cut with robots. Or you just dump a plastic sheet on the ground before installing the panels and forget about plant growth altogether.

Since all the panels are packed, it’s easy for a robot to clean them with little human interaction.

This uses 75% less steel, 90% in machine cost, 50% less in logistics, for an overall reduction in CAPEX of 40% and OPEX (the operation of the farm) of 20%.

Of course, with such a straightforward installation, you don’t need much civil work or design and engineering.

This is just one example of the types of changes we can make to reduce costs of design, logistics, BOS, direct labor, and even civil engineering. We’re already at a point where we can halve these costs of the solar farm (as a reminder, ~43% of the total). We will probably be able to reduce these costs further in the coming years.

PV Inverter

Once all these solar panels are installed, they generate a constant stream of electricity, which is called direct current or DC. But the electric grid uses electricity going back and forth quickly, called alternating current or AC.⁵ The inverter transforms electricity from DC to AC.

This cost has been shrinking and will continue to shrink:

Cost of PV inverters over time. Notice that this is a logarithmic curve. Source.

Other Costs

I assume that taxes, permits, interconnection, and inspection costs won’t shrink significantly. But they account for just 5% of costs.

Meanwhile, we can expect the overheads and margins to shrink as companies become experts in managing solar farms, costs decrease (and hence risks), and competition erodes these margins. Since these account for $0.08 of our panel costs today, we can safely assume they will at least halve.

There are yet other costs of solar farms that don’t appear on the graph I showed early on.

Interest on Capital

A lot of the costs we’re discussing are paid upfront, and then the solar farm makes money over time. This means the cost of capital is important. But if the cost of building a solar farm shrinks, so does the upfront cost, and hence the risk and the cost of capital. 

The little remaining capital has to pay interest according to official rates. The lower the interest rates, the more money these farms will make. 

Today, solar panel interest rates are not too high, so capital costs are not immense.⁶ But we can assume that, when interest rates fall, solar farms will be a bit cheaper, and investment in them will increase.

Efficiency

A lot of the things I’ve mentioned focus on reducing the costs, but that’s not the only lever we have. We want a very low cost per watt, so we can either lower costs, or increase watts. 

So far, most of the efficiency gains were focused on transforming more photons from the Sun into electricity. 

From 2010 to 2020, average panel efficiency increased from ~10% to ~20%. Output from a panel has doubled just from efficiency gains. Theoretical efficiency for current panel designs is limited. Silicon, which the vast majority of today's PV cells use, is limited at 32%. Adding multiple layers optimized for different wavelengths can theoretically raise efficiency up to 68%. Using technology other than today's p-n junction cells could reach theoretical efficiencies of 90%+. It would be a significant achievement to get mass-produced efficiencies up to 25% for single layer and 40% for multi-layer p-n junction panels. Near-term efficiency will continue to increase from increasing silicon purity and the emergence of lower-cost multi-layer panels, among hundreds of other small advances.—Solar PV's Path to Dominance, Austin Vernon

We can increase efficiency by improving the core photovoltaic panel, but there are also other ways.

For example, you can have two-sided panels, which will not just capture solar energy from the Sun, but also from the light that bounces around in the environment.

Double-sided panels. Of course, this setup is likely incompatible with the one in which panels would be on the ground, but you can put it on fences though. Source.

Other examples include half-cut cells, shingled cells, back-contact cells…

We can probably double our efficiency in the coming years. It will take a breakthrough to increase efficiency beyond 40%, though.

Land

From what I can tell, land costs are usually not included in these cost calculations because they’re so low and vary a lot depending on the location. But most of the sources⁷ I have looked at quote costs between 2% and 20% of the overall installation.

Land is not deflationary though. It’s the one thing we can’t make more of, and yet there are more and more humans on Earth. Fixed supply and increasing demand means growing price. So the cost of solar farms will converge towards the cost of land. 

On top of that, the best sites are already taken. Future ones might need to be more expensive or farther out—which increases costs in other ways.

On the other side, in most places, land is plentiful and cheap. And the same force as the one we’ve shared before will come into play: As solar farm costs shrink, land will become a bigger share of the cost, so solar farm operators will have a much stronger incentive to seek the cheapest land. These places will get plenty of solar panels, thus keeping overall solar prices low.

How Fast Will These Improvements Come?

One of the big advantages of solar photovoltaic is how fast its cycles are. It doesn’t take long to set up a solar farm, and it’s cheap to build, so the learning loops are very fast, and hence cost reductions tend to be achieved much faster than in other industries, like nuclear. 

How Much Will Solar Prices Evolve Then?

If you add up all these costs, you can expect in a few years:

• Solar panel costs that shrink from $0.39 to $0.05 per watt_DC.

• Buildup costs (design, engineering, civil engineering, balance of system, direct labor, and inverter) going from about $0.47/watt_DC to ~$0.10/watt_DC. Why? Today, these costs are already halving. I think it’s reasonable to expect this cost to halve again.

• Taxes, permits and the like will likely remain fixed (about $0.05 today), but margins will likely compress in absolute terms.⁸

Just with these measures, we will witness a drop in solar costs from ~$1 to ~$0.25, for a 4x cost drop.

Then you have the falling cost of capital (smaller loans, less risk, faster installation, probable lower interest rates in the future), and cheaper land that will likely be pursued. Overall, let’s assume these all shrink by 4x too.⁹

Then we have the increases in efficiency, which might get us to a doubling of energy production, which would halve again our cost per watt. Further reductions will be difficult to realize unless there’s a breakthrough in PV panel technology.

If we put all this together and my assumptions are reasonable, it means we should see a reduction of solar costs of about 8x, after which gains will slow down.¹⁰

In other words: We have a bright future before us.

But This Won’t Directly Make Our Electricity Cheap!

Everything I’ve told you so far talks mostly about the DC electricity that solar panels generate while the Sun shines. But there’s a few more facts that change completely the conclusion, and tease up how the world of energy will evolve in the coming decade: