Large-scale vs. small-scale solar: Why residential rooftop solar may be threatening the energy transition

Before diving into this argument, let's clarify four things:

Building fast - Where are we today?

Germany had a record year for PV installations of 15 GW in 2023 with a year-end capacity of 81 GW solar and 2024 looks on track to become a close second place for new installations.

Small scale rooftop solar (i.e. plants <100kWp), which is under the governments feed-in tariff and therefore has neither the technical capability nor the economic incentive to react to market prices, accounts for roughly 2/3 of added capacity, meaning that around 10 GW of rooftop solar were added in 2023. Germany and the Netherlands are special in this segment. Most countries, Spain for instance in Europe but most importantly China, are opting for large scale installations while in our country we have heavily subsidized small-scale installations and the problems are starting to show.

Today, of the roughly 81 GW PV, one third – or 24 GW – are under the Direct Marketing (“Direktvermarktung”) scheme where PV power is actively traded on spot markets and 57 GW are small-scale assets that are not actively traded because they are under a fixed feed-in-tariff and have no incentive (or ability) to react to the grid’s or market’s signals. So far the facts.

First of all, this is evidence of the energy transitions success story and to this point most kWp installed were great. So, what is our beef with residential rooftop solar?

Fundamentally there are two problems:

• The most important one from our perspective is their ridiculously high relative cost and (maybe the one more relevant from a policy point of view)
• Their uncontrollable feed-in is threatening the stability of the energy system.

Let us start with costs

It is not 100% straightforward to get firm prices for small-scale rooftop solar, but after checking some sources (we got an offer from Germany’s best known PV/storage installation startups) we find that rooftop solar with good quality cells has all-in costs (including construction, inverter, mounting system etc.) of around 1.500 EUR/kWp or 15.000 EUR for a 10 kWp rooftop asset in Germany.

For comparison, a large scale 10 MW PV plant has turnkey costs of around 500 EUR/kWp (according to many of our customers and publicly available information), so they are about one third of the cost of rooftop solar. This is a HUGE difference. The reason for this difference is that module prices have decreased so fast that the non-module cost, such as inverters, mounting systems, electrical wiring, installation and commissioning etc. are becoming relatively more important. Sending a construction worker onto a single roof is simply a very costly way of installing a module that is worth increasingly less.

With costs being volatile and difficult to calculate, maybe we over- or underestimate this difference a bit, but it should not play a role for the overall point. Even a relative cost difference of 20% would be high at the current scale of investment, but we are talking about a 300% difference!

So does this matter in scale? The answer is, YES, BIGTIME. If we had installed the 10 GW we used for rooftop as brownfield solar plants (utility-scale), we would have saved the cost difference of these 10 GW, i.e. 10,000,000 kWp * 1.000 EUR = 10 billion EUR per year (or 7 billion, if we use the numbers from the best case scenario, see editor's note above). That is the annual additional cost of mounting solar on rooftops. Two years of this could finance Germany’s entire planned hydrogen grid, or many other projects.

The German economy is struggling with relatively high energy costs and a recession in the middle of an ongoing war in Europe that is putting additional pressure on the economy and on prices. And Germany is building tiny renewable assets on scale that are uncompetitive by a factor of 3! This kind of cost is a real threat to the economic viability of the energy transition and no country will be competitive in terms of energy costs if it chooses to waste money on an inefficient supply structure in scale.

Why would anyone invest in rooftop solar, if it is so expensive?

The simple answer is that it is heavily subsidized, directly but mostly indirectly. Rooftop solar that partially feeds the power into the grid and partially gets self-consumed has a fixed feed-in tariff of around 8 ct./kWh or 80 EUR/MWh. This is higher than the roughly 60 EUR/MWh paid to large-scale solar grid installations, but it is not actually the problem. No rooftop solar plant would be installed if it had to rely solely on this fee. Their business model is self-consumption and tax savings.

When I am consuming my own solar power instead of consuming power from the grid, I save roughly 170 EUR/MWh or 17 ct./kWh in grid charges, electricity tax and other surcharges that I would otherwise have to pay. If we assume that our 57 GW of rooftop solar produce 57 TWh/year and 50% of this production is self-consumed (i.e. 28.5 TWh at 170 EUR/MWh) then we have unpaid fees of 4.84 billion EUR per year.

This is great news for relatively wealthy homeowners not having to pay for these fees, but kid you not, someone is paying these costs and that is mostly all of us that don’t own our very own rooftop solar installation, and we pay for it through higher grid charges and other surcharges. Now these unpaid costs tend to massively increase for newer plants as 70% of new installations feature a home storage system that increases the amount of self-consumed power and therefore the amount of unpaid fees and surcharges increases. The collection base for these fees further shrinks and everyone who cannot make this investment pays more. This is a pure transfer from the relatively less wealthy to those who own a rooftop and can make such investments.

Segway: Home storage systems face essentially the same problems as rooftop solar, however we focus on one problem at a time.

But we are not finished: Since 2023 there is no VAT tax (usually 19%) for rooftop solar under 30 kWp (another 2 billion EUR subsidy if we assume 8 GW of new installation falls under this category).  In addition, the income for the grid-exported energy is exempt from income tax or trade tax, which is also curious.

