The signing of the 2015 Paris Agreement helped revive a global push to invest heavily in renewable energy. Much of the spending landed at the feet of wind and solar energy, which were seen as more mature technologies with a lower risk of failure.

In the US, both parties’ politicians had previously supported legislation that subsidized wind and solar technologies. Probably the most impactful subsidies have been the federal investment tax and production taxes credits. Both received second lives with the passage of the Inflation Reduction Act in August of 2022.

After years of observation, we now know that wind and solar assets generate at only about 25% of their stated capacity on average. As increasing amounts of renewable energy are piled onto our grids, these low capacity factors pose a serious risk to any claim that wind and solar generation are a workable solution to today’s modern economy.

Some have tried to redress the untrustworthiness of wind and solar power by urging for funding and building out all kinds of energy sources. This is the “all the above” or “energy diversity” strategy that we hear so much about. Despite its popularity, it would be a terribly wasteful strategy to pursue.

As it happens, a competing approach, one increasingly popular in green energy circles, has emerged. It’s to go “all in” on supposedly low-cost solar energy. Its popularity is based on the now almost ubiquitous claim that solar energy is now the “cheapest source” of electricity due to falling prices for panels from China.

Below are a few examples of how this claim has been pushed in the media over the years…

Solar photovoltaic is now the lowest-cost source of electricity in most places around the world.

— The Conference Board, February 21, 2024

A report from Ernst & Young shows that despite inflationary pressures, solar remains the cheapest source of new-build electricity.

— PV Magazine, December 8, 2023

Solar is now the cheapest electricity in history, confirms the International Energy Agency

— Carbon Brief, October 13, 2020

This narrative of falling costs is not limited to solar panels. It is also now attached to the large batteries that make up energy storage systems (BESS) on our electricity grids. Here are a few headlines about the falling costs and rising value of batteries…

The Rocky Mountain Institute… calculates that the cost of a kilowatt-hour of battery storage has fallen by 99% over the past 30 years.

— The Economist, June 20, 2024

Batteries are quickly moving from… niche applications to shifting large amounts of renewable energy toward peak demand periods.

— Helen Kou of BloombergNEF, New York Times, May 7, 2024

If you want more renewables on the grid, you need more batteries.

— Andrés Gluski, CEO of AES, New York Times, May 7, 2024

When the falling costs of solar panels are combined with the narrative of batteries as “cheap forms” of peak and backup power, some want to believe the reliability problem of renewable energy magically disappears.

As the argument goes, solar power is affordable, abundant and clean. And if we move quickly to this wonder source, we save money and the planet at the same time.

But there are big problems with these assertions.

For one, those making them—and the media repeating them—are confusing falling costs of certain solar components with the all-in cost of solar generation. The truth is panels are now a relatively small portion of the overall cost of utility-scale solar installations. The same is true when it comes to the falling costs of battery packs and BESS. Declines in panel and battery prices don’t have the effects on overall costs that they used to.

It’s as if the cost of sheetrock, say, fell by 80%, and the housing industry suggested that the resulting cost of building a house, and thus the cost of housing, had also fallen. Who in the housing industry would ever try to make such an outlandish claim? Yet, this is what renewable energy advocates say about the supposed viability—even the inevitability—of solar panels and battery energy storage systems (BESS) on our grids.

Another sleight of hand from the “solar as a savior” crowd is the implication that the sometimes very low cost of solar energy – for example, in the midday hours of a clear summer day – can be the cost of solar power over 24 hours, day in and day out. This is clearly not the case.

The truth is that solar power is unpredictable, and betting on it to meet the demands of modern society is risky. Creating a functioning system requires more than stacking panels. We would also need expensive batteries, a lot of money from the public’s wallet, and tons of luck.

Testing the Claim of Solar + BESS

Fortunately, we are not without the means to test the claim that solar is the cheapest source of energy, provided we are willing to do some work.

One obvious way to start to get a handle on the true cost of solar energy is to calculate the all-in cost of electricity for a hypothetical grid powered exclusively by unsubsidized utility-scale solar panels and BESS. Such a grid is called a “PV hybrid system” and allows us to isolate the cost of solar power when it is not supported by the availability of other energy sources.

