After a rocky start, the transition to electric vehicles now seems unstoppable. Battery prices have fallen by roughly 90% since 2010. The International Energy Agency (IEA) had to revise its projections upwards again and again because deployment kept surpassing previous forecasts. In 2015, 330,000 fully battery electric cars (BEVs) were sold globally or around 1% of all new cars. In 2025 it was 14 million, a 27% increase from the year before and nearly double the figure of just three years earlier.

That was close to a fifth of all cars sold globally. When we include plug-in hybrids, 5% of the world’s car fleet is now electric and more than two thirds of all electric cars sold are now fully electric. Most recent IEA analysis shows electric car sales prove resilient despite a weakening global car market.

Critics will point out that the internal combustion engine still rules both in terms of new car sales and the stock of cars in operation today. And they are right. But it is worth reminding ourselves how quickly the progress we have seen has happened. In 2008 the Tesla Roadster was launched, an expensive concept car that managed a 200-mile range. This was followed by the Nissan Leaf, the first widely successful mass-market car still in production today. Other smaller and more affordable cars entered the market with the Renault Zoe in 2011. The Tesla Model S offered the first premium long-range saloon in 2012 with lasting commercial success until the present day. But it was Tesla’s Model 3 launched in 2017 that would transform the car market and reach truly high production volumes. Many carmakers followed and by 2025, there were around 650 different fully electric car models globally and this number keeps increasing every year.

The people who ran the car industry did not see this coming. In 2015 the former CEO of Daimler, the company named after one of the inventors of the modern automobile, publicly dismissed Tesla as a joke, not a serious company, an enterprise that would never take off. This was not an outlier view. It reflected the prevailing perspective in an industry that developed for more than a century perfecting the internal combustion engine. A rapid shift away from it seemed inconceivable. Reality proved the incumbents wrong: within a few years Tesla was the world’s most valuable carmaker by market capitalisation. In 2023, the Model Y became Europe’s best-selling car, the first battery electric vehicle ever to do so.

The endgame

If you want to see where this ends, look at Norway. Norway is often portrayed as a special case. One of the richest countries in the world with vast oil and gas resources. That certainly helps. But it is not the full story. Norway had been incentivising EVs since the early 1990s, decades before most countries developed policies to support EV uptake. The results are impressive. In the first quarter of 2026, fully electric cars accounted for 98% of all new sales. That’s of course only cars sold. But even of the stock 32% are now fully electric.

How did they achieve this? It was a combination of different measures. Electric cars were exempted from the purchase taxes imposed on ICE cars. Electricity is cheap relative to petrol. Free parking, reduced tolls, access to bus lanes, and cheaper ferries were all softer measures deployed to make EVs more attractive to people.

China dominates

As with so much of the clean energy space, China dominates global electric car markets. In 2025 8.2 million fully electric cars were sold in China. That is more than half of global BEV sales. China also built an industry that barely existed fifteen years ago. China has now risen to the world’s number one car exporter by volume. BYD, a company most Europeans had never heard of five years ago, has overtaken Volkswagen as the best-selling brand in China.

Emerging markets

The real news story in the EV space right now is not that Norway has transitioned and China is the largest market. It is the growth we are seeing in emerging markets. Take Ethiopia. In January 2024 it became the first country in the world to ban the import of petrol and diesel cars. It also introduced a zero customs duty on locally assembled electric cars. By the end of 2024 more than 60% of new cars sold were fully electric. The logic was not so much climate policy but energy security: Ethiopia spends a lot of money every year on imported fuel while its own grid runs on more than 90% renewable electricity. But Ethiopia is a small country in terms of the numbers of cars being sold and could be an outlier. Just that it isn’t – there are many other countries in emerging markets now where electric cars are catching on.

Nepal, another mountainous country rich in hydropower and dependent on imported fuel, went from electric cars making up less than 1% of new four-wheeler imports in 2019–20 to more than 75% in 2024. It took only four years to achieve. Early adopter Norway needed almost two decades to achieve something similar.

The pattern that is emerging is that in more and more countries BEVs start to outsell ICE vehicles following the typical X shape of transitions. Robbie Andrew helpfully tracks this and his data allowed me to produce the chart below.

Where next?

By 2035, the IEA expects close to 50% of all cars sold globally and close to 30% of the car fleet to be electric.

The consequences will reach far beyond the car industry itself. Road transport accounts for nearly half of global oil demand. Every electric car that replaces a petrol or diesel car permanently removes a small amount of continuous oil consumption. Over time, it all adds up.

A good example is China: In H1 2026, EVs displaced an estimated 33.7 Mtoe of oil consumption, up 42% from 23.7 Mtoe in 2025. This is more than three times the EV-driven oil displacement in H1 2023 which was 11.6 Mtoe. The fossil fuel displaced by EVs in the recent six months was roughly equivalent to 6% of China’s total crude oil imports in 2025, around 22 days of imports at last year’s average rate.

Anyone today who judges the future of transport from the current share of electric cars on the road will be surprised. The fleet on the road tells you more about the past. The sales figures tell us about the future and they are only going in one direction.

This complex undertaking is often characterised by “non-linearity” and “feedback loops”, where small changes can go on to have disproportionately large impacts and where seemingly straightforward paths encounter unexpected roadblocks.

Interventions can be self-amplifying – leading to runaway change, or they can be self-defeating – when progress seems impossible to attain.

Our policy brief sheds light on these intricate dynamics, which can be overlooked when governments use analytical frameworks based on standard economic thinking.

The brief sets out the most common archetypes of system change and behaviour, as well as the underlying feedback loops that drive them, with the aim of helping policymakers to understand the recurring patterns that can either accelerate or impede progress.

Governments that can recognise these patterns – as well as the ways they can be harnessed or sidestepped – are likely to be better equipped to manage structural change.

This article delves into three key examples from the policy brief, exploring how they are influencing the energy transition and what lessons can be drawn for effective policymaking.

Reinforcing feedback loops

At the heart of the energy transition lies a powerful engine: the reinforcing feedback loops inherent in the development and diffusion of many clean-energy technologies.

This virtuous cycle operates through several mechanisms.

First, “learning by doing”, which means that as more units of a technology, such as solar panels or wind turbines, are produced and deployed, manufacturers and developers become more efficient, processes are refined and costs fall.

Second, economies of scale kick in: as production volumes increase, unit costs decrease due to efficiencies in manufacturing and more developed supply chains.

Finally, wider deployment can trigger network effects and the emergence of complementary innovations. This means that as the adoption of a given technology grows, it can foster an ecosystem of supporting infrastructure, skilled labour and supporting technologies, which can further boost its attractiveness and viability.

Together, these three elements create a powerful reinforcing loop: initial investment drives innovation and cost reduction, which spurs increased demand, attracting further investment.

Solar photovoltaics (PV) and wind turbines are prime examples of this dynamic.

The astonishing growth of solar offers a particularly vivid illustration of the way in which reinforcing feedback loops can blindside experts and policymakers alike.

Solar growth has far exceeded projections made in the early 2000s. Indeed, the world’s actual installed capacity in 2020 was over 700 gigawatts (GW), more than ten times the level expected in outlooks published in 2006, as shown in the figure below.

Global solar deployment has exceeded expectations due to disparate trends and drivers in individual markets that, together, all point in the same direction. China, for instance, met its 2030 target for wind and solar capacity six years ahead of schedule in 2024.

Batteries are also riding this wave, with costs plummeting by around 85% over the past decade as deployment, particularly in road transport, scales up.

However, not all clean-energy technologies benefit from this self-amplifying pattern.

Nuclear power and hydropower, for example, have historically not shown the same rapid cost declines, due to their large, complex and site-specific nature. This contrasts with the smaller, modular and replicable characteristics of technologies, such as solar PV.

This does not negate the potential role of such technologies, but it does mean that they are less likely to see disruptive, exponential and self-reinforcing growth.

There are a number of potential conclusions for policymakers.

Early in the transition, interventions such as feed-in tariffs and public procurement were crucial in kick-starting these reinforcing feedbacks for solar and wind.

As these technologies mature and become cost-competitive, the focus shifts to removing other barriers, such as streamlining permitting processes, investing in grid expansion and reforming markets so they are better able to integrate variable renewable output.

These same principles could now be applied to newly emergent clean-energy technologies. Policies that directly nurture these reinforcing loops, such as deployment subsidies and clean technology mandates, can be expected to be most effective in the initial stages.

Turning again to the example of solar energy, while such initial efforts appeared to be expensive, they paid off over time by unlocking future cost reductions and, thus, kick-starting the self-amplifying feedback loops that are now driving further progress.

This contrasts with the idea that carbon pricing is necessarily the most efficient policy for decarbonisation. It may well be helpful, but as it will not drive rapid early technology adoption, it is less likely to have a self-amplifying effect in the initial stages of the transition.

Renewable ‘cannibalisation’

While the growth of renewable energy is the driving force of the energy transition, another system dynamic, termed “renewable cannibalisation“, can act as a dampening feedback loop. This can potentially slow progress long before full decarbonisation is achieved

This cannibalisation process results in variable renewable energy (VRE) sources, such as solar and wind, receiving decreasing prices for the electricity they generate.

Essentially, the more solar and wind capacity that is connected to the grid, the more they undermine their own revenue. This happens through three main channels.

First, the merit order effect, whereby solar and wind, which have very low operating costs, push more expensive fossil-fuel generators out of the market when supply is abundant.

In markets with marginal pricing, this leads to lower wholesale electricity prices during periods of high renewable output. While this cuts prices for consumers – at least in the short term – these lower prices also reduce revenues for renewable generators, potentially undermining the economic case for further investment.

For example, in California, solar power unit revenues fell by $1.30 per megawatt hour (MWh) for each percentage point increase in solar penetration between 2013 and 2017.

Second, price volatility, where uncertainty over future trends in the generation mix and the balance between supply and demand can make long-term revenues difficult to predict.

This increased uncertainty can raise the cost of capital for new renewable projects, again acting as a brake on investment

The UK, for example, experienced this before the introduction of “contracts for difference” (CfDs), which helped stabilise revenue expectations for renewable developers.

Third, volume risk, where rising VRE capacity increases the likelihood of more frequent curtailment – periods when renewable generation exceeds demand or grid capacity, forcing generators to scale back output and lose potential revenue.

Curtailment in itself is nothing new, but the scale and frequency is changing. Recent analysis by University College London suggests that without significant flexibility or storage, UK renewable generation could exceed demand for more than 50% of the time by 2030.

The analysis found that installed wind and solar capacity is set to surge beyond current levels of electricity demand, as illustrated in the figure below, finding that this could “deter investment” in new projects if no action is taken to address the problem.

These dampening feedback loops illustrate a classic “limits to success” scenario. The very success of renewables, if unmanaged, can create conditions that hinder their continued expansion.

The policy implications here are nuanced. One solution is CfDs, which offer renewable generators a fixed price and have been effective in many countries at mitigating the merit order effect and price volatility, thus maintaining investment.

However, as VRE penetration becomes very high and surplus generation becomes a regular occurrence, other solutions are likely to be needed. This is because existing CfD designs often include clauses that stop payments when market prices drop below zero.

As a result, alternative CfD designs, guaranteeing revenues based on installed capacity or potential – rather than actual – electricity generation might be considered, for example, even though these have other drawbacks.

More fundamentally, our research suggests the solution to this challenge lies in fostering the co-evolution of renewables with technologies such as energy storage and green hydrogen production. These can absorb surplus generation and turn a problem into an opportunity.

Whereas, traditionally, it might be assumed that the market on its own can optimally allocate risk, research suggests that a redesign of market structures may be needed to enable investment and fully realise the cost-saving opportunities of the new technologies.

This is one of several sets of feedbacks discussed in a separate new report published today, looking at the power sector transition in China.

The power of connection

The energy transition is not a series of isolated changes in different sectors. Instead, it is an interconnected system, where progress in one area can catalyse shifts elsewhere. Shared technologies can create reinforcing feedbacks that accelerate decarbonisation across multiple fronts, generating cross-sector synergies.

The relationship between clean power and transport electrification is a powerful example of this. As batteries are deployed at scale in electric vehicles (EVs), their costs fall, enabling ever-wider deployment and further cost declines, as shown in the chart below.

This is due to the learning-by-doing and economies-of-scale feedbacks discussed above.

This cost reduction then makes batteries more viable for grid-scale energy storage, which, in tur, helps integrate more low-cost VRE into the power system.

Cheaper, cleaner electricity then further incentivises the electrification of transport, as well as heating and light industry. This increased electrification boosts demand for renewable power, driving further deployment and cost reductions in solar and wind. It also expands the potential for demand-side response, where consumers adjust their electricity use to help balance the grid.

A similar dynamic is anticipated for “green” hydrogen. As deployment in one anchor sector – perhaps fertilisers or refining – drives down the cost of electrolysers, it makes green hydrogen more competitive for other applications, such as shipping or even long-duration energy storage in the power sector.

Each sector’s adoption of green hydrogen contributes to the shared learning and cost reduction, benefiting all.

The policy implications of these cross-sector synergies could be significant. Their existence suggests, for example, that there is no need to wait for decarbonisation of the power sector to advance further, before beginning the electrification of transport, heating or industry.

This is in contrast to the argument that transport should only be electrified after cutting power sector emissions, since increased EV charging will drive up demand for gas- or coal-fired generation.

While there will be a marginal increase in emissions from plugging a new EV into the power grid, the insights described in our brief imply that it is still likely to be more effective to pursue the transition away from fossil fuels in multiple sectors in parallel, because it can activate beneficial cross-sector feedback loops that are greater than the sum of their parts.

As such, our research suggests that policymakers hoping to take advantage of cross-sector synergies could aim to deliberately strengthen technological linkages between different parts of the energy system. Examples include electricity tariffs and market structures that reward “smart” EV charging and vehicle-to-grid (V2G) services, encouraging industrial participation in demand-side response and promoting integrated home energy systems. These interactions can amplify the benefits of early investment in the transition.

Policy insights from system dynamics

Archetypes such as the self-reinforcing growth of clean technologies, the potential for renewable cannibalisation, the accelerating power of cross-sector synergies and seven others described in our new report paint a picture of a transition that is far from linear. Instead, we find that it is governed by complex interdependencies and feedback loops.

Consequently, our research suggests that policymakers will be much better equipped to manage and steer the transition, if they adopt a systems thinking approach.

Recognising these recurring patterns allows for the design of more robust and effective policies that anticipate challenges and leverage opportunities.

For instance, understanding the power of reinforcing feedback loops in technology diffusion underscores the value of early-stage support for nascent clean-energy technologies.

Conversely, anticipating the dampening effects of renewable cannibalisation highlights the likely benefits of combining renewable buildout with evolving market designs and strategic investments in flexibility solutions, such as storage and demand-side response.

