Rethinking the energy transition

A decade on from the Paris Agreement, the goal to hold the increase in the global average temperature to below 2C — and ideally below 1.5C — above pre-industrial levels is at risk of failing. According to Nasa, 2024 was the warmest since record-keeping began in 1880, at 1.28C higher than the average from 1950 to 1980. The past 10 years have been the warmest on record, as the world struggles to limit greenhouse gas emissions — the main cause of climate change.

The Global Carbon Project estimated that in 2024 the world was on track to produce 37.4bn tonnes of CO₂ emissions from the use of fossil fuels, up 0.8 per cent from 2023. With 2024 also the first year of increase in land-use emissions in a decade, the aggregate rise was 2.5 per cent.

There has been some progress. Developed nations are deploying renewable and low-carbon technologies at a record pace and pursuing energy efficiency policies, which over the past two decades broke the historic link between economic growth and rising emissions.

In the US, the second-largest emitter of GHGs after China and the highest emitter per capita, a combination of retrofitted pollution controls and a shift towards renewables has seen CO₂ emissions drop by a quarter in the three decades to 2022 even as power generation has risen, according to the Environmental Protection Agency.

Emissions in China may have peaked last year, according to analysts, as a surge in electricity supply from wind, solar and nuclear energy was enough to cut output from heavily polluting coal plants.

Jan Rosenow, leader of the energy programme and professor of energy and climate policy at the Environmental Change Institute, says that while “it’s completely correct that we are not on track when it comes to reducing emissions”, it is important to look at what is happening in different sectors, technologies and geographies in order to understand what has been working.

He highlights the UK, where the electricity sector is close to being on track to decarbonise by the government’s goal of 2030. That was, he says, “unthinkable” 20 years ago when he first began analysing the UK energy sector. The rise in sales of electric vehicles is another bright spot despite it being “not that long ago that there wasn’t really any viable alternative to the internal combustion engine”.

But on a worldwide level, efforts to decarbonise are falling short, as Russia’s full-scale invasion of Ukraine and geopolitical tensions prompt policymakers to prioritise energy security and affordability over reducing emissions. Rising power demand due to increased use of air conditioning and artificial intelligence, as well as economic growth in developing nations, is complicating the energy transition.

There is no easy alternative to reducing the amount of GHGs emitted. The MIT Climate Project says that although the planet naturally removes roughly half of humanity’s emissions, any man-made carbon sequestering method would have to capture and store nearly 20bn tonnes of current emissions, never mind the existing gases in the atmosphere. Currently, sequestration methods are dealing with about one-tenth of this volume. Rather than undertake this “incredibly tall order”, it recommends we “just not emit it to begin with”.

Electricity demand is rising faster than expected

But cutting emissions is challenging due to surging power demand, particularly for electricity.

Global electricity demand surged by a whopping 4.3 per cent in 2024, well ahead of previous years. It was the main factor behind a similarly accelerated growth in energy demand of 2.2 per cent last year, compared to an average annual rate of 1.3 per cent between 2013 and 2023. Record global temperatures, which boosted demand for cooling measures in many countries, as well as rising consumption from industry, the electrification of transport and growth in data centres were key reasons behind this increase.

Higher electricity usage is not necessarily a bad thing — the International Energy Agency says that “the more climate-aligned a scenario, the more electricity demand rises”. But this is good news only when the source of that electricity is renewable and efficient. Fossil fuels still account for 60 per cent of generation and the sector is responsible for 36 per cent of energy-related emissions.

In 2024, as electricity use rose, the portion of end-use energy supplied by fossil fuels such as oil did indeed decline: oil’s share fell below 30 per cent for the first time on record (although volume rose by 1 per cent as the overall demand for energy expanded).

Yet even so, demand for oil has outpaced the IEA’s 2015 Stated Policies Scenario forecast of 100mn barrels per day for 2025. The latest usage projection for 2025 is nearly 104mn b/d, the level the 2015 forecasts had in mind for 2040.

The backdrop for coal, the most polluting of energy sources, is even worse. Coal demand in 2025 has blown through historical 2015 forecasts for 2040.

Natural gas has emerged as a so-called bridge fuel, whereby gas-fuelled power plants can be deployed in the short-term to meet surging electricity demand and replace coal generation. In the longer term the argument goes that these plants can be phased out, or modified to run on emission-free hydrogen or include carbon capture technologies.

But most analysts agree that switching to more renewable generation, combined with battery storage to provide reliable power supplies, and boosting energy efficiency, is critical if electricity is to provide a clean energy solution.

