Levelised Cost of Energy Models are Junk

Fake LCOE model results are being used to poison the debate about the cost of renewables

Introduction

When confronted with the actual costs of renewables, net zero lobbyists often trot out Levelised Cost of Energy (LCOE) calculations to claim that renewables are cheaper than gas. These LCOE models include those from Lazard, IRENA and figures produced by the Government in their latest Generation Cost 2025 report.

The trouble is LCOE calculations are based on models and like all models they are subject to garbage-in, garbage-out syndrome. For instance, these models do not compare like-with-like and often use false assumptions to arrive at their results. This is important because ignorant policymakers, desperate for evidence to confirm their bias towards intermittent renewables, cite these models to justify their decisions. Let’s dig into why LCOE models are junk.

What is the Levelised Cost of Energy (LCOE)?

LCOE models began in the 1970’s as a way of comparing the unit cost of electricity produced by different technologies. They work by adding up the capital and operating costs over the lifetime of a power plant, discounting the costs by an appropriate cost of capital and dividing by the amount of electricity generated over the lifetime. The result is expressed in pounds (or dollars) per MWh or pence (or cents) per kWh and often displayed in charts like those in Figure 1, taken from Lazard’s LCOE Version 18.0.

Figure 1 - Lazard LCOE Version 18.0 Results

LCOE Does Not Compare Like-with-Like

In the early days, they were comparing gas, coal, hydro and nuclear which are all dispatchable power plants and can, to varying extents, vary their output to match changes in demand. More recently, the same models have been used to compare the costs of intermittent wind and solar. Here we find the first problem, because wind and solar are not dispatchable – their output varies with the weather, not with demand. The value of a kWh of electricity produced by solar at midday when supply is already above demand is zero. Whereas the value of a kWh produced by a gas plant at 6pm on a cold January evening at time of peak demand is enormous.

LCOE models are not comparing like with like. They ignore the timing of supply and the ability to match supply with demand.

As we shall see below, some models attempt to correct this weakness by calculating the cost of firming intermittent renewables, but these corrections are only partial and leave a lot to be desired.

Inconsistent LCOE Calculations

Bodies such as IRENA publish global average LCOE measures, but others like Lazard focus on the US while the UK Government focuses on Britain. As a result they arrive at wildly different values for LCOE for different technologies. Part of this can be accounted for by geography. For instance, the load factor (LF) of solar in Texas might be around 25%, but in Britain the LF is about 10%. Differences in load factor will make a big difference to cost because higher assumed load factor means more generation which means the costs are spread over more MWh, reducing the cost per unit.

However, differences in geography are not the only reason LCOE varies. Figure 2 below shows the difference LCOE values from IRENA, Lazard, the UK Government Generation Cost report and the results of the latest AR7 auction.

IRENA’s results are consistently lower than the other LCOE calculations and much lower than the actual results achieved in the recent AR7 and AR7a auctions. For instance, IRENA estimates the global average cost of onshore wind in 2024 was 3.4c/kWh or about 2.5p/kWh (at an exchange rate of $1.35/£). This equates to just £25/MWh and is little more than a third of the £72/MWh 20-year index-linked contract strike price for onshore wind achieved in the recent AR7a renewables auction. Similarly, IRENA says solar costs just 4.3c/kWh or £32/MWh, less than half the £65/MWh AR7a contract price. The mid-points of Lazard’s calculations are also much lower than recent UK auctions, although their high estimates are much more reasonable. The Government’s Generation Cost report 2025 is much closer to the actual results of the AR7 and AR7a auctions.

This shows the danger of comparing global averages or US specific figures with UK conditions. The differences can be accounted for by looking at some of the key factors that impact LCOE most.

Impact of Capital Cost Estimates

The LCOE of wind and solar is extremely sensitive to the initial capital cost of building the power plant. This is because the capital cost is relatively high and incurred at the outset of the project before any revenue is earned from selling electricity. In the discounted cashflow model, the present value of the capital outlay is hardly discounted at all, but the revenue from the electricity generated 20 years hence is discounted heavily.

Part of the reason the LCOE results vary so much across models is the different capex assumptions used by the different organisations, see Figure 3 (all costs in £/kW, where appropriate converted at $1.35 per pound).

IRENA assumes onshore wind costs just £771 per kilowatt of capacity, well below the Lazard mid-point at over £1,500/kW. The Government’s Generation Cost report 2025 assumes £1,693/kW which compares to the capital expenditure of £1,865/kW for the Sneddon project that activated its CfD in 2024. Similar gaps exist for offshore wind and solar, although it has to be said Lazard is getting much closer to reality in its latest estimates.

Cost of Capital Delusion

Another reason IRENA is such an outlier in its LCOE estimates is that it assumes a ridiculously low cost of capital as seen in Figure 4. Using a low cost of capital artificially reduces the LCOE.