Policymakers that understand and work with these dynamics are likely to be in a better position to spark self-amplifying changes – achieving maximum value for minimum effort – and to avoid self-defeating interventions that go nowhere.

This is a reprint of an earlier version of this article published in CarbonBrief with Simon Sharpe and Max Collett.

That number gets cited constantly as proof that replacing fossil fuels with clean energy is a) not happening and b) unrealistic. Renewables, the argument goes, are simply added on top of a growing amount of fossil fuels. The apparent logical conclusion drawn is that we are not in an energy transition but in an energy addition with little chance of ever phasing out fossil fuels.

The claim has serious voices behind it. In a prominent Foreign Affairs piece called ‘The troubled transition’, Daniel Yergin and co-authors point out that the fossil fuel share of primary energy fell only from 85% in 1990 to about 80% today, while fossil fuel demand keeps growing. Their conclusion from these trends is that what we’re seeing is a clean energy addition, not an energy transition. And the chances for that changing in the future are presented as slim. Widely-read energy analyst Vaclav Smil argues that replacing a system supplying more than 80% of the world’s energy within decades would be an energy transition of unprecedented speed, something never done before in history. Others go further still, calling the idea that wind, solar, and batteries can replace fossil fuels pure fantasy because the scale of demand is too great and renewables lack the density to power modern civilisation.

At first glance these arguments may seem very compelling and I see them every day on social media. Let me be clear: the scale of the challenge is huge and it is real. Pretending otherwise would be very naive and no serious energy expert disputes this. It is also true that for most of history, clean energy has scaled far too slowly to replace fossil fuel demand also because energy consumption kept growing driven by an increase in world population and economic activity. At the global level and in absolute terms, renewables have indeed been added on top of fossil fuels. One could, and many do, decide to call the discussion concluded at this point and simply accept that the energy transition is a futile undertaking.

I tend to differ. But look sectorally and regionally, and the picture changes.

I’ve written about example after example of energy systems already changing beyond recognition, whether that is electric vehicles in Norway, the rapid rise of batteries in places such as California, Pakistan’s bottom-up solar revolution, Spain’s successful phase-down of natural gas in power markets, and the shift to heat pumps in the Nordic countries and elsewhere. Looking only at global data obscures these remarkable shifts and leaves us with the impression that nothing has changed.

There are clear indicators of the energy system moving. Since 2016 more money is invested in clean energy every year than in fossil fuel technologies, today around twice as much.

Roughly 90% of new electricity capacity is renewable.

Of course capacity is not generation but generation also increasingly shows how fossil fuel-generated electricity is declining as a share of global electricity production.

Another example is that the global car market has shifted with one in six cars sold last year being fully battery electric. And many markets in 2026 reported record growth in electric vehicles suggesting a further upswing is just around the corner.

Often overlooked but critical, feedback loops, which can be both positive and negative, mean change doesn’t happen linearly. Technology transitions often happen exponentially and unexpectedly, as I discuss with colleagues in Nature Reviews Clean Technology.

There’s a second problem with the “energy addition” case put to us. The claim that a large-scale shift is unrealistic because of the sheer scale of fossil fuel use overlooks fundamental differences in efficiency, conversion, and end use. The Canadian energy expert Paul Martin called this the “primary energy fallacy” and many including myself have written about it. This fallacy arises when comparing energy sources in primary energy terms rather than their end-use utility in terms of energy services. This is important, as around two thirds of global primary energy consumption is lost during combustion and conversion processes.

Primary energy represents the total energy content of natural resources before undergoing any conversion processes, including coal, oil, biomass, and renewable sources. The primary energy fallacy arises when high primary energy inputs are mistakenly equated with high energy services. Measuring energy systems solely in primary energy terms distorts comparisons. It inflates the perceived contribution of fossil fuels while underestimating the efficiency of renewable energy and the energy savings achievable through electrification. Michael Liebreich estimates that today clean energy sources contribute around 30% to useful energy (energy services) up from 10% in the 1970s. This is not shown in statistics on global energy demand.

And this isn’t just an accounting quirk or a subject of debate amongst energy nerds. This has practical consequences: policymakers may hesitate to support clean energy technologies if they incorrectly believe replacing fossil fuels is impossible. Likewise, certain policies focused on reducing primary energy demand (such as those promoting fossil fuel combustion efficiency improvements), rather than maximising carbon reduction, could inadvertently discourage investment in renewables and electrification.

Ultimately, the test whether or not the transition succeeds is straightforward. Can clean energy, electrification and efficiency drive sustained declines in global fossil fuel consumption? That remains to be seen. But the rapid transitions already underway across technologies, sectors and countries suggest that today’s energy additions need not remain additions forever. The past is a poor guide to technologies that improve exponentially, and for that reason I’m optimistic about the prospects a genuine energy transition.

First some context

A German household installing an air-to-water heat pump in an existing home pays somewhere between €23,000 and €40,000. In the UK the same installation costs around €13,800. Why Is there such a huge gap? Marek Miara and I have just published a paper in the journal Energy & Buildings trying to understand where that difference might come from. TLDR: The answer is less about German heating engineering culture than I expected when we started writing the paper and potentially much more about policy design.

Our starting point was a specification-matched comparison done by my collaborator Christian Vering and colleagues at RWTH Aachen. They found that an air to water heat pump in a similar type of home was priced very differently: €29,719 net in Germany compared to €12,095 in the UK. Because Germany applies 19% VAT to heat pump installations while the UK has applied a zero rate since 2022, the price per unit installed before any grant is €35,366 compared to €12,095. In total, that is a gap of €23,271 or 192%. Note that this is not because we compare much larger units or different types of heat pumps (air to air heat pumps are much cheaper than air to water systems).

What drives the difference?

Getting a clear answer to a simple question turned out to be a complicated endeavour. Here are some pointers:

Compared to its neighbours German equipment prices contribute €2,000 to €4,000, partly because the German market is dominated by European manufacturers selling units at €9,000 to €18,000, while Asian brands with €3,000 to €8,500 units hold larger shares in the UK, France and the Netherlands.

An additional factor are the installation regulations. In Germany, VDI Guideline 4645 requires outdoor units to be installed on solid concrete foundations, which costs €976 to €2,700 where a wall mount allowable in many other countries costs under €500 (although from my experience many UK installs are also built on concrete foundations or similar).

Since January 2024, §14a of the Energiewirtschaftsgesetz requires a distribution board upgrade for almost every installation which adds another €1,800 to €4,000.

If we include VAT, which accounts for around a quarter of the gross gap, this explains perhaps up to 60% of the difference.

We expected labour costs to make a big difference but were surprised to find that they didn’t. German installers charge €60 to €70 an hour compared to €45 to €60 in the UK. Government-commissioned analysis finds mean labour costs per German installation to be €4,700 which is actually below the UK figure of €4,800. Instead of being a major factor that could explain the price difference labour is a small and genuinely uncertain contributor rather than a key driver, which is not what we expected.

The role of subsidies

So what makes up the remaining 40%? Under the Bundesförderung für effiziente Gebäude, households receive 30 to 70% of eligible costs. If an installer raises a €25,000 quote to €30,000 and the household receives a 50% subsidy rate, the household’s net cost rises by €2,500 while the installer receives €5,000. This means the pressure to negotiate a better deal is halved because of the way the policy is designed.

Could it be that the generous subsidies available may help inflate prices? For heat pump advocates like myself this is a somewhat uncomfortable question as I’m a firm believer in the need for strong policy support for quite some time. But we should be able to have honest conversations about policy effectiveness and sound design. The most effective policies deliver more uptake for every Euro spent.

Germany’s own policy history provides useful evidence for this. Mean total customer costs were around €17,000 through the capacity-based MAP scheme up to 2019. In 2020, when the scheme was redesigned to a percentage-based mechanism the funding share jumped from 11% to 40%. Shortly after that mean costs increased to around €40,000. Could there be other factors? Minimum efficiency requirements for eligible equipment did not change at the same time. Material costs over the same transition rose about 20% and helped drive up prices but installation costs roughly doubled over the same period.

German exceptionalism seems unlikely

Many will rightly point out that 2020 to 2023 was an unusual period. Pandemic, supply chain disruptions, commodity inflation, then a demand boom that allowed German installers to quote much higher. But the UK and Sweden experienced the same shocks and their installation costs rose only by 18% and 14% respectively, while Germany’s rose by about 100%.

Where does this leave German households?

With grants factored in how much do households pay? On the German gross price of €35,366, with €30,000 eligible and a 50% rate (the 30% base plus the 20% climate-speed bonus), the household pays €20,366. Even at the maximum 70%, they pay €14,366. A British household on a gross price of €12,095, taking the Boiler Upgrade Scheme’s flat £7,500, pays about €3,200. Four to six times less, under a scheme worth a fraction of Germany’s in nominal terms, and costing the British taxpayer far less per installation.

The UK’s Boiler Upgrade Scheme is not perfect by any means. It was slow to start, its budget was underspent, and £7,500 is a somewhat arbitrary number arrived at by a somewhat convoluted process. But its simple structure does something important to keep prices in check: every pound above the grant is a pound the household pays in full, so households behave like buyers, and installers price like sellers in a competitive market. France and Poland use fixed or income-differentiated amounts and are positioned at the cheaper end of the European distribution too.

The puzzle remains at least in part unsolved

Of course our attempted decomposition analysis and the conclusions we draw are not without uncertainty. Because we do not have a comprehensive dataset of heat pumps costs and their components our work should be treated as illustrative rather than representative. The large residual (unidentified factors) suggests substantial unexplained variation rather than strong proof of any particular mechanism. And the subsidy evidence is observational: a policy change that coincides with a price change is suggestive only and correlation does not necessarily mean causation.

Are there other countries we could look at? Austria is instructive here. Costs there are at around €32,000 which is close to the German level of heat pump prices. Austria also has generous and partly proportional support schemes in place. Of course this may well be a coincidence. However, without further analysis we won’t be able to say whether or not it is indicative of the effect of policy design.

If anything, we learned that understanding heat pump prices and what drives them is far from trivial and many of our preconceptions might be wrong. Critics may also argue that our analysis makes the case for undermining heat pump policies in Germany. That is not our intention. If anything, our work should serve to improve the effectiveness of policies and address price inflation thereby driving faster heat pump uptake.

The full peer-reviewed paper is available through open access in Energy & Buildings: doi.org/10.1016/j.enbuild.2026.117998

But it is also true that we are far from having transitioned away from fossil fuels. In fact, fossil fuel consumption has remained stubbornly high due to increased energy demand across the globe. However, for the first time since the Industrial Revolution, carbon emissions from fossil fuel combustion may finally plateau.

That’s good news for the climate. However, what is needed is not a plateau, but a steep drop of fossil fuel use to drive emission reductions. Whilst we have made significant inroads with electricity production, the share of clean energy in the industry, built environment and transport sectors has barely increased. Sceptics will point to global primary energy consumption data and claim that it will be impossible to replace all of that with renewable energy sources. Fortunately, we don’t have to do that.

The primary energy fallacy is a term coined eloquently by Paul Martin, a Canadian energy analyst. It is the idea that all primary energy from fossil fuels must be replaced by an equivalent amount of renewable energy. However, this is not necessary because more than two-thirds of primary energy is lost as waste heat during conversion processes. Think about a car engine: the majority of the fuel put into the car ends up as waste heat rather than moving the car. A fossil fuel boiler typically wastes 15% or so in converting fuel to usable heat. Electrification of those end uses on the other hand improves the efficiency dramatically. A heat pump will provide the same amount of heat as a gas boiler for 3-5 times less energy input because it moves and compresses pre-existing heat rather than burning a fuel. An electric car operates at almost 90% efficiency, making it multiple times more efficient than an internal combustion engine car which operates at about 20% efficiency.

What this means is that through electrification, we can dramatically improve the efficiency of our energy system and reduce energy being wasted. Recent analysis calculates that the improved energy efficiency of electrification alone (moving to technologies that use electricity as a source of energy such as electric vehicles and heat pumps) would result in a global reduction of final energy demand of around 40%, which is a significantly lower level of demand to be met through renewable energy production.

Another argument often made is that the big emitters in the world such as China aren’t committed to the energy transition, so why should other, much smaller countries be? The reality is that the amount of wind and solar power under construction in China is now nearly twice as much as the rest of the world combined. Yes, China also continues to build new coal plants, but the share of coal in the electricity mix is falling.

Too many argue that the energy transition is supposedly impossible instead of looking at how far we have already come. Yes, huge challenges remain, and we are currently not on track to meet climate commitments. There is a need to collectively rise to the sustainability challenge and the world remains resolutely on a deeply unsustainable path. But at the same time, we should find some comfort in the rapid transformation we have already seen in many countries around the world, and we should learn from what works and what has impact to shape our efforts going forward. We are at a turning point, and this should help us focus on making what seems impossible, possible. Together, we can lead the way in achieving the energy transition.

Innovation within the energy industry – supported by policy and by public and private investment – has driven and will continue to drive down costs of clean energy technologies. This trend is now unstoppable, according to the International Energy Agency. The challenges and hurdles present opportunities for innovation at the technical and political level. We will need to build out and optimise the use of our electricity grids. We can create the conditions in which electrification is the most economic option. And we must make sure that everyone can be part of and benefit from the transition. Unlike the German energy utilities in the 1990s, if we focus on how to get things done rather than on why they can’t be done, I believe we stand a chance of moving away from fossil fuels for good.

This article has previously posted on the Cambridge Institute for Sustainability Leadership (CISL) website as a blog.

For more than two decades I’ve been circling the same question: what does economics actually have to say about this? My first energy research project, as a student at the Wuppertal Institute in 2004, was a study for the German Environment Agency asking whether lignite really deserved its reputation. Germany’s utilities liked to call brown coal the country’s only subsidy-free source of power. We went looking for the hidden support, from tax breaks to the cost of resettling entire villages that stood in the path of the diggers, and the subsidy-free story fell apart fairly quickly. We put the quantifiable social costs of lignite power at up to 21.7 eurocents per kilowatt-hour. Even on the more cautious estimates, the hidden damage came to roughly as much as generating the electricity in the first place.

The lesson stuck. The price of energy on paper and its cost to society are not the same number, and often not even close. A few years ago a large study finally put figures on that gap for electricity worldwide, and that is what this issue is about.

What is an externality?

Markets exist to put a price on things. You want bread, the baker wants money, you settle on a number, and that number is meant to capture the cost of making the bread plus a margin.

Sometimes, though, an activity creates costs that fall on people who were never part of the deal. A factory tips its waste into a river: good for the factory and its customers, bad for everyone living downstream, none of whom were asked and none of whom were paid. Economists call this an externality, a side-effect landed on someone who had no part in the transaction.