Furthermore, small-scale PV often benefits from cheap and subsidized KfW credit programs providing lower-than-market rate interest and PV can benefit from special depreciation methods that provide a further tax benefit. However, going into detail here would require a much longer article.

Putting all this together, the simple fact is that rooftop solar is extremely expensive relative to grid-scale installations and the main reason why these assets are installed is a bouquet of subsidies paid by the general public.

But one can ask: is this subsidy justified?

This may be the case if rooftops were the best places to put solar assets because we simply do not have enough space, and/or if the (indirect) subsidies paid in terms of uncollected grid fees and surcharges were justified because these rooftop assets massively alleviated grid costs.

Now while using already sealed surfaces on rooftops may seem intuitively clever, we simply have much less of a land scarcity issue than many may think. You can easily buy land for PV in Germany at a cost of less than 3 EUR per square meter.

Let’s say I use the 10 billion EUR wasted on rooftop PV in a single year to procure land. I could purchase 333,333 hectares of land, which I could use to build 333 GW of PV (twice the total amount installed in all of Germany today). According to the Umweltbundesamt we only need an additional 63,000 hectares to reach the PV installations goals or in total 0,3% of Germany’s land area. In comparison 1,410,000 hectares of land or 4% of German land area is currently used for the energetically much less effective biogas production, so finding 0.3% of German landmass is a relatively minor problem. Again: land is not the problem!

What about the grid?

The second argument we often hear to justify the heavy subsidies for rooftop solar is the also intuitive sounding point that self-consumption reduces grid costs. This is not only false, but the opposite is the case.

I doubt that of the 4 million rooftop solar households there are more than 100 that are completely self-sufficient and can do without a grid connection. With grids being CAPEX heavy, it does not really matter to grid costs whether you use the grid 24 hours a day or only 12 hours a day. The costs are the same. Therefore, using the grid less due to partial own consumption doesn’t really decrease costs in any fundamental way. The argument that own-consumption fundamentally reduces grid costs is about as convincing as arguing that road construction costs decrease substantially if some people only use roads half of the time. The expensive part in CAPEX investment is the first-time investment, not the upkeep. Which brings us to the more important point namely that uncontrollable rooftop solar continually tends to threaten grid stability.

Rooftop Solar threatening the grid

Another false argument in favor of rooftop solar is that it alleviates the grid during peak demand. We tend to have maximum demand in evening hours from 19-20 in December, and as you might have guessed: the sun isn’t shining at those times. Rather we tend to have ever more situations in which too much uncontrollable rooftop solar is threatening the grid.

The power market is structured in a way that power plants only produce if they make money by selling their power on the so-called spot markets, such as the Day Ahead or Intraday market. This means that prices have to be higher than their variable costs for production to be worthwhile. So, when there is overproduction, prices should drop to a level to decrease supply until it meets demand. If this does not happen, we are running into real trouble.

Renewables don’t really have variable costs, so they tend to produce until a price of 0 EUR/MWh and those with subsidies to a level of maybe until -100 EUR/MWh (link article). After this they should shut down production. So far the theory and at FlexPower very much our everyday practice with larger-scale assets. So, in theory, everyone shuts down production and the market regulates itself, but there are a few issues.

Inflexible supply

We have around 8 GW of conventional must-run conventional capacity in the market that is providing power as a by-product or is providing various grid services or is simply too inflexible so it will not shut down even if prices turn extremely negative. 8 GW was the lowest Day-Time conventional production observed at extremely negative price on May 12^th 2024, hence the number.

Additionally, we have some share of grid scale renewable installations that for reasons we cannot really understand do not respond to price signals, which we see running even during extremely negative prices.

And most importantly we have a fleet of rooftop solar plants that have already been built that have neither the technical ability nor the economic incentive to shut down power production when they are crowding the grid. These assets cannot and would not respond to an overcrowded grid.

Let’s look at an example

On days with low demand such as Easter Sunday, demand during solar hours can go as low as 40 GW say between 1-2 pm. Now let’s say it is a very sunny day and of our 57 GW of rooftop solar 60% is producing (which would be a lot) i.e. they would be producing 34.2 GW. We also have the 8 GW conventional must-run capacity.

In addition, let us say we have another 14 GW of grid-scale solar production and 20 GW of wind production and 5 GW of Biomass production (i.e. 39 GW of grid scale renewables). Now these renewables would, if prices drop low enough, react to this signal and curtail (shut off their production) but not all of them do so for various reasons. However, let’s be optimistic and say a record 70% of this capacity is curtailed leaving 11.7 GW of grid scale production.

Now we have:

Demand = 40 GW

Supply = 8 GW conventional must run + 34.2 GW uncontrolled rooftop solar + 11.7 GW of grid scale renewables

= 53.9 GW

Let’s be optimistic and say we can export 8 GW to our neighbors which is very high, because they also have low demand and lots of renewables on Easter Sunday, but OK. Now we have excess supply of:

53.9 GW- 8 GW export – 40 GW demand = 5,9 GW excess.

What happens now? We have an excess capacity of the equivalent of about 5 nuclear power stations running at full force and crowding the grid. The last market resort is 3 GW negative balancing reserves kept by the TSOs, which they can regulate down. So after this last emergency measure we are left with 2,9 GW excess supply.