When modeling a PV hybrid system, there are four basic rules to remember:

Every electron consumed on the grid is the product of sunlight hitting a solar panel, even those generated during the day, stored in BESS and later dispatched to serve nighttime demand.
Since solar panels can only produce when the sun is shining, nighttime power consumption must be met by BESS. There’s no other option.
As with any grid, the level, timing and duration of peak consumption on the system plays a major role in determining the system’s ultimate configuration. That is, we must design the system to, among other things, be able to reliably meet peak demand.
To ensure an apples-to-apples comparison with the costs of other forms of electricity, the system’s reliability must match that of the grids they are attempting to replace.

Assumptions: From Idealized to Real World Conditions

We start our analysis by determining the cost of solar energy under the most favorable of conditions—for both solar generation and consumption. This Scenario A allows us to identify the lowest imaginable cost for a PV-Hybrid system, as unrealistic as it may be. Once this “baseline” breakeven cost of generation is established, we can then incrementally add various real-world conditions to the grid, tracking the change in the breakeven along the way.

Modeling a PV-hybrid system in this manner allows us to both isolate what it would take for solar energy to power an entire grid, and understand which real-world conditions have the greatest effect on the real costs of solar.

Beginning Grid Characteristics

We settle on a grid with annual daytime consumption of 100 GW and average nighttime consumption of 50 GW. For context, this is about 50% more power than consumed in Texas in 2023.

For the idealized scenario, nature is assumed to be perfectly cooperative. The sun rises and sets like clockwork across the entire system at 6 a.m. and 6 p.m. There are no clouds and no variability. The sun is always positioned for maximum panel output in daylight hours. Solar panels produce a steady flow of energy, and the BESS takes care of the nighttime demand. Everything works like a dream. See footnote 10-A for visualization of supply and demand.

The system’s beginning installed panel capacity is 150 GW. This is sufficient to generate both the 100 GW of energy consumed in the daytime and the 50 GW stored in the BESS to serve nighttime consumption.

We assume panel capital costs average $1MM per MW and have an average lifespan of 20 years. BESS costs are assumed to be $3.6 MM per MW for a 12-hour duration with a 15-year lifespan.

Breakeven Cost of Generation

The math shows our idealized system, Scenario A, has a breakeven cost of generation of $38.90 per MWhr. As context, under real-world circumstances, the average all-in newbuild gas- or coal-fired plant has a breakeven of about $50 per MWhr, with a range of between $30 and $60 per MWhr depending on location, commodity prices and other factors.

Alas, the dream of generating at $38.90 per MWhr doesn’t last long once the following real-world conditions are systematically added to create a system closer to what we might see in a typical US Sunbelt state. The breakeven rises as the following conditions are added to the system:

Scenario B — Intraday Variability… or the kind of basic hourly change in consumption that occurs on virtually all systems. It comes about as households and businesses ramp up—and then ramp down—their activities each day. It adds $4.40 per MWhr to the breakeven cost of our system. See footnote xx for details.

Scenario C — Seasonal Variability… or higher than-average consumption during “peak” months of the year and lower-than-average consumption in the “off-peak” months. Specifically, we assume summer demand in the summer rises by 20% and winter demand falls by 20%. The remaining six months of the year show no change in consumption. The seasonal imbalance adds another $12.00 to the breakeven costs as rising capital costs begin to take their toll on the economics. See footnote xx for details.

Scenario D — Basic Weather… in the form of periodic clouds, rain and snow. These conditions serve to reduce by 20% to 30% the amount of sunlight assumed to hit the system’s panels. The assumed reduction in generation ranges from 30% in winter months to 20% in the summer. It adds $6.90 to the breakeven. See footnote xx for more detail.

Scenario E — BESS Energy Losses… need to be accounted for. We now assume that batteries discharge at 90% of their capacity, rather than the 100% in prior scenarios. We also adjust for the fact batteries have 15-year standard lives versus the 20 years assumed for solar panels. The costs of these adjustments are not trivial and add up fast, boosting the breakeven by $11.60.