The idea is more than a century old. I studied in more than 25 years ago when I was interested in how economics as a discipline approaches environmental degradation. Alfred Marshall set it out in the 1890s and Arthur Pigou sharpened it in the 1910s: where there’s an external cost, the market gets the price wrong, and a firm left to its own devices will make too much of the harmful stuff, because someone else is footing the bill for the harm.

Electricity is the textbook example. Burn coal and you get heat, which becomes power. You also get fine particles that lodge in people’s lungs, sulphur and nitrogen oxides that turn into smog and acid rain, and carbon dioxide that warms the planet for centuries. The plant owner sells the electricity, but it doesn’t pay for the heart attacks, the ruined crops or the flooding. Those costs are real. They’re also spread thinly across millions of people, many of them distant and poor, which makes them easy to overlook.

Putting a price on the invisible

For years these costs were treated as essentially unknowable, the kind of thing you can't read off a meter. But researchers have spent three decades chipping away at them, study by study and pollutant by pollutant (our small lignite paper was one early contribution), and in 2021 Benjamin Sovacool and colleagues pulled the whole literature together into something genuinely useful. They combined 139 studies and 704 separate estimates, 83 of those studies and 318 of the estimates dealing specifically with electricity, into a meta-study.

Their headline figure for electricity is an average external cost of roughly 7 cents per kilowatt-hour. Scale that up to global electricity production and the hidden costs reach about $11.6 trillion a year, or $3.8 trillion even on their more conservative central estimate.

The part that changes everything

This is what makes it a Bright Spots story rather than another entry in the climate-gloom file: the hidden costs are wildly uneven. They cluster around burning things. When Sovacool’s team broke the external cost down by source, the gap between fuels they uncovered was enormous.

Coal tops the table at a median external cost of about 8.1 cents per kilowatt-hour. Oil comes in around 6.6, natural gas at 2.9. After that the numbers collapse. Everything else sits below a cent: biomass at 0.8, solar PV at 0.7, nuclear at 0.4, solar thermal and wind at 0.2, hydro and geothermal at 0.1, which is to say next to nothing. Coal’s external costs run to roughly 12 times those of solar PV, 41 times those of wind, and 87 times those of geothermal.

We spend a lot of energy arguing about whether clean power is cheap enough to compete on its own. That argument leaves externalities out entirely. Put them back in, price coal and gas at their true cost to society, and it stops being a close contest.

There’s a revealing asymmetry in how this gets debated. People who dislike renewables work hard to draw attention to system costs, the grids, the storage, the back-up capacity you need to integrate wind and solar. Those costs are real and should be counted, no argument from me there.

The same voices go remarkably quiet about the externalities of fossil fuels, which are several times larger. You can’t demand a full accounting of every integration cost of a wind farm and then leave 8 cents of coal damage per kilowatt-hour off the books.

The bigger picture

Coal alone accounts for most of electricity’s hidden costs worldwide, something like 60% of them. So in narrow economic terms, the single most valuable thing we can do is stop burning it. Every coal plant that shuts, and every wind farm or solar array that takes its place, cancels a bill that would otherwise land on hospitals, farms and coastlines, one that never showed up on anyone’s electricity statement but was being paid all the same.

These costs are real but diffuse, which is precisely why policy finds them so easy to ignore. Adding the health costs would roughly double the price of coal in most large economies, and yet almost nowhere does the market do it. Twenty years on from our lignite study, Germany is still mining and burning brown coal, so I hold no illusion that pointing at the numbers is enough on its own.

Still, the clean option and the cheap option turn out to be the same option, once you count properly. The only reason that isn’t obvious is an accounting failure a century old. Carbon pricing, pollution standards, an end to fossil-fuel subsidies (all of them are ways of putting the hidden bill back onto the price tag) would let the market stop pretending burning things is free.

One thing to look out for

The same study points to something that cuts the other way: energy efficiency. Every kilowatt-hour we don’t use carries a positive external benefit of around 7.8 cents, roughly $312 billion a year globally. The cheapest and cleanest kilowatt-hour is still the one we never had to generate, and it barely features in the debate we actually have.

The history of lithium batteries is fascinating. In November, just a stone’s throw from my department, I listened to a lecture by, and had dinner with, Stan Whittingham, who built the first working lithium battery at Exxon in the 1970s and shared the Nobel Prize for it with Goodenough and Akira Yoshino. Stan described how those early cells came about. He couldn’t have imagined then what would follow.

For years the technology stayed niche: Sony commercialised the first lithium-ion cells in 1991 to power its camcorders. Then came laptops and mobile phones, and lithium-ion swept through consumer electronics.

But the first grid-scale lithium-ion battery was only connected to the grid in 2012: a single 5 MW project in Oregon. Today, 267,000 MW have been deployed globally. By 2030, forecasters expect 800,000 to 1,200,000 MW, depending on how seriously the world takes its climate commitments.

Every so often a technology stops being a pilot and scales rapidly. Solar did it. Onshore wind did it. Grid-scale batteries are doing it right now but and most people haven’t noticed yet.

To see how, I’ve pulled fresh data from Modo Energy (thank you for the access) covering the four biggest benchmarked battery markets: Great Britain, Texas (ERCOT), California (CAISO) and Australia’s National Electricity Market. Together they have gone from a rounding error to around 45 GW in six years. (A note on the numbers: these figures count grid-scale batteries above 6 MW, so the large domestic and behind-the-meter fleets found in places like Germany sit outside them. And “Australia” here means the National Electricity Market; Western Australia’s separate grid — the WEM — adds roughly another 1.5 GW.)

The build-out is exponential, not linear

Start with the headline chart. Installed battery power capacity across these four grids has grown from just under 5 GW in 2019 to around 45 GW today. Texas stands out: from roughly 100 MW at the end of 2019 to over 14 GW now — a more than hundredfold increase in six and a half years. California has the largest battery fleet, at almost 16 GW.

It isn’t just the level; it’s the slope. Taken together, the steepest stretch of the build-out is the most recent one — the combined curve is still bending upward, not flattening. Over the past year alone, every one of these markets added double- or triple-digit percentages to its fleet, and Australia more than doubled its capacity in 2025.

A caveat, though: this won’t be a smooth ride upward. In the most saturated markets — Australia and Texas — merchant revenues for short-duration batteries are already struggling to clear the hurdle for new investment, and the case increasingly depends on longer duration and contracted or structured revenue rather than pure energy arbitrage. High returns attract capacity; capacity competes those returns away; building slows until prices recover. High prices, in effect, cure high prices. Expect that cycle to repeat through the buildout. The direction of travel is unmistakably up — but the path will be bumpy, not a straight line.

Big enough to move the grid

Gigawatts of capacity on their own don’t tell a system operator much. The number that matters is the size of the battery fleet relative to the peak demand the grid has to meet — and on that measure, storage has gone from novelty to material in all four markets. California leads, with batteries now equivalent to about a third of peak demand.

The Financial Times published a version of this picture a few months ago; the climb has continued since — most sharply in Australia. There’s a second way to cut the data that I find even more revealing. Rebase each line to the quarter in which batteries first crossed 1% of peak load, and you can compare how fast each market scaled regardless of when it started. California didn’t just go highest — it went fastest. Great Britain took almost twice as long to reach the same share.

More batteries and longer batteries

The megawatt numbers mask a second shift: the batteries being built can increasingly discharge for longer. Texas shows the split between one- and two-hour systems cleanly. Capacity for two-hour delivery has gone from under a third of the fleet to the majority, overtaking one-hour outright in mid-2025.

Why? Value shifts to systems that can move energy across more of the day. As short-duration markets fill up, developers build longer. Australia’s one-hour fleet has flatlined since 2024; California, shaped by its four-hour capacity rules, barely builds anything shorter.

A quick word on the economics

The build-out has a side-effect. As fleets grow, the price spreads batteries feed on get arbitraged away — more storage chasing the same volatility means thinner margins per megawatt. Australia shows it most vividly, where a wave of new capacity in late 2025 collided head-on with falling revenues.

In twelve months, Australian battery capacity rose about 115% while the per-megawatt revenue index fell roughly 64%. That’s not a market failing but market maturing. Falling revenue-per-MW is exactly what you’d expect as storage stops being scarce; it’s a sign the technology is doing its job. But it’s a side-plot. The main story is still the build-out.

What’s coming next

The pipeline says the fastest chapter is still ahead — though not without friction. Great Britain’s fleet reached 7.2 GW at the start of 2026, and the project pipeline now stands at around 30 GW. But Modo’s latest buildout report is blunt about the bottleneck: how much gets built, and how fast, depends almost entirely on the connection queue. Its base case is roughly 20 GW by the end of 2029 — up to 22.8 GW if connections reform goes smoothly, closer to 17 GW if delays bite. The direction is up and to the right; the open question is the slope.

And the character of the build is changing. Batteries of 200 MW or larger made up 43% of the 2025 pipeline and 74% of 2026’s. The era of small, standalone units is giving way to grid-scale megaprojects.

Put the three threads together — more capacity, longer duration, bigger units — and the direction is unmistakable. Grid-scale storage is quietly rewiring how the electricity system works.

An earlier version of this article was reviewed extensively by Ed Porter and colleagues at Modo Energy who kindly gave me access to their data.

Today the European Commission launched its Electrification Action Plan. The headline number, still a bracketed placeholder when I analysed the leaked draft last week, has now been filled in: electricity is to meet 46% of the EU’s final energy consumption by 2040. Getting there means to speed up 9x compared to the annual increase of the electrification rate since 1990.

That would double today’s share. Electricity currently meets just 23% of final energy consumption in the EU, and it has sat there for a decade while China, Korea and Japan have moved past 30% and continue to pull away. The stagnation is not an abstract statistic. It is the reason Europe keeps getting caught out by fossil fuel price shocks, most recently when the Iran war sent oil and gas prices soaring. The EU still imports more than 80% of its gas and over 90% of its oil, and the Commission’s own analysis puts a price on that dependency: an accelerated transition with electrification at its core could cut the fossil fuel import bill by up to €260 billion per year by 2040 and reduce gas imports by more than 70%.

The leaked draft mentioned a target “enshrined in EU legislation”. The adopted text calls it an indicative target that will be “subject to an impact assessment” as part of the Energy Union package due in the fourth quarter. The number survived inter-service consultation but the legal anchoring, for now, did not.

A target on its own delivers nothing, of course. What makes this plan more credible than previous announcements is that it goes after the biggest barrier directly. Electricity costs companies in Europe almost three times as much as gas, and in some member states the ratio is closer to four. As I set out in my analysis of the draft, the plan pairs the headline target with a KPI to bring electricity-to-gas price ratios down to 2.5 for households and 2 for industry by 2030, backed by a legislative proposal on electricity bills adopted alongside the plan. That proposal now carries the taxation fix too: a provision addressing the tax differential between electricity and gas, built on the principle that electricity should not be taxed more than gas. In most member states electricity currently carries the heavier tax burden. Reversing that signal is where the plan touches the real economics rather than staying at the level of aspiration.

Most of the rest survived as leaked, but not all of it. The heat pump KPI in the draft was to double installation rates by 2030. The final version commits to raising them from 2.4 million a year to around 4 million, a two-thirds increase. The draft’s EU framework for reducing VAT on electrification assets, from heat pumps to industrial equipment, has shrunk to a gentle reminder that member states may already use the options in the VAT Directive, with the framework itself now confined to corporate fleets.

Other elements were strengthened. The storage ambition of 200 GW by 2030 (up from around 55 GW today) gained a 500 GW milestone for 2040, and the Commission will assess the EU’s long-duration flexibility needs out to 2050. The plan also commits, for the first time, to publishing the uncomfortable arithmetic on fossil fuel subsidies: €97 billion in 2024, of which more than half has no end date before 2030. Vehicle-to-grid requirements for new EVs arrive by the end of 2027. And the commitment to explore a clean heat market mechanism obliging manufacturers to grow their share of clean heat sales survived intact, an instrument colleagues and I have long advocated through academic analysis and a handbook for policymakers. For the full detail, including what is genuinely new and what is being re-announced, my analysis from last week still stands.

Watch this space

The Energy Union package in the fourth quarter is the moment of truth. That is when the impact assessment lands and we learn whether 46% hardens into law or stays a number in a communication. Between now and then, the fight will be over the electricity bills proposal, because if taxes and levies are not rebalanced, the target will stay out of reach no matter how often it is repeated.

The direction is right and the case has never been stronger. Now comes the delivery.

I’m grateful for the enthusiasm. But allow me to point out three ironies.

Irony one: air conditioners are heat pumps in disguise

Many of the people now cheering me on have spent years telling us that heat pumps don’t work. Here is the thing: an air conditioner is a heat pump. It uses the same refrigeration cycle, the same compressor, the same refrigerant, and the same physics. The only difference is that a heating heat pump is designed to move heat into the building, while an air conditioner moves heat out. Many modern systems can do both at the flick of a switch. More on this later.

If you believe air conditioning works (and it does) then you already believe heat pumps work. Because they are the same machine and they work very well in all building types – more on this here.

Irony two: you can’t deny the heat and demand the cooling

For years I was told that climate change is a hoax and that British summers aren’t getting hotter. Now some of the same voices demand to know why on earth Britain never installed air conditioning. You cannot have it both ways.

The rising demand for cooling is driven by the warming climate. Europe has warmed by around 2.5°C since pre-industrial times (the fastest-warming continent on Earth), and we simply have far more hot and also hotter days than we used to. The reason Britain never bothered with air conditioning is the same reason it is now scrambling to think about it: for most of living memory it was cool enough that only about 4% of English homes have installed mechanical cooling systems. In Greece it is 70–80%, in Italy nearly 60%, and in the United States over 90%.

Irony three: affordable cooling depends on the things this crowd loves to attack

Running air conditioning affordably depends on exactly the things the anti-net-zero crowd likes to campaign against: cheap solar power, a modernised grid, and efficient electric technology.

Air conditioning will require more electricity. The good news is that solar generation peaks at almost precisely the moment cooling demand is high: in the hot, bright middle of the day. The sunshine that makes a home unbearably hot is the same sunshine that can provide cheap electricity to cool it. Fight cheap solar and a smarter grid, and you are fighting the cheapest way to keep cool.

Where there is not a good match of solar and cooling is during hot nights but in Northern Europe those are still relatively rare and the cooling demand is much lower than during the daytime. And increasingly batteries can help with this too.

What I actually said about cooling

In case it got lost in the noise: my argument is that rising heat is a genuine risk to health and productivity, and that much of Europe’s building stock was simply never designed for these temperatures. Carbon Brief notes that cooling averted roughly 190,000 heat deaths a year between 2019 and 2021, and Europe’s heat-death rate is several times that of the better-cooled United States. The June 2026 heatwave alone is estimated to have killed more than 20,000 people across Europe.