IRENA assumes a cost of capital between 3% and 3.7%, whereas Lazard and the Government are much more realistic with cost of capital in the range 7.6-8.9%. It is worth noting the Government uses a hurdle rate of 8.9% for gas-fired generators, well above their estimate for solar and onshore wind which artificially increases the cost of gas-fired electricity.

Of course, commercial companies will need to factor in higher discount rates if they are to make a profit over and above their weighted average cost of capital (WACC). For instance Orsted targets a return of 1.5-3% above WACC.

As an aside, the Climate Change Committee assumes a discount rate of 3.5% in its calculations which is one reason why they arrive at costs for offshore wind two and a half times lower than the recent AR7 auction.

Load Factor Fantasies

Load factor (LF) is another significant factor in LCOE calculations. High load factors mean more electricity is generated and so the costs can be allocated across more output and the levelised cost falls.

Figure 5 shows the assumptions made by IRENA, Lazard and the 2025 Generation Cost report and compares them to the actual achieved by the UK renewables fleet in 2024 according to Energy Trends data (Table ET6.1).

IRENA, Lazard and the UK Government all assume load factors for all technologies much higher than the UK renewables fleet achieved in 2024. This means they are systematically overstating the amount of electricity generated and so under-estimating the cost per MWh.

Asset Life Optimism

The various bodies calculating LCOE also make optimistic assumptions about the asset life of renewable technologies as shown in Figure 6.

The 2025 Generation Cost report assumes an asset life of 35 years for onshore wind, 30 years for offshore wind and 38 years for solar. The contract duration for projects awarded contracts in AR7 and AR7a has been extended to 20 years. After that period they will be reliant on market prices for their revenue. However, if there is a lot of renewable capacity on the grid, solar panels synchronise their generation and wind farm output tends to vary in unison. So, when it is sunny the value of the solar output will be close to zero and similarly the value of wind generation will be close to zero when it is windy. If the revenue they generate is close to zero, then they will be unable to cover their operating cash costs and will no longer be economic once the CfD contracts expire. The economic life of these generators is unlikely to extend beyond the subsidy period. This means they will generate less, and so the LCOE is again under-stated.

When it comes to gas-fired generators, the opposite is true. The Government assumes a life of just 25 years for CCGT generators, even though they are scrabbling around to extend the life of the existing fleet to 35 years or beyond to ensure we have sufficient backup when the sun isn’t shining or the wind isn’t blowing. This has the impact of over-stating the LCOE for gas-fired generation.

Carbon Costs Tilt the Playing Field

As well as making optimistic assumptions about capex, cost of capital, asset life and load factors to make renewables look artificially cheap, the UK Government makes ridiculous assumptions about the cost of carbon to make fossil fuels look artificially expensive as seen in Figure 7.

Figure 7 - LCOE of Dispatchable Technologies Commissioning in 2030

According to the Generation Cost report, gas-fired electricity commissioning in 2030 will cost £147/MWh using a 30% load factor. Gas generators would only deliver such a low load factor because of large amounts of intermittent renewables on the grid. Close analysis of the detailed spreadsheet they provide shows they expect carbon costs to be £41/MWh of the total. They base this calculation on their assumption that traded carbon prices will rise more than five-fold from £44/t in 2025 to £235/t in 2050. They say that carbon prices rise sharply after 2035 because abatement (emissions reductions) beyond the power sector will be more expensive to achieve. In other words, they must make carbon expensive to make it appear cost-effective to reduce emissions. This is sometimes referred to as target-consistent pricing used to make Net Zero look cheap.

Impact of Battery Storage

To partially address the problem of intermittency being ignored for wind and solar in LCOE models, Lazard attempts to calculate the cost of firming intermittency. IRENA and the UK Government do not even bother to try. First, Lazard calculates the cost of adding batteries, typically 2 hours of storage, to solar and onshore wind installations. Figure 1 shows the LCOE of solar rises from £38-78/MWh to $50-131/MWh when batteries are added. Similarly onshore wind costs rise from $37-123/MWh to $44-123/MWh when batteries are added.

We should note that in Lazard’s costs of standalone storage calculations, they assume that batteries can be charged at just $33/MWh, much less than the levelised cost of energy from any of the technologies they assess, except for an existing, depreciated gas plant. Low costs of storage rely on gas for their economics, or for renewables to sell power to them below cost.

However, the capacity of the batteries is not enough to give proper firm power, so Lazard estimates the cost of proper firming of intermittent renewables too. Figure 8 shows the results of Lazard’s work.

Figure 8 - Cost of Firming Intermittency (Lazard)

Their calculations show that for California (CAISO), unsubsidised solar costs $51/MWh, unsubsidised solar plus some storage costs $77/MWh. However, the storage is not enough to give proper firm capacity, so the cost of solar (or solar plus storage) rises to $142/MWh when the costs of capacity payments to a firming resource are considered. The $142/MWh is above the $48-109/MWh range for a combined cycle gas turbine. However, this too flatters the cost of “firm” renewables because the high-end cost of gas reflects a load factor of just 30%. The need to run gas turbines on such low load factors only arises because of renewables.