Scenario F — All assets and infrastructure carry operating and maintenance costs. PV-hybrid systems are no exception. So, we now add fixed O&M costs of $25 per kilowatt for panels and $15 per kilowatt for batteries are added. Both escalate at 2% per year. The change adds $12.70 to the breakeven. See footnote xx for more detail.

Scenario G — What about the cost of capital for a PV-hybrid system? Lenders and investors don’t really hand out debt and equity financing with a pre-tax weighted average cost of capital (WACC) of 7.00%, as we’ve so far assumed. We increase it to a more realistic 8.40%, about the lowest we can justify. This increases the breakeven another $7.20 per MWhr.

Scenario H — Finally, to make the system work in the real world, we need asset overbuild. This is the additional assets (panels and/or BESS) needed to ensure the system won’t suffer prolonged blackouts in a worst-case scenario—which we define as panels receiving only 50% of a clear day’s sunlight over the course of five consecutive days. Analysis shows overbuild of panels is more economic than an overbuild of BESS, which carries a staggering cost. Nevertheless, the solar panel overbuild adds a whopping $33.50 to the breakeven.

The addition of these seven real-world conditions result in a system with peak hourly consumption of about 140 GW (at around Noon on a peak summer day) with off-peak, or minimum, hourly consumption of about 55 GW around midnight. The final demand profile also includes a natural ramp up to the daytime peak and ramp down to the nighttime low-point. See footnote xx for more detail.

So, how do these real-world conditions change things in terms of the system’s capacity requirements and the resulting capital costs? Pretty significantly it turns out. In fact, it’s massive increase in capital costs and the resulting decline in capacity factors, both of which arise because there are no dispatchable generation assets to rescue customers. To wit…

Required solar panel capacity needs rises from 150 GW to 503 GW… an increase of 235%

Required battery capacity jumps from 50 GW to 72 GW… or 44%

Total invested capital required increases 151%… from $330 to $828 billion

See the table in footnote xx for more information on changes in assumptions and output across scenarios.

Conclusion

All totaled, the addition of the real-world conditions pushes capital costs up 151% and the breakeven of the system to $129.25 per MWhr. This breakeven is over three times that of the idealized system, and well in excess of real-world coal- and gas-fired generation breakevens, driven higher by low capacity factors and high capital costs.

Chart showing the rise in capacities and capital costs across scenarios — Chart 5: Capacities and capital both rise as real-world conditions are added

Chart showing the rise in breakeven generation costs across scenarios — Chart 6: Generation breakeven rises as real-world conditions are added.

So you tell us, if solar energy is in fact the “cheapest” form of electricity in existence, why is it that a grid powered only by solar power—without the benefit of taxpayer subsidies and an installed base of dispatchable thermal plants to step in when mother nature doesn’t cooperate—requires a breakeven price of generated power so much higher than that of newbuild natural gas or coal-fired units?

Solar power may dazzle those who refuse to run the numbers. But once they are run, solar quickly begins to look like an uber-expensive non-solution to our energy needs.

The stakes are too high to keep kidding ourselves about reality—solar energy is too unreliable, and thus too expensive, to depend upon. There are better low carbon solutions.

Footnotes

1. Additional Assumptions & Inputs

A) The only kinds of assets on the system are unsubsidized utility-scale photovoltaic solar (panels) and battery energy storage systems (BESS). This configuration is what’s referred to as a PV-hybrid system.

B) Importantly, total annual demand on the system doesn’t change across scenarios, only the shape of demand. That is, regardless of the scenario, average annual demand remains 75 GW, or 657 TWhrs per year.

C) The breakeven cost of generation is the average price of generated power needed over 20 years to exactly meet investors’ return expectations after paying capital, operating and financing costs. The breakevens for all scenarios are expressed in 2024 dollars per MWhr.

E) Determining the capital costs for BESS was a challenge. After a thorough search, we found a quality 2023 study by the National Renewable Energy Laboratory (NREL), a US Department of Energy unit that analyzed data from 16 respected sources. The study acknowledges the range of estimates and fast pace of change in costs for BESS. It also recognizes differences in BESS cost elements between studies. For example, one estimate might include inverters, while another might not. In the end, NREL chose a simple average of all 16 estimates for BESS with four-hour duration and 15-year lives. The study also includes O&M estimates, which we will use in Scenario F.