Passive measures should come first wherever possible, because they avoid the need for excessive mechanical cooling in the first place: shading, shutters, insulation, better design. This is not abstract for me. We have just fitted 13 external shutters to our windows for exactly this reason: external shading is far more effective than internal blinds and we recently supported a mature tree in our garden to stop it from falling, because shade is cooling you don’t have to pay for.

But in many cases even after passive measures have been considered and implemented, active cooling will still be needed. It should be efficient, clean and affordable. The alternative to planning for this is not that there will be no air conditioning. Instead the problem is millions of cheap, inefficient units bought in a panic. We already see that mobile air conditioning units in the UK sold out.

Cooling can decarbonise heating

There is an opportunity too though: The growing demand for cooling is potentially a route we have to decarbonise heating.

A reversible air-to-air heat pump cools in summer and heats in winter from the same unit. This is not exotic. Most heat pumps in Scandinavia are air-to-air systems; the same is true across much of the United States and many other parts of the world. Once a household has installed a unit to survive August, it also owns a low-carbon heating system for January, if it is an air-to-air heat pump.

Ground-source heat pumps are naturally reversible and exceptionally efficient for cooling, using the stable temperature of the ground as a heat sink in summer and a heat source in winter, and, unlike individual air conditioners, they don’t dump waste heat into the street.

Manufacturers are also now bringing out combined units that pair air-to-water heating for radiators, underfloor heating and hot water with air-to-air cooling in a single system. That matters, because it lets homes keep the wet heating they already have while adding efficient cooling, rather than choosing between the two. For those not familiar with European radiator heating systems: many homes use a wet heating system rather than ducted forced air heating technology.

There are real constraints on the role air-to-air can play in the UK and other European countries. Many of our homes are traditionally heated with radiators, our building stock is older and more complex with many heritage buildings, our planning rules are not yet set up for it. But the direction is clear. As more and more countries move into a dual mode of needing both heating and cooling, we have to stop discussing the heating transition and the rise of cooling as if they were separate conversations. They are the same conversation.

And it doesn’t have to mean a unit in every window

There is also a smarter way to cool a city than millions of individual boxes each pumping hot air onto the pavement. If everyone in Paris installed their own air conditioner, the city’s streets could warm by more than 2°C cooling indoors by making outdoors worse.

Paris offers the alternative. The city runs one of the largest district cooling networks in the world, operated by Fraîcheur de Paris, drawing cold water from the Seine to chill the Louvre, hospitals, offices and stations through a network of underground pipes and it plans to expand it dramatically over the coming years. Cooling is generated centrally and efficiently, often using off-peak and low-carbon electricity, and the waste heat is managed rather than dumped into the streets. It is the cooling equivalent of district heating, and for dense cities it is a serious option.

None of this is anti-net-zero. It is net zero. Cheap clean power, efficient machines, passive design first, and a bit of joined-up thinking between how we heat and how we cool. If that makes me a hero of the anti-net-zero crowd, I’ll take the applause. I suspect they’re clapping for the wrong reasons.

For ten years, the share of electricity in Europe’s final energy consumption has barely moved. It sits at 23%, roughly where it was 10 years ago. China, Korea and Japan are already past 30% and pulling away fast. The lack of progress with electrification explains why Europe keeps getting caught out by fossil fuel price volatility caused by geopolitical shocks. The now leaked draft of the Commission’s Electrification Action Plan (COM(2026) 595), due for adoption at the end of the year, is the first serious attempt to catch up.

I’ve now read the whole thing, and the framing is broadly right. The Commission positions electrification as the route to lower prices, more security and a stronger industrial base, not just a cost to be managed. In this piece I set out what is genuinely new, what is being re-announced, and the one or two ideas that could actually move the needle.

The key barrier the plan gets right

Let’s start with the number that matters the most. Electricity in Europe costs companies almost three times as much as gas. Only in Finland and Sweden does the electricity-to-gas ratio drop below two and in some countries it is closer to four. As long as that gap persists, the economics of switching to a heat pump or an electric process remain challenging. This shows in the data: For example, Member States with the lowest price ratios record triple the heat pump sales of those where the ratio is above three.

So the headline commitment is a KPI to bring national electricity-to-gas ratios down to 2.5 for households and 2 for industry by 2030, backed by a legislative proposal on “future-proofing electricity bills” landing alongside the plan. I went through that proposal in detail when it leaked separately a few weeks ago (my write-up is here), and it is worth recapping what it does, because it is the machinery that would make this KPI real.

The proposal treats the whole bill rather than the wholesale price alone. On network charges, which already account for between a quarter and a third of the average household bill and will only grow as grid investment climbs toward €100 billion a year, it rewrites the rules so tariffs reward flexibility and demand shifting, with ACER benchmarking transmission operators and national regulators doing the same for Europe’s 2,600-plus distribution operators. It sets a smart meter floor of 50% of customers by 2030 and 65% by 2033, which matters because time-of-use tariffs do very little without them (Germany is still below 5%). The cleverest move is on tax. By stating that electricity may be taxed no higher than gas, and anchoring that principle in the Electricity Regulation under Article 194, the Commission shifts the decision to qualified majority voting and sidesteps the unanimity that has kept the Energy Taxation Directive frozen for years. In several countries electricity currently carries the heavier tax burden, so this reverses the signal at source.

That is where the plan either touches the real economics or stays at the level of aspiration. Taxes and levies fall more heavily on electricity than gas in most Member States, and fossil fuel subsidies are still distorting the choice. If the Commission follows through here, it could be a game-changer.

A big idea: a clean heat market mechanism

The heat pump section is where the plan really surprised me. There is a KPI to double the installation rate by 2030 compared to 2025. We have seen similar announcements before as part of the REPowerEU Strategy.

What caught my eye is not so much the KPI but the key mechanism proposed: By 2027 the Commission will explore a clean heat market mechanism, obliging manufacturers to grow their share of clean heat sales and pushing installers toward heat pumps while narrowing the upfront cost gap. This is an instrument I and colleagues have long promoted through detailed academic analysis and a handbook for policy makers.

If that moves from “explore” to “propose”, it could fundamentally reshape the market in a way subsidies alone never have. A supply-side obligation changes the incentives of the people actually selling heating systems, rather than waiting for households to make the leap on their own. It is the kind of instrument that has worked for other technologies such as EVs, renewables, and energy efficient appliances. Seeing it named so prominently is a step I did not expect.

Two smaller heat pump measures are also worth flagging because they solve specific but important problems. The Commission will look at mandating heat pumps in public buildings through procurement rules, which creates guaranteed demand and helps the value chain. And it will support national online comparison tools so households can get a transparent, comparable “heat pump quote” from different brands and installers. Anyone who has watched quote inflation kill an installation will understand why price transparency matters as much as any subsidy.

Green VAT, flexibility and other parts that make sense

A few other measures are new and useful. As part of the Green VAT initiative from the Clean Industrial Deal, the Commission will present a framework for Member States to cut VAT on electrification assets: EVs, heat pumps, household batteries and industrial electrification. Tax is one of the fastest levers a government has, and only six Member States currently apply a lower VAT rate to heat pumps than to gas boilers.

On flexibility, the plan raises the storage ambition to 200 GW by 2030, up from around 55 GW today. Alongside it, the Commission will introduce vehicle-to-grid requirements for new EVs by the end of 2027, including the interoperability standards that make V2G actually work, and will assess smart charging by default in EV supply contracts by mid-2027. Both are sensible and overdue. There is also a commitment to develop, by 2027, a methodology for assessing flexibility potential across industrial processes and data centres, which fills an important and longstanding knowledge gap.

The parts that are being re-announced

Not everything here is new. Much of the industry section restates commitments already on the table. The ETS Directive revision that channels more national ETS revenue into industrial decarbonisation, the €100 billion Industrial Decarbonisation Bank with its €30 billion ETS Investment Booster, and the second industrial heat auction under the Innovation Fund in 2026 were all announced before. I am of course glad industrial electrification sits at the centre of the plan, since it remains the most underappreciated lever we have, but there is less new policy reform in this section than the length suggests.

Some of the EV measures are similarly incremental. BEVs have climbed to a record 20.4% of new sales so far in 2026 and saved an estimated €4.1 billion in avoided oil imports last year, so the trajectory is strong. The plan promises a review of the Clean Vehicles Directive by Q4 2027 and a recommendation on demand-side incentives by Q4 2026. Helpful, but this is fine-tuning rather than a new direction.

The global ambition

Internationally, the plan builds on Electrify Now, the platform launched on 23 June 2026 with the COP presidencies and the IEA. The Commission wants to assemble a coalition committing to global clean electrification, anchored in the tripling of renewables and a new objective to reach 35% electrification worldwide by 2035, which the COP31 presidency is championing (I covered this elsewhere). Europe positioning itself as the first “electro-continent” is a good story, and export markets for European heat pumps, batteries and grid technology are the prize if the manufacturing base holds up.

The bottom line

The diagnosis in this plan is excellent and directionally correct. The electrification target itself is still a bracket, [X]% of final energy by 2040, and many of the verbs are “explore”, “assess” and “will support”. The effectiveness of the plan will depend on whether ambition translates into action on the ground.

What I will be watching between now and adoption in Q4 2026 is two things: whether the target lands at a level that forces the pace rather than ratifying the status quo, and whether the electricity bills proposal genuinely rebalances the taxes and levies that keep electricity expensive. Get those right and this is the moment Europe’s electrification rate starts climbing again.

It starts in February 2023, when Germany’s biggest tabloid, BILD, put a leaked draft of the revised Buildings Energy Act on its front page and gave it a name: the Heizungshammer, the heating hammer. Then came one of the most divisive political battles in recent German history, and it ended with the policy completely gutted.

But before we get to the coverage, it helps to remember what the law actually said. In principle, the Buildings Energy Act (GEG) was nothing new. It had been in place since 2020, introduced by a CDU-led government (this is the conservative party). The 2023 reform by the coalition of social democrats, greens and liberals proposed that all new heating systems installed from 1 January 2024 should run on at least 65% renewable energy. Existing boilers could remain. Nobody was going to rip a working heating system out of anyone’s cellar, and there would have been a wide menu of options for households to comply, backed by subsidies that were later increased even further, particularly for lower-income households. None of this should have come as a surprise either: the 65% target had been in the coalition agreement all along.

That’s not the way it landed. In the first months of 2023, the German press produced a few hundred articles a month on the heating law. By June that number had risen to over 7,500. One of the biggest stories in the country for much of the year was a technical amendment to building regulations.

For the paper we analysed all 333 articles BILD published on the law between January 2023 and March 2024, using computational text analysis to chart what the coverage was actually about.

The most obvious trend is personalisation. The most common word in the whole corpus is, by a long way, Habeck. Robert Habeck, then economy minister, was mentioned 839 times, an average of more than twice per article, ahead of the law itself, the coalition and every institution involved. The name was even fused directly onto the BILD-invented name for the heating law (”Habecks Heizungshammer” and variants, 44 times). A draft law from two ministries, agreed by three parties and passed by parliament, became in the telling one man’s personal crusade.

Then there’s the money. “Euro” and “costs” are high on the frequency list, and false claims about forced retrofits and total bans on boilers circulated widely, often with little resemblance to the draft. At one point figures of 300,000 euros for a single house were doing the rounds. Beneath both ran an ideological frame: the law as a project of detached elites imposing their worldview on everyday people. About 660 articles in the German press linked the legislation to the word “ideology”.

The thing I notice most is how far the framing went. The term BILD coined appeared more than 250 times in its own pages, and in more than 1,100 articles elsewhere. “Heizungsverbot” (heating ban), for a law that banned no existing heating, appeared in 800. References to public Verunsicherung, uncertainty, linked to the law surfaced in more than 2,500. The AfD launched a campaign website to “stop the heating hammer”. Weeks later the CDU launched its own version.

What made the spread possible is that almost nobody went back to the source. Of all the outlets covering the story in those early months, only a handful, Klimareporter and Volksverpetzer among them, actually worked through the leaked draft to check what it said. None of the big papers did. Much of the debate played out on the surface, arguing over BILD’s version of the law rather than the law itself. Part of the explanation is that buildings had been a neglected beat for years. Wind and nuclear got the headlines; heating was the boring bit. I have worked on buildings policy for more than two decades and it was always the stepchild of the energy debate, so when it suddenly became the biggest story in the country, few journalists knew the material well enough to push back.

And it produced real legislative results. The 65% requirement was first narrowed to new buildings in new developments, and then tied to municipal heat planning, with deadlines that extend to 2028, which for most rural municipalities means a delay of many years. Biomass limits, which had been in previous drafts for sound air quality reasons, were dropped. The planned tightening of efficiency standards for new buildings, due to take effect in 2025, was scrapped altogether at the “building summit” in September 2023. Germany then overturned its long-standing support for minimum energy performance standards in the EU buildings directive and opposed the Commission’s proposal to phase out stand-alone fossil boilers, which parts of the press had already dubbed a “European heating hammer”. This spring the present CDU-led government said it would do away with the 65% rule altogether. A newspaper campaign, in other words, worked its way through the debate into the statute book, and from there into Germany’s negotiating position in Brussels.

I don’t want to pretend that the original law was perfect politically. The government has itself acknowledged two mistakes. There was no communication strategy worth the name, so the first time most people heard of the policy was through a hostile frame, with no counter-narrative in place, and none coming for weeks. And the money was missing. The draft regulated first and left the subsidy package to be worked out later, which meant that for months there was an obligation on the table with no answer to the obvious question of who pays, and what it all means for the pensioner in a poorly insulated house. The answers came eventually. By then the debate was fully ablaze, and millions of people already believed their boiler was about to be confiscated.

Government mistakes alone don’t explain what happened, though. The coverage ran alongside intense lobbying against the law, including from parts of the gas industry, whose Zukunft Gas initiative was promoting hydrogen for heating at exactly this time. Sowing fear, uncertainty and doubt to stall regulation is an old playbook, familiar from tobacco and oil, and here it found a very large megaphone.

So what’s the good news? Two things, perhaps. First, the heat pump market speaks for itself. Once the subsidy scheme was in place German sales bounced back strongly, suggesting the underlying economics survived the political fire. People react to what heating actually costs them, not headlines, and the subsidies altered that equation. Second, other countries have been looking on. The Danish and Dutch approaches, built on district heating rollouts and long lead times communicated years ahead, have faced nothing comparable. Policy design and communication cannot be separated, and Germany has learned this the hard way.