Batteries and Solar in the UK

When pointing out the intermittency and inter-seasonal variation of solar power in the UK, the retort from renewables advocates on X can often be characterised as: “haven’t you heard of batteries.” We should therefore look at the impact of battery storage on the cost of grid-scale solar in the UK.

We can reverse engineer an LCOE calculation using the parameters of a load factor of 12%, asset life of 38 years and the expenditure profile. This gives a baseline LCOE of £60/MWh, as per the 2025 Generation Cost report.

We can then use this model to test the sensitivity of LCOE to the additional capital expenditure of adding batteries. These sensitivity calculations have been more than kind to the cost of batteries for two reasons. First, the cost of batteries is assumed to be a one-off cost during the construction of the solar farm, but in reality the batteries will have a shorter life than the solar panels, meaning more capital will need to be spent to replace the batteries at least once during the life of the panels. Second, there will be losses from charging and discharging the battery, thus reducing the useful output of the solar panels. The impact of these losses has been ignored.

Lazard produce estimates for the cost of storage that range from $127-326 per kWh of capacity for a 100MW/200MWh battery, which translates to a midpoint cost of £150/kWh when converted into sterling at $1.35/£. However, we also have another datapoint for a large storage facility in the UK – Zenobe Capenhurst. This is a 100MW/107MWh battery, that according to their accounts cost £33.25m to install, giving a unit cost of £311 per kWh of installed capacity. Figure 9 shows how the LCOE of solar in the UK varies with the addition of 2-hour or 4-hour batteries using Lazard estimates and the actual costs from a real project in the UK.

The basic LCOE of £60/MWh rises to £84/MWh by adding a 2-hour battery and to £108/MWh for a 4-hour battery assuming the mid-point of Lazard’s battery storage cost estimate. However, using the costs of storage from Zenobe Capenhurst, the LCOE of solar plus batteries rises to £110-160/MWh. As we saw above with the Lazard costs of firming, the cost of solar plus batteries is far from the cost of a fully firm system. Extra balancing and backup costs will still be incurred over and above the LCOE.

Full Cost of Firming Renewables

As discussed above, intermittent renewables like wind and solar generate electricity according to weather conditions and do not respond to variations in demand on the grid. We therefore incur costs of backup, grid balancing and extending the grid to accommodate these technologies.

If we add the £1.96bn of extra balancing costs to the £1.25bn cost of backup from the capacity market in 2024 and divide by the 97.6TWh of generation from intermittent wind and solar, we can arrive at a cost of about £33/MWh for grid balancing and backup which should be added to the LCOE figures. The cost of expanding the transmission network to connect remote renewables should be added too. Once you add these costs, wind and solar no longer look attractive.

In Figure 7, the Government also inadvertently revealed the impact of renewables on the cost gas-fired generation. Without renewables. we might typically expect a gas fleet to run on about 60% load factor because of inter-seasonal variation. Their figures show the LCOE of gas (excluding carbon) varies from £70/MWh at 93% LF, to £106/MWh at 30% and £383/MWh at 5% LF. Running more renewables on the grid makes reliable backup more expensive because the capital costs must be recovered from less generation.

Full System Cost of Energy

A much better measure of the cost of renewables is the Levelised Full System Cost of Energy (LFSCOE) which attempts to quantify all the extra costs of intermittent renewables but is more difficult to calculate and subject to even more assumptions. However Bank of America (BofA) produced the results in Figure 10 back in 2023.

Figure 10 - Bank of America Levelised Full System Cost of Energy

According to BofA, the cost of a 100% wind grid in Germany would be $504/MWh (£373/MWh) and a staggering $1,548/MWh (£1,146/MWh) for a fully solar system. Their costs fall by nearly half if the aim is to achieve a grid using 95% of the chosen technology, but are still eyewateringly expensive.

Conclusions

The results obtained from LCOE models are highly sensitive to the input parameters. As a result they are subject to garbage-in, garbage-out syndrome. Sadly, many of the bodies promoting these models intentionally put garbage into the models, get garbage out, yet claim the favourable results for intermittent renewables are pristine and useful.

There are signs that some realism is creeping into the Government Generation Cost report from 2025, however, results of these calculations ignore the additional costs of integrating renewables into the grid to deliver firm power. Lazard is starting to get there, but even now, its results fall well short of the LFSCOE calculations carried out by the Bank of America. The best we can say is that LCOE models are junk.

---

This Substack now has well over 5,200 subscribers and is growing fast. If you enjoyed this article, please share it with your family, friends and colleagues. Please consider signing up for or gifting a paid subscription.