F) Capacity factors of generating assets measure actual output relative to the potential of an asset over time. Factors of 70% or above usually indicate base-load operations. Factors below 15% are common for peaking assets. Intermediate services fall between the two.

2. Scenario A: Baseline System

A) For our idealized system, the sun rises and sets completely and immediately throughout the system at 6 a.m. and 6 p.m. All days are 100% cloudless, with the sun always positioned so that panel output is maximized during daylight hours. Power supplied by panels is level at 150 GW across all daytime hours, and zero during nighttime hours. Daytime consumption is level at 100 GW. Nighttime consumption is half as much at 50 GW.

B) Installed panel capacity to serve baseline needs is 150 GW. This covers both daytime consumption and the charging of the BESS. We assume panel capital costs average $1MM per MW and have economic lives of 20 years.

C) The system has 50 GW of 4-hour duration BESS that services nighttime demand. Capital costs of installed BESS are assumed to be $300 per KWhr per estimates by NREL. This comes to $3.6 MM per MW for a 12-hour duration. The BESS is assumed to have an economic life of 15 years. See Table XX for more information on BESS capital costs.

D) For the baseline scenario, total system costs are limited to capex and investor returns. Neither panels nor BESS incur O&M costs. We also ignore the costs of energy leaks, degradation and replacement costs of BESS,. which are not assumed to carry any O&M costs at his point.

E) We select a generous 7.00% as the project’s pre-tax weighted average capital cost (WACC).

G) The system’s resulting capacity factors are 50% for the panels and 16.7% for the BESS.

3. Scenario B: Adding Intraday Variability

A) In the real world, demand for energy isn’t constant. People use more power during the day and less at night. And, of course, the sun doesn’t shine all day. Rather than coming online at 100 GW at 6 AM, daytime demand now begins at 65 GW at 6 AM and rises to a peak of 125 GW by 11 AM. It remains at 125 GW for two hours. At 1 PM, demand begins to drift back lower. It’s 65 GW by 5 PM, which reverses the AM ramp-up. Total daytime consumption of 1,200 GWhrs does not change. See footnote xx for more information on Scenario B’s daytime supply and demand profile.

B) Nighttime consumption begins at 60 GW in the 6 PM hour. It drifts to its low of 40 GW by 11 PM, where it holds for two hours. It then drifts upwards beginning at 1 AM, reaching 60 GW in the 5 AM hour. Total nighttime consumption remains 600 GWhrs. These patterns repeat each day, every day. See footnote xx for more information on Scenario B’s daytime supply and demand profile.

C) The system’s new challenge is to handle a variable daytime demand peak of 187.5 GW, up from 150 GW previously. To accommodate this peak, we increase panel capacity to 187.5 GW. This pushes total capital investment $367.5B, up from $330B. Panel capacity factor drops to 40%.

D) We assume BESS can vary discharge rates as needed to match hourly nighttime demand. This means, since nighttime demand does not change, BESS’s costs and capacity factors don’t change.

4. Scenario C: Adding Seasonal Variability

A) In the peak summer months, we now assume demand jumps, driven by the demand for electricity to keep us cool. In July, August, and September, we assume demand of 90 GW. We need more solar panel capacity to handle this summer load, even though it means having more capacity than we need during the remaining nine months. In the off-peak months of January, February and March, demand is just 60 GW. The remaining “shoulder months” experience no change in demand.

B) Adding seasonality has big implications. For one, peak hourly demand (consumption plus BESS charging) rises to 235.7 GW. This is now the new required capacity for panels. We also now need 15 more GW of BESS, bringing the total to 65 GW.

C) Capital costs increase to $469.7B with this new configuration. Since capacity has risen even as annual demand hasn’t changed, panel capacity factor falls from 40.0% to 31.8%. BESS’s capacity factor falls from 16.6% to 12.8%. The declines reflect the fact that the extra panel and BESS capacity to meet peak demand for three months is a costly burden in the other nine months.