The paper, “Heating up the headlines: How tabloid framing reshaped Germany’s Buildings Energy Act”, is published open access in Energy Research & Social Science: https://doi.org/10.1016/j.erss.2026.104832

Twenty years ago renewables were a quarter or so of annual additions, easy to wave off as a niche. That line has climbed steadily ever since. Most of this is now solar and wind which together made up 96.8% of net renewable additions in 2025, largely driven by the economics and falling costs.

But you will notice that even though record renewable energy capacity additions were made in 2025 the share of renewables of new capacity dipped slightly: Renewables took 86% of all new capacity last year but it was 92% the year before. Fossil additions increased compared to the year before. For example, China, alongside its vast clean build-out, added more than 100 GW of non-renewable capacity, roughly four-fifths of it coal. However, the utlilisation rate of coal plants in China is dropping which means whilst more capacity is being added overall coal generation has started to decline.

Was 2025 fast enough?

The harder question is whether a record year is actually a fast enough year? The answer is yes if we use the tripling renewables capacity as a benchmark. At COP28 the world agreed to triple renewable capacity to more than 11 terawatts by 2030. Meeting the COP28 pledge requires sustaining growth at roughly 15-17% per year globally, every year to 2030. In 2025 capacity growth was 18%.

Who leads?

In terms of sheer scale of renewable capacity additions the answer is clear: China. It added 440 GW on its own. Europe installed 77 GW, North America around 55 GW. Africa had its best year on record at 11 GW, with most of it in Ethiopia, South Africa and Egypt. That is worth celebrating given how slow historic deployment on the continent has been.

What the absolute figures miss, though, is how fast some of the late adopters are now moving. Saudi Arabia drove a 29% jump in Middle East renewable capacity last year, a figure that would have looked fanciful a decade ago. Africa grew 16%, the quickest on record so far. Being near the bottom of that chart in 2025 tells you where a country stands today, not where it is heading, and the places lagging historically are now speeding up deployment substantially.

Capacity is not generation

At this point a careful reader is already objecting: Capacity is not generation. A 49% share of capacity does not mean renewables are producing 49% of our electricity. If we want to know how the transition is really going, capacity is the wrong number to use. Generation is what really counts.

Measured by generation, non-fossil sources (renewables plus nuclear) supplied 43% of the world’s electricity in 2025, up from below 33% two decades earlier. Renewables alone reached 33%. Wind and solar together hit 17% and their increase is particularly impressive because it was close to zero at the start of the century. In absolute terms wind and solar generated 5,490 terawatt-hours in 2025, up 18% in a single year.

Fossil generation fell in 2025, the first decline since the pandemic slump of 2020. It is a small dip and it may not hold. But for the first year in memory, all that new clean generation did more than accommodate rising demand: It started, just barely, to push fossil output down. If that turns into a trend rather than a blip it means legacy fossil generation declines in absolute terms not just as a share of generation. Only then will we see a continuous and sustained reduction in power sector emissions.

So what is holding the pace back? Not the technology, and increasingly not the cost. Solar modules and wind turbines have kept getting cheaper. The binding constraints today are grids, permitting, and storage. Transmission lines take the better part of a decade to plan and build. Connection queues stretch for years in many markets. There is a huge pipeline of grid-scale storage capacity waiting to be connected.

Some claim that this shows the failure of the expansion of renewables. I’d argue it is a feature of the transition not a bug - we just did not build grids out fast enough as we underestimated the speed at which particularly wind and solar would expand. Speeding up requires regulatory, market design and policy reform around the world. When you look closely, many of these issues are solvable and are already being solved today.

It is important to point out that electricity is only 21% of global final energy consumption. If we want to reduce fossil fuel consumption in transport, industry and buildings we will have to use more clean electricity in those sectors. Electrification will have to increase substantially in the future as I explain in my recent Substack article. This is why the next decade will not only be about increasing the share of renewables in the power sector but also about how to use more electricity instead of fossil fuels in the end-use sectors.

EU final energy consumption in 2024 was 854 million tonnes of oil equivalent. Without the efficiency improvements made since 2000 it would have been closer to 1,103 Mtoe. The difference, 249 Mtoe, is 29% of what the EU actually used last year. The data comes out of the Odyssee-MURE decomposition., a project that has been tracking energy demand for more than three decades and to which I contributed 10 years ago. Importantly that is energy nobody had to generate, import or pay for. It built up slowly, over twenty-five years, across 27 countries, in warmer buildings, more efficient factories, better appliances, the switch to LED lighting and steadily more efficient transport. At any given price Europe now pays around a third less than it would have without those gains.

This shield holds whichever direction the next shock arrives from. If anything, the figure undersells the benefit, because it counts only the energy saved and not the power stations, pipelines and grid connections nobody had to build. It is one of the most effective protection measures against future energy crisis: the energy not consumed cannot be withheld, weaponised, or made more expensive by a supplier you do not control.

The crises of the past few years have all produced the same response. Russia’s invasion of Ukraine, the shock of Covid, the latest trouble in the Middle East, each set off a scramble for alternative supplies, but none of them sufficiently addressed the root cause. Europe still imports too much fossil fuel, and it wastes far too much energy, a habit that cheap fuel let it put off dealing with for decades.

Electrofficiency: Why electrification shrinks the problem

The good news is that the potential for further efficiency improvements is vast. Electrification is efficiency because electrifying something usually cuts the energy it needs, often by a lot. A heat pump delivers three or four units of heat for every unit of electricity it draws. An electric car turns most of its energy into motion, where a petrol engine throws away 80% of it as heat before the wheels even turn. So when people line up today’s fossil system against a future clean one, unit for unit, they tend to overstate how much new generation Europe has to build, because each clean kilowatt-hour does more useful work. Part of the transition is a fuel switch, and part of it is the problem getting smaller through system compression and electrofficiency.

Uneven fossil fuel phase-out across sectors

Electricity is where Europe has seen real success when if comes to reducing import dependency. The fossil share of EU power generation has dropped from 53% in 2005 to below 30% last year, on Ember’s latest figures, which is real progress and not a statistical artefact. The catch, in the Eurostat balances, is that electricity is still only 23% of final energy use. Transport is 92% fossil-fuelled. Buildings are at 71%, industry at 61%. Away from the power sector, most of the work is still in front of us.

Three levers, one system

The mistake I keep seeing is that efficiency, electrification and clean power get treated as rival claims, argued over by different directorates and different lobbies. They are really one system: Efficiency lowers the demand that has to be met, which makes the grid smaller and cheaper and shrinks the import bill at every price. Electrification then shifts what is left onto electricity, the one energy carrier Europe can produce at home from clean sources. Renewables and nuclear are what make that electricity clean in the first place. Weaken any one of them and the other two get harder and dearer, which is the practical reason the usual political trade-offs between them tend to backfire.

Take one ordinary example. A heat pump in a well-insulated home on clean electricity cuts the energy the building needs, replaces the gas boiler, and runs on a grid that keeps getting cleaner. That single device also pulls the household out of the gas import chain for good, and as a side effect keeps it cooler through the sort of summer Europe now gets most years.

Industry has large potential to improve

The part that gets least attention is industry, but the potential is vast. Electric motors run more than half of the EU’s entire electricity use, and a large share of them are older and less efficient than current standards allow. These are the pumps, fans and compressors humming in the background of every factory. Upgrading releases clean power for all the things we do want to electrify. Our recent analysis suggests up to 90% of industrial energy use could be electrified in the long-term.

What is adds up to by 2040

Stack the three levers together and the Commission’s own 2040 modelling gets interesting. Gas demand falls from 340 billion cubic metres in 2025 to about 101 by 2040, roughly a 70% cut, shared between efficiency, electrification and renewable supply. Oil falls further in proportion, from 350 to 98 Mtoe, a 72% drop, mostly because road transport goes electric. Across those fifteen years the avoided import bill works out somewhere between €1.4 and €2 trillion, depending where prices settle. That is money currently leaving the continent every year that would instead stay in it.

There is also a human story here. Europeans spend roughly 90% of their lives indoors, and on the Healthy Buildings Barometer more than 160 million of them live with damp, mould, cold, poor light or noise at home. Renovating those buildings cuts energy use and improves health in the same act, and the wider gains, in productivity and avoided healthcare costs, run into the hundreds of billions by mid-century. Efficient buildings also flatten the daily demand peak, which recent modelling finds lowers the cost of the whole system and not just the bill of the household inside.

The risk now

None of this needs a new strategy. The targets exist, the directives exist, the supply chains are being built, and in efficiency and electrification that industrial base is an asset Europe would be foolish to let drift offshore. What has been missing is the steadiness to see the plan through, and the policy lurches of recent years have left investors wary and made the people who train installers think twice. Most of the road to energy security is just a matter of delivering what is already agreed, and paying for it.

If there is a weak link in all this, it is the grid. Clean generation and new electric demand are both arriving faster than the wires and substations to carry them, and in several member states the connection queue has become a genuine brake on industrial electrification. Efficiency helps here as well, by holding the peak down so that less new network has to be built, but the honest answer is that grid investment and permitting still need to move up a gear to keep pace with the rest.

This article is based on a policy brief published 16 June 2026.

The reason why electrification has risen up the global energy agenda is that because electricity is less than a quarter of global final energy use a clean grid only matters a if transport, heating and industry increasingly draw their energy from it rather than rely on coal, oil and gas burned on site. I cover similar ground to what is in this week’s Bright Spots in a forthcoming Aurora energy podcast.

Of critical importance is this: Electrification is not just a fuel switch from molecules to electrons. In short, electrification is efficiency. This is because electric machines are far more efficient than the combustion they replace. An electric motor turns around 90% of its input into useful work against a combustion engine’s 25%, and a heat pump delivers three to five units of heat per unit of electricity. So every point of electrification shrinks the total amount of energy the system actually needs. I call this “Electrofficiency”.

This is the direct counterpart to the Primary Energy Fallacy: judging the transition against today's fossil throughput overstates the task by a wide margin, because more than two thirds of that energy is simply wasted as heat and this waste heat can be reduced dramatically in a highly electrified world running on clean electricity.

That is what makes the electricity share of final energy consumption, the share of the energy we use that arrives as electrons, one of the most important measures of how the transition is really going.

Electricity supplies about a fifth of global final energy consumption today (21% in 2024). The share has risen by roughly 0.3 percentage points/year since 2010. But a single global economy-wide number obscures the structure underneath. At an aggregate level the world moves far too slowly to meet climate commitments. But sectors and countries move at very different speeds.

Buildings and industry rise, transport stagnates

Let’s split final energy use into its three big sectors using IEA data.

• Buildings: In 1971 electricity met about 11% of the energy used in the world's homes, offices, shops and public buildings. Today it meets 36%.

• Industry is a bit of a surprise. The conventional wisdom is that industry is stuck, too hard to electrify. The global data says otherwise. Electricity's share of industrial energy has more than doubled, from 15% in 1971 to 30% today, and it has been rising fastest in the last decade. Electric alternatives exist and the potential for industrial electrification of energy use is much higher than people think, potentially as much as 90% or more in the long run as I wrote in a previous article.

• Transport is the outlier at 1%, essentially unchanged since 1990, because liquid fuels held an advantage that only collapsed once batteries became cheap more recently. This is likely to pick up speed now with electric car sales growing by 20% globally exceeding 20 million in 2025, meaning one-quarter of all new cars sold were electric last year. Around 70% all electric cars sold last year were battery electric vehicles rather than hybrids and their share is increasing year by year.

It is clear that the global average cannot accelerate on one sector alone. Further electrification has to come from all sectors at the same time.

Almost everyone is behind on 2035

On current policies the IEA projects it reaches just 25% by 2035 or 4 percentage points of additional electrification in a decade. The COP31 “35 by 35” target requires reaching 35% over the same period which is equivalent to an increase of around 1.3 percentage points/year or roughly three times the pace of the IEA’s Current Policies trajectory and four times the historical trend.

The COP31 target is very similar to the IEA’s Net Zero Emissions by 2050 scenario which implies an electricity share of a third of final consumption by 2035 and 55% by 2050. Other 1.5 °C-aligned scenarios span a similar range at mid-century, from approximately 54% in IRENA's 1.5 °C pathway to as much as 70% in the Energy Transitions Commission's analysis. The three climate-aligned trajectories converge on roughly one-third around 2035 before diverging, which indicates broad agreement on the near-term pace of electrification and genuine uncertainty only about how far direct electrification extends into heavy industry, aviation and shipping by mid-century.

Where to countries sit in respect to the 2035 ambition level called for by the COP31 hosts? Line the major economies up and most have substantial room for improvement. Japan is closest at 30.6%, only 2.4 percentage points short. China is at 29.3%, and notably is closing its gap faster than anyone because its additions are concentrated in fast-moving end uses. The United States and the EU are stuck around 22%, some 11 percentage points away from the IEA 2035 milestone, held back by continuous use of gas in heating and a slow industrial transition. Brazil and India trail at 20.4% and 19.1%. Whilst the target is physically achievable, of the large economies only China is currently moving at the pace required to meet the 35% target.

An important caveat: Which metric to use to track progress on electrification and how define any milestones is not trivial as my long-standing collaborator Michael Liebreich discusses in his excellent essay published earlier this week. This is because of the Electrofficiency effect I mentioned above: electric technologies use less energy for the same amount of useful energy output and when they replace one percent of fossil fuels their share does not increase by one percent but in case of electric cars and heat pumps only by 0.25-0.3 percent. More in Michael’s essay.

A further complication is that rising electricity demand from new loads such as data centres help bring up the electrification rate without displacing fossil fuels in industry, buildings and transport. And energy efficiency improvements of electric appliances point in the other direction. In other words, choosing the right metric is messy but in my view having an aspirational global commitment for electrification is both directionally correct and helps concentrates minds.

The same three sectors but very different races at country level

Just as different countries the three end-use sectors - buildings, industry and transport - move at very different speeds too.

• Buildings had some success with electrifying, with dozens of countries climbing steadily and the front-runners well past 50%, because the enabling technologies are mature. It is Asia that has seen the highest increases since 2000 (Indonesia: +44 percentage points, China: +36 percentage points, Thailand: +31 percentage points) but the highest shares today can be found in countries such as Norway (80%), Singapore (92%) and Malta (85%). The most unexpected is perhaps Norway given the cold climate and heating needs. Much of Norway’s heating today is from heat pumps as I wrote in this piece.

• Industry has seen significant progress in a number of countries. China is up from 20% in 2000 at 35% today. Mexico increased the share of electricity in industry from 26% to 40%. Indonesia has seen one of the steepest increases from just 10% in 2000 to now 31%. One standout country is Iceland with 95% industrial energy coming from electricity. But Iceland does not have an industrial sector representative of global industry.

• Transport has not made much progress in most countries at all except for a handful such as Singapore now at 13% electrification, and Norway at 7%, Sweden at 6% and China at 5%. But as I said earlier in this piece, this is likely going to change at a much faster rate going forward.