And now? The honest answer is nobody really knows. The intraday market would fail to clear, and prices would be stuck at -9.999 EUR/MWh. Grid frequency could rise to threatening levels of 50.2 Hertz where solar inverters would shut down, however machinery would already start getting damaged and local grid providers may have to take whole areas off the grid (local brown outs) in order to protect the system. I think my home area of solar-rich southern Bavaria would be the first to go black.

You may think that we are painting horror scenarios here but none of this is exaggerated or even far away. We have seen minimum possible prices of -500 EUR/MWh on Day Ahead and minimum possible prices on intraday of -9.999 EUR/MWh in 2023 and 2024. If this bad-case scenario does not happen this year, it will happen next year or the year after if the rooftop solar expansion continues. It is really not a question of if but when. The German regulator BNetzA has recognized this problem and is also warning of the consequences.

Oh and of course: who pays for selling energy for -9.999 EUR? That would be the transmission grid operators who would recuperate this money by higher grid charges for everyone. Mind you: we as traders usually benefit in these situations as we like price volatility; it is the customer who pays.

So what to do?

The elegant answer is of course: From now on we only install solar plants that react to market prices, we install lots of (hopefully) grid scale storage and we make sure millions of EVs increase demand in these situations to balance the grid. This is all correct, but it is years away. In the status quo most households do not have smart meters, smart meter gateways, and remote-control units for their solar, all of which would be a minimum prerequisite to implement any of these elegant solutions. So, face it: we simply do not have a short-term solution at this time!

Therefore, in the short-term government needs to do one thing, and that is taking away all subsidies that are specific to rooftop solar. This would level the playing field between rooftop and brownfield solar and significantly slow down the buildout of this very expensive (and currently) inflexible technology called residential rooftop PV. Additionally, the following make sense from our point of view:

• Implement variable grid fees of 0 EUR/MWh during peak solar infeed hours decreasing the relative value of the rooftop subsidy and start charging more for grid connection capacity instead of energy consumption.
• Start charging VAT for solar equipment and levy income and trade tax for small scale solar developments and
• Regulate for rooftop solar plants to be remotely dispatchable (like all bigger installations) in order to be able to safely shut them down in times of overproduction. The current EnWG draft goes into that direction, however is not yet forceful enough to mandate active trading on all solar PV assets.

I am well aware that this would likely slow down the residential rooftop solar market significantly. However, we should really start questioning whether it is a good idea to put small-scale units of an uncontrolled technology that threatens system stability into individual homes and subsidize relatively rich members of society at the cost of everyone else while threatening the efficacy of our power system. If the” New Solar” revolution works and we can cost effectively have smart and regulated solar systems without any (hidden) subsidies, we will be the first ones to applaud the change. But despite what some sexy start ups claim, we are not there today.

The private person: Should I put PV on my roof?

If you think that spending 25k EUR on a 10kWp PV and a 10kwh battery is a good and green investment even with all the subsidies, think again. If you had put 25.000 EUR into the standard iShares Core S&P 500 ETF 20 years ago, you would have 152,622 EUR today, a return of 9.5% annually.

Let’s compare this with one famous German upstart’s offer that guarantees you an all-in energy price of around 5ct/kWh for roughly 5.000 kWh/year if you make this investment and use their services. With the ETF you would have made 152,622 EUR – 25,000 EUR = 127,622 EUR. With this money you could have bought 425.406 kWh of energy at a price of 30 ct/kWh or 21.270 kWh per year! So, your choice is between 5.000 kWh for 5 ct./kWh, or 21.270 kWh for free. But you want to put your money into green energy? Great, you could have bought, or still buy Global X Renewable Energy Producers ETF (RNRG), which only invests in green installations (large scale) with an annualized return of 7% and you would have made 96,742 EUR, leaving you 71,742 EUR to buy power, which at 30 ct. per kWh would still have allowed you to buy 16,124 kWh (so free power for this amount). Oh, and after 20 years, in the case of rooftop + battery you have to start replacing the equipment on your your roof and in the basement while in the ETF case you still have 25,000 EUR in the bank. So, PV/battery in your house even with all the subsidies in place: bad deal if you ask me.

Putting all this together, I believe the result is clear: Our beloved baby called Renewable Energy has grown to become an adult and is now an industrial sector, and it makes sense to take advantages of economies of scale and to expect it to behave like an adult (no more pampering needed). Also on the level of installations. I think it is great that homeowners want to participate in this project, but just like we shouldn’t base our food supply on people gardening in their backyard, we shouldn’t build our energy system on small, expensive and inefficient units that threaten system stability.

And now, we welcome the discussion.

Bias Disclaimer: FlexPower trades in large-scale wind and solar, so we may naturally lean against small-scale integrations. Despite this, I believe the facts support our stance.