5. Scenario D: Addition of Basic Weather

A) Anyone who lives outside the desert understands that the weather doesn’t always cooperate with the needs of solar energy. Clouds, storms, rain, sleet and snow are part of Earth’s climate. They all serve to reduce the ability of a PV-hybrid system to generate reliable energy.

B) In this scenario, the amount of sunlight hitting panels changes as the weather changes, which is assumed to be different in each season. To wit, we imagine 10-day periods for each season to test weather’s effect. Winter is our off-peak season. It enjoys three days of pure sunshine. Three days are 75% sunny with some clouds. Another three days have 50% sunshine and partial clouds. One day has heavy clouds, rain, or snow with 25% sunshine. On average, winter is 70% sunny. Summer is peak season. It’s assumed to be 80% sunny on average. Spring and Fall, both shoulder months, have sunshine levels of 77.5% and 75.0%, respectively.

C) Since we must build to meet peak demand, we scale-up the system using summer’s 80.0%. sunshine factor. The calculations are simple. The weather reduces panel performance by 20%, so more panels operating at 80% are needed to meet peak demand. This pushes installed panel capacity to 287.8 GW (235.7 GW / 0.80). Total capital employed jumps to $528.6B. The capacity factor of panels falls to 25.5%. BESS’s factor remains at 12.8%.

D) Bear in mind, the factors we’ve accounted in this scenario are rising costs and reduced efficiencies introduced by basic weather, not full-on storms that produce enough hail or heavy winds to take out an entire solar farm of panels.

6. Scenario E: Addition of BESS O&M & Inefficiencies

A) In the baseline, we assume the system’s BESS charges and discharges at 100% efficiency. And that it does so every day for 20 years. This isn’t realistic. We need to account for the wear and tear on batteries, inefficiencies and ongoing maintenance costs. So, we now assume BESS charges and discharges 90% of its capacity, after losses and inefficiencies. This affects the configuration of the system in two ways. First, nameplate BESS capacity increases by 7.2 GW. This allows net BESS capacity to meet nighttime load of 65 GW. Second, panel capacity increases by 7.2 GW in order to charge the additional BESS.

B) We also resolve the mismatch in assumed lifespans of panels (20 years) and BESS (15 years) in this scenario. To do so, we adjust capital invested in BESS at the beginning of the project. For BESS cost in year 16, we adopt NREL’s estimate (in 2024 dollars) of $225 per KWhr. This is below the $300 per KWhr for the original BESS installation. We need five years (year 16 thru 20) of replacement BESS to match the 20-year life of panels. So, we add a third of the $225 cost, or $75 per KWhr, to the upfront capital costs. BESS capital costs rise to $375 per KWhr, up from $300 initially. See footnote xx for more information on future capital costs of BESS.

C) In total, BESS inefficiencies and replacement costs increase initial capital investment by $98.2B, or 19.1%. This breaks down to $91.0B for additional BESS assets and $7.2B for additional panels.

D) Capacity factors for panels and BESS fall to 24.8% and 11.5%, respectively.

7). Scenario F: Addition of Solar Panel Fixed O&M

A) Until now, we’ve assumed no variable O&M for panels and BESS. But fixed O&M (FOM) is too material to ignore. For it, we select $25 per KW for panels, and $15 per KW for BESS. Both are at the low end of NREL’s mid-point estimates. NREL bases its FOM estimates for panels on a mid (moderate) case over 20 years and mature technology. For BESS, the assumption is 4-hour utility scale storage and a 20-year life. These assumptions appear generous, as EIA data from 2024 show lifespans of solar panels spans in the US of only nine years. And seven years for BESS. See footnote xx for more information on O&M cost estimates.

B) For panels, major FOM costs are insurance, land lease payments, property taxes, inverter specialists, transformers, connection fees, and replacement parts. Inspection fees, cleaning, electricians, array and module specialists, general maintenance, and other miscellaneous costs are also included. BESS categories are similar but include battery augmentation and exclude inverter costs.