Richer doesn’t always mean more electrified

One potential explanation for different electrification rates could be that richer countries use more electricity compared to other energy carriers. But this is not an explanation that holds. Norway leads near 50%, but the United States and Germany are around 22%, below China and only modestly above India, while Indonesia at 27% outperforms economies many times richer. The correlation with GDP is weak and wealth clearly is not a binding constraint.

Much of it is national context, climate, industrial structure, economics and policy. What moves a country up the y-axis is the structure of its power sector, the carbon and energy prices facing industry, building codes that mandate electric heat, and the speed of EV adoption, none of which are determined by income alone. Norway is electrified because of decades of cheap hydro and deliberate transport policy, not simply because it is rich (although it clearly helped). Wealth buys more options to electrify but policy has a major impact on whether a country does it and at what speed. That is encouraging, because policy is the one variable governments can change relatively quickly.

The Global South is moving fastest

Much of the electrification debate focuses on the Global North. But rank countries by how much their electricity share has grown since 2000 and the largest gains are not in the Global North. China leads with a rise of nearly 20 percent since 2000. Indonesia, India, Morocco, Egypt and Thailand have all added substantially too. An important distinction: These are systems electrifying as they get built rather than retrofitting. In Global North countries it is the opposite.

Most of the century’s growth in energy demand will come from outside the OECD, which means the marginal unit of new energy infrastructure is being decided now, in exactly these countries. If it is built electric, the world avoids locking in decades of fossil fuel infrastructure and the long replacement cycles that come with it. If it is not, that capital is stranded later at far greater cost. China is the proof of concept: it electrified its end uses while adding clean generation faster than almost anyone had forecast (although not enough yet to replace significant amounts of coal generation but there are promising signs of peak emissions). It increasingly looks like the economic development pathway and the climate pathway are converging, which is perhaps the single most underrated trend in the transition.

How to accelerate

Most of the technologies required already exist today. A global electrification goal is laudable and has important symbolic meaning. But the key challenge is deployment at the scale and pace required. One of the main obstacles is the cost of electricity and in particular how it compares per unit of energy to the cost of fossil fuels. Some countries have already made reforms to bring about lower electricity prices but many have not. This week a forthcoming proposal from the European Commission leaked on how to reduce electricity prices in Europe. Many of the key ingredients in the draft are universally applicable: tax electricity less than fossil fuels, reform network regulation and tariffs to incentivise smarter use of the grid, and encourage the use of electricity when it is cheap and abundant. The next decade will show whether policy makers can move from aspiration to implementation.

(Note that this is an extraordinary Substack article not part of my usual newsletter.)

What I like about this package is that it treats the electricity bill as a whole rather than chasing the wholesale price alone. The energy component is only one part of what people pay. Grid costs, taxes and levies shape the rest, and they tend to get ignored in the public debate even though they make up a large share of the bill.

One caveat before I go into the detail. This is a draft. It will change in negotiation, and those bracketed figures may well change. But the direction of travel reads as a serious attempt to align the cost of electricity with where Europe says it wants to go.

Here are the key proposals broken down into network charges, energy taxation and smart meters.

Network charges

Grid costs already make up between 24% and 29% of the average household bill, and ACER expects annual grid investment in Europe to roughly double by 2050, reaching up to 100 billion euros a year. As the energy component of bills falls, this is the part that grows.

Making better use of the grid we already have is something I and various collaborators have been raising for years, so it is good to see it reflected so prominently. The point is simple: non-wire, innovative contractual and digital solutions should be considered alongside conventional reinforcement, and ideally before it, rather than as an afterthought once the wires have been ordered.

The draft replaces Article 18 of the Electricity Regulation with a tighter set of rules. Charges have to be cost-reflective and transparent, and they have to reward the behaviour that keeps system costs down, whether that is shifting demand, providing flexibility or reducing peak load. Regulators would set performance indicators and ACER would benchmark transmission operators against each other, with national regulators doing the same for the 2,600-plus distribution operators. There is also a real push on transparency, with regulators required to publish the cost categories and tariff values behind their methodologies.

Member States may also use public money to cover part of network costs, so that less of the grid bill falls on system users through tariffs. The draft explicitly mentions the general budget and, more interestingly, Cohesion funds or other available EU funding. In effect it is a way to socialise some grid investment onto the taxpayer, or onto EU funds, rather than the bill-payer, at a moment when those costs are climbing steeply.

The conditions attached read like an attempt to pre-empt State aid concerns, but how much room states really have will depend on how this sits with State aid rules and the Clean Industrial Deal aid framework, and the text does not fully resolve that.

Smart meters and smart grids

Without smart meters, time-of-use tariffs and flexibility incentives are unlikely to do much of anything. The draft sets an EU-wide floor: at least 50% of customers by 2030, rising to 65% by 2033. Many member states have already passed this, though there are laggards, with Germany still below 5%. Cost-benefit tests would only apply above that floor, which closes off the previous option of effectively opting out of the rollout altogether.

Alongside this, the draft proposes a framework for grid data exchange and a voluntary route for operators to develop shared digital and AI tools for running the system. More transparency can help facilitate faster grid connections and smarter use of the grid.

Taxation

The draft does what the Energy Taxation Directive failed to do because it is currently stuck in the legislative process due to the requirement of unanimity.

The proposal sidesteps that. It does this by stating that electricity must be taxed no higher than natural gas, and by placing the principle inside the Electricity Regulation under Article 194 it shifts the decision to qualified majority voting.

This proposal if adopted would have teeth: In a number of countries today the position is reversed, with electricity carrying the heavier tax burden. Member States would keep the freedom to set their own rates, but they could no longer hold electricity above gas. There is also scope to cut electricity tax for energy-intensive industry, down to zero, to support electrification where the cost gap with gas is the widest.

That legal base of this may well be fought over, and the “tax harmonisation in disguise” argument is likely to be made. But the signal it sends is the right one. If the goal is to electrify homes and industry, the tax system should not be pulling in the opposite direction.

The satellite image below shows the scale of change in a powerful way: Same rooftops in 2021 and 2025. Within the matter of 3 years rooftop solar has grown exponentially now visible from space (using Google Earth images).

We are used to energy transitions that happen because governments engineer them, with incentives and mandates and targets. Pakistan’s solar boom is the opposite. It is probably the fastest deployments of distributed solar anywhere in the world, and it happened largely in spite of the state rather than because of it. But Pakistan is also a cautionary tale for what happens in an unmanaged transition and by no means is this a blueprint for other countries to follow. But it shows just how rapidly solar can be deployed around the world.

An important note: I am NOT a Pakistan energy expert and this is very much someone looking in from the outside from high altitude. I hope that I got most of what I write here right but appreciate feedback and will include nuance where it is needed.

The numbers behind the story

Pakistan imported roughly 7.6 GW of solar panels in 2023, then 16.4 GW in 2024, then 16.9 GW in 2025. Around 55 GW have now arrived in total. That makes Pakistan the second-largest importer of solar panels in the world in 2025. Pakistan, a country of around 241 million people with a long history of power cuts, now absorbs a remarkable share of everything China’s solar industry makes.

And it is showing up in the grid data. By the summer of 2025, solar had become Pakistan's single largest source of electricity, generating around a quarter of the total during peak months. Official numbers understate the boom of solar because most of the panels were installed on rooftops and behind meters, in a distributed system no one is fully measuring.

Most of that hidden capacity is overwhelmingly residential. Of an estimated 33 GW of distributed solar, around half sits on the roofs of homes, with the rest split across industry, agriculture and commerce. Contrary to it being a story of big developers and utility contracts it is millions of households making the switch.

Why it happened

A key reason for the boom of solar in Pakistan is about resilience and grid reliability. Load-shedding, the scheduled rationing of power when supply cannot meet demand, has been a daily fact of life in Pakistan. How long the lights stay off depends on where you live, and the gap between a wealthy Islamabad suburb and rural Balochistan is enormous.

For years the outages were blamed on capacity shortages and not enough power stations. That was once true but it is not any more. Pakistan now runs a capacity surplus and in 2024 its power plants ran at barely a third of their capacity.

There is a long chain of interlinkages that exacerbate the problem of missing money as an excellent academic study analyses in great detail: Expensive imported fuel pushed generation costs up. Contracts with private producers guarantee them capacity payments whether their plants run or not, now around 2.5 trillion rupees a year for electricity that is often never generated. The public utilities cannot recover enough to cover those bills, so the debt piles up into a circular debt of close to 2 trillion rupees. Faced with that, the system rations supply, cutting power longest where losses are highest and bill payment is lowest.

Scheduled outages in April this year were permitted by the regulator for several hours a day - now a common feature in Pakistan. The distribution companies serving Islamabad and Karachi keep cuts to a handful of hours a day. In Sukkur and Quetta, rural customers can be without power for up to two-thirds of the day. The chart below shows the spread across the country’s main distribution companies, and the gulf between urban and rural service ( I could not get hold of the 2025 data yet but here is the April 2026 data).

To see how extreme this is, hold it against Europe and the US: In Europe, customers typically face around a few minutes of outages per year (both planned and unplanned). The average American, battered by hurricanes in 2024, lost about 11 hours over twelve months, and closer to two hours in a normal year. In some parts of Pakistan people can lose more than that before lunchtime. A household in Quetta can face 16 to 20 hours of outages in a single day. When that is the alternative, a roof full of panels and a battery offers energy resilience.

But not only outages drove solar demand. Electricity in Pakistan also became very expensive. The average effective tariff roughly tripled in a decade, from about 12.5 rupees per unit in 2015 to over 34 rupees in 2025. Pakistan’s power sector now carries a circular debt of close to 2 trillion rupees, the accumulated result of unpaid subsidies, under-recovery, and capacity payments owed to independent power producers for plants that often sit idle. Around 37% of a typical bill is now made up of surcharges tied to that debt.

So households and businesses were being asked to pay more and more for electricity that arrived unreliably, while watching the price of solar panels collapse. China’s manufacturing glut pushed module prices to record lows at exactly the moment Pakistani consumers were most desperate for an exit. A rooftop system could pay for itself in two to three years. For a factory facing industrial tariffs, the payback was faster still.

This is the engine of the boom Pakistan experienced. When grid power is expensive and unreliable, and the alternative is cheap and getting cheaper, people vote with their wallets.

The catch

A bottom-up solar revolution sounds like an unambiguously good news story. It partly is. But it also creates a genuine problem:

When the customers who can afford solar leave the grid, fewer kilowatt hours are consumed from the grid. But many of the high fixed costs remain. As a result, costs, including the capacity payments owed to idle power plants, get spread across a shrinking base of remaining customers. Their bills go up. That pushes more of them towards solar. Which shrinks the base again. This is the classic utility death spiral, and Pakistan is one of the clearest real-world examples of it now unfolding.

The government’s response tells you how seriously it takes the threat. In February 2026 the regulator, NEPRA, replaced the decade-old net metering regime with a new “net billing” system. Under the old rules, solar owners were credited for exported electricity at the same rate they paid for it. Under the new ones, utilities buy surplus solar at the national average purchase price of around 11 to 13 rupees per unit, while selling power back at the full retail rate of 40 to 55 rupees. Existing net-metering customers were eventually grandfathered in after a backlash, but the direction is clear: The regulator is trying to slow the exodus.

You can read this two ways, and both are true. It is a regressive move that makes solar less attractive and protects a failing centralised model. It is also a rational attempt to stop wealthier consumers from offloading the cost of the grid onto everyone who cannot afford to leave it. The regulator’s own projection is that without reform the losses would reach hundreds of billions of rupees over the next decade. The uncomfortable reality is that a transition this fast, and this unplanned, breaks things on the way through.

And there are equity concerns too. Solar adoption in Pakistan rises steeply with income. Wealthier households are far more likely to have already left, which is exactly why those still on the grid end up carrying more of its cost.

The bigger picture

Most of the world’s future electricity demand growth will come from countries that look more like Pakistan than like Germany. Hot, growing, energy-hungry, often saddled with creaking grids and expensive imported fuel. The conventional assumption was that these countries would electrify the way the rich world did, through large centralised power stations and slow grid expansion. Pakistan suggests another path is not only possible but may be unstoppable once the price signals line up.

The optimistic view is that a country that spent decades short of power has, in the space of three or four years, built one of the most solarised electricity systems in the developing world, funded overwhelmingly by private money. There is also another more positive side to this. Not everyone adding solar is leaving the grid; some are gaining access to electricity for the first time. According to the regulator, more than a quarter of Pakistanis had no grid connection in 2023. For these households, a panel and a small battery can provide power where the grid does not reach. The same cheap imports that let a factory cut its grid use also make it easier to electrify homes that were previously off-grid. Distributed solar, then, is not only a way for better-off consumers to exit the system. It can also help narrow Pakistan’s electricity access gap, provided the panels reach the households that need them most.

But Pakistan is also a warning. Solar deployment is the easy part now. The hard part is the system around it: the grid, the tariffs, the market design, the question of who pays for the shared infrastructure when generation becomes something millions of people own themselves. Pakistan ran ahead on deployment and left the institutions trailing. The result is a boom and a crisis happening at the same time, in the same place.

The lesson is not that distributed solar is dangerous. It is that policy has to keep pace with it. Solar will continue to be deployed in large numbers regardless. Whether it strengthens the existing system or hollows it out depends on choices that are being made now.

Further reading:

Shedding light on Pakistan’s distributed solar revolution

Investigating power outages in Pakistan

How Pakistan’s people-led solar boom is easing impact of Middle East energy crisis

Pakistan’s solar surge lifts it into rarefied 25% club

Industrial electrification potential much higher than commonly assumed

The report is called High Voltage, and the question behind it is simple: How much of industry energy demand could actually run on electricity? Industry accounts for 29% of global CO2 emissions, most of it from burning fossil fuels to make heat. For years that has been filed under “hard to abate”, a label that too often becomes an excuse to wait. A lot of analysis has already been produced on industry and potential energy pathways. So rather than build another model, we synthesised the best evidence that already exists: a meta-review of bottom-up engineering studies, and a systematic analysis of 1,600 global mitigation scenarios that report how much of industrial energy comes from electricity.

Here is the bright spot. Today, electricity meets only about a fifth of industry’s final energy. Across the full set of scenarios, the median rises to 35% by 2050. But that average hides the more interesting story. In the scenarios built around strong climate ambition, fast power-sector decarbonisation and supportive industrial policy, electrification reaches a median of 51% by 2050, with the upper tail going as high as 85%. Look further out, and by 2100 the high-electrification pathways cluster around 68 to 70%, with some reaching near-complete electrification.

Policy matters

The number that matters most is not any single percentage. It is what explains the gap between the low and high pathways. The spread does not reflect a technical ceiling. It reflects enabling conditions. The scenarios at the top consistently combine early, large-scale infrastructure investment with clean, affordable power and policy that points in the same direction for long enough to matter. In other words, the difference between an industry stuck at a third electrified and one approaching the high tail is not physics. It is choices we make this decade.

That reframing is the whole point. Electrification is a structural fossil fuel phase-out and decarbonisation lever, not a marginal one. Swapping an industrial gas boiler for a heat pump cuts emissions on day one, and then keeps cutting them automatically as the grid gets cleaner. Industrial heat pumps, electric boilers, electro-thermal storage and resistance heating can already replace large shares of low and medium-temperature heat. These are not prototypes. The barrier is rarely whether the technology works. It is usually that running electricity costs more than burning gas, because of how we price and tax the two.

None of this means it is easy. Steel, cement and chemicals need very high temperatures and process changes that are genuinely harder, although a suite of technologies such as plasma torches and induction heating is becoming available also for very high temperatures. Industrial assets are long-lived and capital-intensive, and firms are understandably wary of committing to higher upfront costs when future energy prices and carbon policy look uncertain. In sectors like pulp and paper or refining, electrifying also means losing the on-site power generation that currently offsets operating costs. These are real constraints. But they are reasons to design policy carefully, not reasons to wait.

The report sets out a short list of immediate priorities. Make electricity the safer, more stable option by removing the levies and charges that fall disproportionately on power. Scale the technologies that already work, using tools such as auctions, contracts for difference, operating grants or clean heat premiums to bridge the price gap. Fast-track grid connections and permitting, because a project that cannot get a grid connection is not a project. De-risk first movers with faster depreciation, tax credits and concessional finance. And use public procurement and industrial clusters to anchor demand so early projects can scale.

Then there are the structural priorities for the longer haul: making clean electricity genuinely abundant and affordable, aligning industrial strategy with electrification rather than against it, and focusing scarce innovation effort where electrification is hardest, without slowing down everywhere it is already viable. Harder to electrify sectors deserve targeted demonstration support. Low and medium-temperature heat, which is most of the demand, deserves to be rolled out today.

Treating industrial electrification as a strategic opportunity rather than an add on is also what I told the 27 EU energy ministers in my intervention in the European Council in October last year - you can watch the recording below.

The durable answer to fossil fuel volatility is to depend less on internationally traded fuels and more on electricity that can be generated at home from wind and sun. Every factory that electrifies is a factory less exposed to the next price shock. That is the framing I think gets industrial electrification out of the “too hard” box and into the “why are we waiting” one.

What about the economics?

But what does this cost? Some of this I have looked at directly. In recent work with colleagues at the Technical University of Denmark, published in Cell Reports Sustainability, we compared the lifetime cost of heat across the main clean options for low and medium-temperature industrial heat: high-temperature heat pumps, electric boilers, hydrogen boilers and biomass boilers. The chart below shows the levelised cost of heat for each in the EU using average price data under different electricity prices and operating hours.

Two things stand out. First, high-temperature heat pumps come out cheapest of all the clean options, and pairing them with thermal energy storage cuts the cost of heat by up to a further 15% by letting them run when electricity is cheap. Second, hydrogen boilers are not close. Even under optimistic assumptions, including free and unlimited hydrogen storage, they remain far more expensive than the alternatives, and even a 100 €/tonne carbon price does not make fossil gas more expensive than hydrogen.

The important part is the gas comparison. As the grey bands in the chart show, heat pumps are not yet universally cheaper than burning gas, but they close in fast, and at lower electricity prices the best configurations get there. That gap is not about the technology. It is about how we price electricity against gas, and about the upfront cost of the kit. Which is exactly why targeted support, for thermal storage, heat network connections and first movers, does so much work here.

The world’s first industrial heat electrification auction

Last year the EU launched the world’s first industrial heat electrification auction. The EU's Innovation Fund Heat Auction is the bloc's first dedicated instrument for industrial heat electrification. In May it was announced that the auction is set to run again: Following the €1 billion pilot launched in December 2025, which attracted 85 applications worth nearly €10 billion in bids and ultimately awarded around €400 million to 65 projects across ten countries, the Commission has confirmed a second round will take place in 2026 with another €1 billion budget. The signal matters as much as the money. In my intervention at the European Council (see video above) I urged the Commission to make the auction recurring rather than one-off. It is great to see Brussels committed to continuing the scheme - a recognition that long-lived industrial assets need investment predictability, not a single competitive window. The pilot already covered the full temperature range, with a dedicated high-temperature basket above 400°C and eligible technologies spanning heat pumps, induction, resistance heating, and plasma torches. Making it a standing instrument turns a proof-of-concept into the embryo of the future Industrial Decarbonisation Bank, and tells the supply chain that electrified process heat is now core EU industrial policy rather than a niche climate experiment.

The thing I will be watching is the 2020s themselves. The decisions made now on grids, retrofits and which equipment gets replaced when it wears out will largely determine whether industry can reach the high pathways later in the century. The technology is ready. The potential is bigger than most people realise. The question is whether policy keeps up.

The price shock did not come out of nowhere. Even before Hormuz, European gas prices had not returned to pre-2022 norms. TTF peaked at a monthly average of $74/MMBtu in August 2022 during the European energy crisis. By 2024 prices had eased back to around $9, still well above the pre-Ukraine baseline of $4 to $6. The Hormuz closure pushed them back up to $20. By May 2026, front-month TTF was still averaging around €48/MWh, roughly $16/MMBtu.

What does the future hold?

The question is: What happens next? How long will the Strait remain closed or at least severely constrained? In a new commentary published a few days ago in the journal One Earth with Sebastian Zwickl-Bernhard, Chi Kong Chyong, Anna Creti and Anne Neumann, we model three Hormuz scenarios. None of them produce a physical shortage in Europe. But all of them produce a price shock. Even partial reopening parks European LNG prices around $14/MMBtu, above pre-crisis norms. A six-month closure pushes them to $28. And if the United States were to respond to its own domestic price spike by restricting LNG exports, European prices reach $54.9/MMBtu, above the 2022 peak.

A word on where those numbers come from. The model treats the global LNG market as one connected network: producers, tankers, regasification terminals and buyers, all linked by physical constraints (how much each country can liquefy, ship and receive). It then asks a simple question. What is the cheapest way to keep supply and demand in balance, given a particular shock? Take Qatari and Emirati cargoes off the table because they have nowhere else to go, force the remaining buyers to compete for what is left, and the price Europe would have to pay to attract the marginal cargo is what falls out. We focus on flexible, spot-traded cargoes rather than long-term contracts.

Russian gas out, US LNG in

The post-2022 narrative was that Europe had won. We replaced Russian pipeline gas with seaborne LNG. We rebuilt storage. We diversified. What we didn’t do is reduce our reliance on the molecule. We swapped one source of vulnerability for another.

Russian gas supply to the EU collapsed from 165 bcm in 2019 to around 20 bcm in 2025. US LNG climbed from 5 to 81 bcm over the same period. IEEFA projects that by 2030, the United States could supply up to 80% of all EU LNG imports, around 115 bcm per year. The July 2025 EU-US trade deal locked $750 billion of US energy purchases in by 2028. Europe traded a structural dependency on Moscow for a structural dependency on Washington.

Short-term measures have limited impact on prices

What’s striking is how little Europe can do about this on its own. A 10% cut in European LNG demand under the worst-case scenario, which is roughly 2% of total EU gas consumption, brings prices from $54.9 down to $48.3. That’s a small reduction for a lot of effort. A coordinated 10% cut by all major importing regions brings prices to $37.2.

The instinct after a shock is to find a new supplier or sign a new long-term contract. But the actual lever is on the demand side.

A structural demand focus

Industry, buildings and power generation account for 94% of EU gas demand. In each of these sectors, electric alternatives are already commercially competitive. Heat pumps for buildings. Renewables and storage for power. Industrial electrification for process heat. Efficiency throughout. What’s changed is the price tag of not doing it. Every year of delay is a year of continued exposure to the next geopolitical crisis.

The bright spot here, if there is one, is that we now know what works and what doesn’t. New LNG terminals and long-term supply contracts do not insulate Europe from global price shocks. They entrench the exposure. Demand reduction, structurally pursued, is the only durable answer. The Hormuz crisis is hard evidence for an argument that climate policy people and energy security people have been making separately for years, and that turns out to be the same argument.

Bright spots

This is no longer hypothetical, and some governments are already moving and responding to the ongoing crisis with comprehensive strategies and not just short-term measures. In April 2026 France doubled its state support for electrification to up to €10 billion a year through 2030, explicitly citing the Strait of Hormuz crisis. The plan is concrete rather than aspirational. It bans gas heating in new buildings from 2027, begins a long phase-out of gas heating across two million social-housing units, targets a million French-made heat pumps a year by 2030, and backs the switch to electric transport with subsidies for high-mileage drivers and up to €100,000 per electric truck or van for businesses. The stated goal is to replace 85 terawatt-hours of gas, about a fifth of the country’s gas import bill, with domestic electricity by 2030.

Canada is working on a longer timescale but in the same direction. Its new national strategy, Powering Canada Strong, sets out to double the country’s electricity supply by 2050, a build-out it estimates will cost over a trillion Canadian dollars, while accelerating electrification across transport, buildings and industry as demand is expected to double. Canada starts from an enviable position, with around 80% of its electricity already from non-emitting sources, so the strategy frames clean, affordable power as the foundation of competitiveness and energy security rather than as a climate measure alone. The planned steps are faster and clearer permitting for grids and new generation, demand-side measures and grid modernisation, domestic manufacturing of system components, and federal financing through investment tax credits and public banks. The government’s own assessment is that most households end up paying less by 2050, because electrified equipment is more efficient.

There is one more thing to watch. In July, the European Commission is expected to publish its Electrification Action Plan, the structural counterpart to the emergency measures it set out in April. Electricity still meets less than a quarter of EU final energy demand, against a reference target of 32% by 2030. The plan is expected to set a new electrification target, tackle the barriers holding back electrification in industry, transport and buildings, and begin phasing out incentives pointing in the other direction locking in continued fossil fuel use.

The pattern across all of this is the same. France, Canada, and soon Brussels are not reaching for a new supplier or a new terminal. They are reaching for the demand side. That is the lesson the Hormuz crisis has forced into the open. You cannot diversify your way out of fossil fuel exposure, but you can electrify your way out of it. The countries that understand this will be the ones paying less, and worrying less, when the next shock comes. And there will be a next shock, the question is only when and where it starts.

Let me walk through what the data actually shows.

Wholesale prices: renewables help, gas dominates

Start with European wholesale electricity prices in the first four months of 2026. Plot the average wholesale price against the total renewables share, country by country, and you see a negative slope. Countries with higher renewables shares tend to have lower wholesale prices. Spain and Portugal sit in the bottom right, with renewables shares above 60% and wholesale prices around €43/MWh. Slovakia, Czechia and Poland sit in the top left, with low renewables shares and prices above €100/MWh.

But strip out hydro and look only at wind and solar, and the relationship weakens. The slope flattens. Hydro-rich Norway and Sweden do a lot of the work in pulling the (weak) headline correlation down.

That doesn’t mean wind and solar fail to lower wholesale prices. The merit order effect is real and well-documented. It means that in any given year, wholesale prices reflect many things at once: gas exposure, hydro availability, weather, interconnection, market design. The renewables share alone doesn’t determine the price.

The single strongest predictor of wholesale prices in 2025 was something else entirely: the share of hours in which gas set the price.

Where gas sets prices, wholesale market prices are higher. That is the structural problem European electricity markets need to solve, and it is not a problem that the renewables debate, framed the way it usually is, addresses.

Gas sets the price less often when there is more non-fossil capacity available at the margin - renewables, storage, hydro, legacy nuclear, flexible demand, imports through interconnectors. The more physically and commercially connected markets are, the more often the marginal unit is something other than domestic gas.

The renewables share is part of that story, but so are the other pieces of the non-fossil system. Spain for example has low prices because it has solar, wind, hydro, and nuclear, and because storage and flexibility increasingly fill the gaps.

Retail prices: the link breaks down

Now look at retail prices, which is what households and businesses actually pay. Here the picture is more uncomfortable for clean energy advocates.

Across European countries in 2025, household retail prices show essentially no relationship with the renewables share (R² = 0.02). The trend line is pretty flat. Portugal and Germany have similar renewables shares and very different retail prices. Norway has 99% renewables and the cheapest household electricity in Europe. Denmark has 91% renewables and prices well above the EU average. There is no clean pattern. But if we only look at wind and solar the picture changes with higher shares of wind and solar being correlated more closely with higher retail prices. However, this is mainly driven by two countries, Denmark and Germany and the correlation is weak (R² = 0.10).

Industrial prices tell a slightly different story. Across European countries in 2025, total renewables share is weakly and slightly negatively correlated with industrial electricity prices. The cheap end is Norway, Sweden and Finland. The most expensive countries are Ireland, Cyprus, Germany and Denmark. For wind+solar specifically the correlation is essentially zero (R² = 0.06) with no meaningful relationship either way.

The same is true across US states in 2024 and as pointed out in her excellent article a while back: “Having lots of solar and wind doesn’t guarantee that you have cheaper or more expensive power than your neighbours. Nor does the absence of solar and wind guarantee that your power will be cheap.”

Wide variation in retail prices, a positive but fairly weak correlation with renewables share, and a slightly negative but also very weak correlation with wind and solar share. So renewables aren’t pushing retail prices up. But they aren’t pushing them down in any visible way either. But why?

The wedge between wholesale and retail

The reason is straightforward once you look at the components of a retail bill. Wholesale electricity is only part of what households and industry pay. Network charges, policy costs, taxes, levies and supplier margins make up most of the bill in many countries. These components don’t move with the wholesale price.

In many cases, the non-energy components of the electricity bill are responsible for unhelpfully large electricity to gas price ratios. In the second half of 2025, the average European household paid roughly 2.5 to 3 times more for a unit of electricity than for the same energy delivered as gas.

For industry, the spread is even wider. The UK industrial ratio is 4.83. Finland’s is 0.78, meaning industrial users actually pay less for electricity than for gas.

These ratios have very little to do with the cost of generating electricity. They have everything to do with how policy costs and levies are distributed across fuels. In most of Europe, electricity carries the bulk of the renewable energy and energy efficiency levies. Gas largely escapes them. The result is a price signal that actively discourages the electrification we say we want.

System costs: real, growing, and often misread

There’s a third element in the wedge that deserves its own paragraph: system costs. As wind and solar shares rise, so do the costs of integrating them. Curtailment when generation outruns what the grid can absorb. Balancing services to manage variability. Capacity mechanisms. Grid reinforcement to move electricity from where it’s generated to where it’s used. Research has long recognised these costs and produced estimates of how high these costs are.