C) We assume a 2% annual escalation in FOM. This leads to average annual fixed O&M of 2.4% ($7.5B / $301.8B) of initial capex for panels. For BESS, it’s 0.6% ($1.1B / $325.0B). Total FOM below 1.4% of the initial capex would likely be well-received by most investors.

D) NREL’s O&M estimates seem to include replacing some large panel and BESS components over their technical lives. This supports the assumption that, once installed, such assets are replaced and upgraded, not abandoned, after 20 years. We therefore don’t allocate any extra O&M or capex for panel or BESS decommissioning. Terminal values also are not included.

8. Scenario G: Addition of Realistic Pre-tax WACC

A) In this scenario, we adjust our assumed pre-tax cost of capital from 7.00% to something more realistic. Let’s first clarify what we mean by the cost of capital. In short, it is the pre-tax weighted average cost of capital (WACC) of the project. Based on the principles of the Capital Asset Pricing Model (CAPM), it captures the total cost of capital from all sources needed to finance the system’s construction and operation. It includes all forms of debt and equity. Simply put, WACC is the average rate that the project pays its lenders and investors. It also serves as the minimum return, or “hurdle” rate, that investors require from the project.

B) Setting a practical WACC begins with determining the capital structure. Despite our doubts, we favorably assume a system of this size can be optimally funded with 60% non-recourse debt. The remaining 40% is equity.

D) For the interest rate on the debt portion of the financing, we look at 20-year US Treasury bonds—the life of the project. In the summer of 2024, these bonds traded at 4.50%. We assume a 75 basis-point spread, or risk premium. This gets us 5.25% as our debt rate. We assume equity investors expect a 10% annual return over the project’s 20-year lifespan.

E) After applying weightings of 60% and 40%, these two assumptions produce an after-tax WACC of 7.15%. If we assume the project’s pre-tax cash flows are 15% lower after taxes, this produces a pre-tax WACC of 8.41% (7.15% / 0.85). We’ll round to 8.40%.

9. Scenarios H1 & H2: Addition of Overbuild/Reserve Generation

A) Though we can imagine worse, a five-day period of adverse daytime weather—especially during a during a peak month—would be devastating to the PV-hybrid system. To avoid this devastation, this scenarios adds the necessary reserve or “overbuild” capacity to the system to protect against blackouts in a worst-case event.

B) We must determine what the worst-case scenario looks like. We settle on a prolonged period of adverse daytime weather during a peak month. We pick five consecutive days of very cloudy, rainy, and/or snowy conditions. During this period, we expect the generation of solar power to decrease by half.

C) We assume daily demand does not change under worst-case conditions. We could argue demand would go higher or lower. To wit, it could fall during a hurricane, but rise during a winter storm. The latter would mean our overbuild is not sufficient.

E) We assume one-third of the overbuild pays for itself. This could be from arbitrage, export, or other opportunities arising from having the extra capacity on the system. Thus, only two-thirds of the overbuild capital and O&M costs are in the project economics.

F) We first calculate the numbers for an overbuild in panel capacity—extra panels sufficient to meet the demand, including recharging BESS, for five days. We designate this scenario as H1. It requires 301.8 GW of extra panel capacity. Net of 100.6 GW assumed to pay for itself, the cost of the overbuild is $201.2B. Total capital investment rises to $828.0 B.

H) Next, we calculate the figures assuming the overbuild is limited to BESS capacity adequate to meet the cumulative unmet demand of 3,600 GWhrs over five days. We call this Scenario H2. It requires 332.8 GW more BESS capacity, after excluding 167.2 GW presumed to pay for itself. So, if 72.2 GW of charged BESS capacity perfectly services 780 GWhrs of demand each night after leakage and inefficiencies, 332.84 GW of charged capacity can be expected to discharge 3,600 GWhrs over five days (3,600 / 780 x 72.2) as needed. This means total BESS required on the system is 72.2 GW + 332.84 GW, or 405.04 GW. The combined capital costs for all BESS surge to $1.82 trillion. Total system capital investment skyrockets to $2.1 trillion. H2’s breakeven is a massive $299.01 per MWhr. We select Scenario H1. The system’s real-world breakeven rises to $129.25 per MWhr.