These costs are real and they are growing. German curtailment compensation reached €2.8 billion in 2024. GB balancing costs have been running above £2 billion a year. In most countries they end up on the network charge line of the retail bill.

But two things are worth saying about them.

First, they are still a relatively small share of total system costs (here is more detail on the UK for example). They are not the dominant driver of retail prices. Levies, taxes and legacy network costs are usually bigger.

Second, they are partly a function of choices, not a fixed property of renewables. Curtailment is high where transmission build-out has lagged generation build-out, as it has in Germany and the UK. Balancing costs are high where flexibility markets are thin and interconnection is weak.

The “renewables make the system expensive” framing gets the causation backwards. It is the failure to build the rest of the system at the same pace required that drives the costs up.

This matters for the policy debate because system costs get used as a catch-all argument against further deployment. The honest answer is that they are real, they are manageable, and the way to keep them manageable is to build the grid and the flexibility and storage infrastructure that renewables need. Not to slow renewables.

What the data is telling us

Three things, I think.

First, the “renewables make electricity expensive” claim doesn’t survive contact with the wholesale data. Where renewables shares are high and gas exposure is low, wholesale prices are low. Spain, Portugal, France, Sweden and Norway all sit at the cheap end. The countries with the most expensive wholesale electricity are the ones still dependent on gas to set their prices. But consumers care about retail, not wholesale prices.

Second, the counter-claim that more renewables automatically delivers cheaper retail prices doesn’t survive either. The pass-through is broken. Households and businesses see their bills shaped far more by network charges, levies and tax design than by what is happening on the wholesale market.

Third, and this is the bit I find most important: the policy debate about electricity prices is the wrong debate if it stops at the wholesale market. The bigger lever, by far, is the structure of retail bills. Whether you load levies onto electricity or spread them across fuels. Whether you tax electricity at a higher rate than gas. Whether network charges are recovered through volumetric or capacity-based tariffs. These are choices, and they vary enormously across countries.

Norway, Sweden and Finland have built retail electricity markets where electrification makes economic sense for ordinary households and industry. Heat pump adoption tracks that. EV adoption tracks that. The technology is the same everywhere. The price signal is what differs.

If we want clean electricity to translate into clean heat, clean industry and clean transport, the wholesale market is doing most of its job. The retail market mostly isn’t. That’s the policy problem worth focusing on.

A reform agenda

So what would beneficial retail reform look like? Four things matter most. Rebalancing levies and taxes across fuels, so that gas carries its share of the policy and environmental costs rather than electricity carrying most of them. Letting temporal and locational price signals reach consumers, so that flexible demand can respond to what the system actually needs. Protecting vulnerable consumers through safety valve mechanisms and targeted support rather than through suppressed energy prices for everyone. And spreading the recovery of grid investment costs over a longer period, so that today’s bills do not have to carry the full cost of building the system tomorrow’s electrified economy will use.

Retail reform is not the only outstanding piece. Wholesale market design itself is still in motion: long-term contracts (PPAs and CfDs), supplier hedging, locational signals, demand-side access to balancing markets, cross-border infrastructure cost sharing. Getting those right will matter at least as much as fixing retail.

A harder question sits behind these: how much of this can competitive retail markets deliver, and how much needs the regulator or government to step back in? Across Europe the political momentum is towards a stronger public role. Whether that is the right direction is genuinely contested in the industry. What is not contested is that the current retail structure in most countries is not fit for an electrified system.

Where prices are heading

So what happens next? Three structural forces are now pushing in the same direction, and a fourth could undo a lot of the gains if policymakers get it wrong.

The first force is the steady decline in gas-setter hours. Spain is the clearest example. In 2022, gas set the wholesale price for roughly half of all hours. By early 2026, that share had fallen below 10%. Wholesale prices fell with it. The same pattern is starting to appear in other markets with high renewables shares. As wind and solar capacity grows and storage builds out, gas spends fewer hours at the margin. The mechanical link between European wholesale prices and the TTF gas price weakens. This is not a forecast. It is already happening in the countries furthest along.

The second force is generation cost. Solar PV is now the cheapest source of new electricity in most of the world. Onshore wind is close behind. Battery storage costs have fallen by roughly 90% in the last decade. New-build renewables plus storage are now cheaper than running many existing gas plants in much of Europe. And the cost curve has not bottomed out.

The levelised cost of energy (LCOE) metric on its own understates this, because it ignores when electricity is produced and what it is worth at that moment. The IEA’s value-adjusted LCOE, or VALCOE, captures that better. It adjusts the headline cost for the energy, capacity and flexibility value each technology actually delivers to the system. On that metric, solar and wind in most markets are not just the cheapest new build. They are the most valuable new build, even after accounting for the integration costs of variable generation.

The third force is the one people forget. Network charges, policy costs and grid investment are largely fixed in the short run. They are recovered by spreading the total bill across the kWh sold. If electricity demand grows, the same fixed costs are spread across more units. The per-kWh charge falls. This is the simple arithmetic of electrification. Heat pumps, EVs and industrial electrification all push electricity demand up. As they do, the network and policy components of the bill get diluted. A grid that delivers twice as much electricity does not cost twice as much to run. Electrification is, in this sense, its own affordability strategy.

Put these three forces together and the wholesale and underlying cost outlook is clear. Average wholesale prices fall as renewables are being added and replace gas. Volatility falls as storage and flexibility are scaled. The gas exposure that has dominated European electricity bills since 2021 fades. Fixed system costs get spread across a larger base. None of this is automatic, and the pace varies by country, but the direction of travel is not in doubt.

The fourth force is the one to focus on. Decarbonising the system requires major grid investment. Transmission build-out, distribution reinforcement, flexibility services and capacity mechanisms all cost money. Those costs land on the network charge component of retail bills. If electrification stalls, those costs get spread across a shrinking volume of electricity sales, and the per-kWh charge rises. The risk is a vicious circle: high retail prices discourage electrification, low electrification volumes drive up per-kWh network charges, higher charges discourage electrification further. Some of that is already visible in UK and German bills.

This is why retail reform matters more than it gets credit for. Ensuring that electricity is not excessively taxed compared to other energy vectors, and recovering costs in a way that supports dynamic efficiency and keeps cost down are the key ingredients. The countries that get this right will see lower wholesale prices feed through into lower bills, growing electrification volumes spreading fixed costs across more kWh, and stronger incentives for the next round of investment. The countries that don’t will see falling generation costs offset by rising network and policy costs, and households and businesses will conclude that the energy transition is making them poorer. Same underlying physics, opposite political outcomes.

The good news is that the bright spots are real. Wholesale markets are decoupling from gas. Electrification creates its own tailwind once it gets going. The harder question is whether retail markets will be redesigned fast enough to let the gains through.

Acknowledgements: I want to thank Bram Claeys from CEER for his helpful review and commentary on an earlier draft. Any errors remain mine and mine alone.

This is not where most observers expected Spain to be. A decade ago, Spain was a cautionary tale of stranded solar investment and one of Europe’s more expensive power markets. Today it sits near the bottom of the price table, and the gap is widening.

The story behind that ranking is, on its surface, simple. Spain increasingly pushed gas increasingly out of its electricity supply, and the price of electricity followed.

The mix has changed beyond recognition

Twenty-five years ago, a third of Spain’s electricity came from coal. Today, coal is effectively gone. Gas, which surged in the 2000s as the replacement, peaked above 30% of generation in the late 2000s and has since been pushed back to roughly 19%. Nuclear has held steady around 19%, hydro and bioenergy together around 14%, and the remaining capacity has been steadily filled by wind and solar.

Wind alone supplied 20% of Spanish generation in 2025. Solar, which barely existed at scale in the early 2010s, hit 22%. Between them, those two technologies now generate more electricity than any other single category in the system, including the nuclear fleet that was once Spain’s reliable workhorse.

2022 was the turning point

If you stack solar and wind against all fossil generation (gas plus the last embers of coal and oil), the lines crossed in 2022. That was the first year wind plus solar generated more electricity than every fossil source combined. Through the first quarter of 2026, the gap has widened further. Solar and wind delivered 44% of generation, fossil fuels 17%.

This is the structural story that many arguments about energy policy circle around. Spain did not just add renewables on top of a fossil base. It substituted. The fossil curve has been falling, year after year, while the renewable curve has been climbing.

2022 also a turning point for wholesale electricity prices in Spain: The Iberian exception capped electricity prices initially to below EU27 average prices but even after the mechanism ended Spain widened the price gap further.

Why this shows up in the price

In a wholesale electricity market, the price in any given hour is set by the most expensive plant that needs to run to meet demand. For most of Europe, for most of the last decade, that has been a gas plant. The merit-order link from gas prices to power prices is the reason European households got an electricity bill shock when Russian pipeline gas collapsed in 2022.

What has quietly happened in Spain is that gas now sets the price far less often. In 2022, gas was the marginal plant in roughly 55% of all hours. In 2024 it had fallen to 27%. By the first four months of 2026, it was just 9%.

A note on methodology: The gas price-setter share is estimated using the method of Zakeri et al. (2023). For each hour, two conditions are checked: is gas actually generating, and is the market price at or above the calculated short-run marginal cost of running a gas plant? If both are true, gas is counted as the likely price setter. The marginal cost is calculated from monthly TTF and EU ETS prices, assuming a modern combined-cycle gas turbine with 52% efficiency. The chart counts hours where gas is directly dispatched and the market price reaches the calculated short-run marginal cost of gas generation, and likely understates the true influence of gas prices on the Spanish market. Feedback I have received on the analysis above suggests that two effects are not captured: hydro generators often bid not at their near-zero running costs but at the opportunity cost - essentially what they could earn when prices are high, which are usually set by gas. And cogeneration plants burn gas but have their costs calculated under a specific Spanish regulatory formula rather than the spot gas price, so they don't show up in the standard test the methodology uses. Gas is genuinely setting the price in those hours, but the method doesn't count it. The more precise claim is that gas generation at the margin is declining fast but gas price influence on the market remains broader than the headline figure suggests.

What this is not

Four things this story is not.

It is not a claim that gas has gone away. Gas still supplies around 1/5 of Spanish electricity, roughly the same share as nuclear. Its role on the margin (the price-setting bit) has shrunk faster than its role in the energy mix (the megawatt-hours-burned bit).

It is only the wholesale price. The €44/MWh figure is what generators are paid in the day-ahead market. It is not what households pay. Network charges, system costs, suppliers’ margins, taxes and policy levies sit on top, and they can easily double or treble the underlying figure by the time it reaches a domestic bill. Wholesale moving cheaper is necessary for retail bills to fall, but it is not sufficient.

Despite having Europe’s cheapest wholesale electricity, Spanish households pay above the EU average €0.265/kWh in 2025, ranking 16th out of 25 countries. That puts Spain more expensive than France, the Netherlands, Denmark, and most of Central and Eastern Europe. Some of this is to do with the amount of taxes and levies put on electricity

The chart below decomposes a typical Spanish household electricity bill into its main cost components. The total (€0.27/kWh, Eurostat, December 2025) and the broad tax-and-levy share (~31%) are anchored in official data; the sub-breakdown between energy, network charges, and system costs is a model estimate consistent with that total rather than a figure drawn from any single source.

Three caveats are worth keeping in mind:

First, the wholesale energy component - the largest slice - moves daily with the power market, so the share shown reflects a December 2025 average and can swing well beyond that range across the year.

Second, Spain has adjusted electricity taxes repeatedly in recent years: VAT fell from 21% to 5% during the 2022 energy crisis before being fully restored to 21% by January 2025 - the rate shown here - while the electricity excise tax (IEE) was cut again to 0.5% in March 2026. A chart drawn for a different month can look meaningfully different.

Third, free-market and regulated tariffs present the same underlying costs differently: on a commercial fixed-price contract, network tolls are typically bundled into the energy rate rather than shown separately, so the relative slices shift even though the underlying economics are identical.

One further simplification: part of every Spanish bill is a fixed daily charge based on contracted capacity in kilowatts, independent of energy consumed. Expressing this as a per-kWh figure - as the chart necessarily does - overstates the unit cost for low-consumption households and understates it for high-consumption ones.

Other system costs are rising. The flip side of getting energy cheap is paying more elsewhere to keep the system stable. Spain is procuring more balancing services, more reactive power and voltage support, and ultimately more transmission to move wind and solar from where the resources are good to where the demand is. Those costs land on consumers through network and policy charges rather than the wholesale price. They are not yet large enough to offset the wholesale gains, but they are climbing.

Nuclear is still doing the work, and the policy is to switch it off. About a fifth of Spanish electricity comes from a fleet of reactors scheduled to retire between 2027 and 2035. If those reactors come off the system without an equivalent low-carbon replacement, gas will move back up the merit order and the gas-setter share will climb again.

What it is

For all those caveats, this is the clearest example in Europe of the price effect that renewable advocates have been describing for years finally arriving in the data. More wind and solar on the grid means fewer hours when gas is the marginal plant. Fewer of those hours means a wholesale price that is decoupled from the gas market for most of the day. And a wholesale price decoupled from gas, in 2026, is a cheap one.

Spain is now a working demonstration that you can take an electricity system that was 33% coal a generation ago, 30%-plus gas a decade ago, and run it on roughly 44% wind-and-solar with the resulting wholesale prices among the lowest in Europe. The remaining questions are about resilience, about who actually captures the savings, and about how far the substitution can go before nuclear retirements reopen the gap. Another key question is also whether rising energy system costs (grids, balancing etc.) will be offset by lower wholesale prices.

The April 2025 blackout was not what the headlines said it was

On 28 April 2025 the entire Iberian peninsula went dark for most of a working day, and nearly 60 million people lost power. On the day, several high-reach outlets immediately blamed renewables with no evidence. The final ENTSO-E investigation, out nearly a year later, finds the opposite. Damian Cortinas, chair of ENTSO-E’s board, told the Financial Times that “the issue is not about renewables” but about the grid’s ability to manage “fast voltage variations” that can destabilise the system. Unusual oscillations triggered a cascade of plant disconnections, and grid managers lost control. The real lesson is not that Spain has gone too far on wind and solar, but that every country in Europe needs to modernise how it handles voltage stability. Spain, in Cortinas’s framing, is “a pioneer” encountering things “everyone else could encounter at some point”. The fixes exist today, statcoms (voltage shock absorbers), better disconnection protocols, and voltage control capabilities required of all generators rather than only renewables. “There is nothing in the recommendations that cannot be done tomorrow,” Cortinas told the FT. The open question is who pays for upgrading the existing fleet, and that, he said, “will need to be looked at in every country”.