While that shift is under way, it is not happening quickly enough. The IEA says that renewables generation is set to grow by 2.7 times between 2022 and 2030, taking the share of renewables in the sector from 30 per cent to 46 per cent, with most new capacity to come from solar and wind. While the forecast growth in renewable generation capacity is higher than the aggregate of current government pledges, it is still lower than the three times growth in renewables necessary to keep net zero targets on track and that the collective nations set as a target at COP28 in 2023.

AI creates a near-term challenge

Unforeseen electricity demand, such as that coming from the build-out of artificial intelligence data centres, is complicating the energy transition. A briefing by the European parliament on AI and the energy sector explicitly linked the bloc’s increased energy demand to AI. While data centres were estimated to consume 1.5 per cent of global electricity in 2024, it was 3 per cent in Europe, with consumption as high as 20 per cent in Ireland. Data centre operators are aware of the problem, and 100 have signed the Climate Neutral Data Centre Pact, including hyperscalers Google, Microsoft, AWS and Meta.

Despite this, there is little usage data available. Brookings Institution, the think-tank, in July 2025 made a call for more transparency from the electric and tech companies behind AI, noting that without this information the environmental impact was hard to assess. Tech companies and data centres that serve them tend neither to disclose AI model energy needs, nor where they are accessing their supply.

However, the surge in demand has been so unplanned-for and so rapid that it is likely to have been supplied by existing grids, whose installed base still has a heavy reliance on fossil fuels. In addition, AI data centres require a constant and consistent electricity source, so the more unpredictable supply patterns of renewable generation tend to be inadequate.

AI’s power needs are especially high in the training phase as foundation models absorb all the data on the internet to educate themselves. But even an AI query requires considerably more energy than a standard web search. In July, Axios reported that ChatGPT users send more than 2.5bn prompts each day globally. As these increase, so will the need for power for AI infrastructure.

In May, MIT Technology Review took a stab at estimating the power consumption by AI. Citing 2024 research from Lawrence Berkeley National Laboratory, it said that data centre energy usage had been largely flat between 2005 and 2017 as efficiencies offset capacity growth, but that consumption had doubled between then and 2023. The same research estimated that US data centre electricity usage had risen from 1.9 per cent of total annual US consumption in 2018 to 4.4 per cent by 2023 and could rise to between 6.7 and 12 per cent of the total by 2028.

There is no doubt that further data centre expansion is coming. In February, President Donald Trump announced private sector investment of up to $500bn in AI infrastructure, while separately, Google expects to spend much of its $75bn capital expenditure in 2025 on AI-related data centres and servers. Some of Apple’s $600bn announced US-based expenditure over the coming four years will also be directed towards the manufacturing of servers in Houston and other Apple Intelligence infrastructure and data centres. With this expansion comes a pressing need to solve the shortfall in electricity supply.

Technology can boost energy efficiency

Improving efficiency is one piece of the emissions reduction puzzle. Rocky Mountain Institute says that almost two-thirds of all primary energy in the fossil fuel system is wasted in energy production, transportation and use “before fossil fuel has done any work or produced any benefit”, worth $4.5tn, or 5 per cent of global GDP.

Fossil fuel power plants and internal combustion engines are responsible for half of this, with the relative loss greatest in the generation phase, where only 30 to 40 per cent of source fuel is ultimately converted into electricity. Here, a switch to renewable sources can make a big difference. Centralised grid delivery of fossil fuel generated power is about 90 per cent efficient.

Efficiency is not only an issue in generation, but also in consumption. While AI is garnering negative press for its energy use, it can be a force for good, helping to integrate energy-efficient systems — virtual power plants, for instance, frequently use AI to optimise battery charging and discharging back to the grid. The European Commission’s department responsible for energy policy believes that AI can “transform” the continent’s energy grid to move the bloc towards sustainability. Using AI, highly digitalised smart grids can predict consumption and production patterns and manage grid congestion, as well as foresee maintenance requirements to help avoid shutdowns.

AI can also help buildings reduce their significant emissions, with the IEA forecasting in a 2025 report that “widespread use of AI in buildings saves more than 300 terawatt hours in global electricity demand in 2035, equivalent to 5 per cent of the total consumption of electricity for heating and cooling”. Further up the chain, AI can analyse building designs for efficiency, preventing costly errors and redesign requirements. Using digital twins AI can also save carbon through reduced use of construction materials.

Technology companies are also attempting to reduce the draw on energy by introducing more efficient chips. Universities are also devising solutions. Stanford is working on neuromorphic chips that allow more functions to migrate to edge devices with lower power requirements, while Princeton is developing a low-power AI chip based on a different approach to computation.

The nuclear renaissance

As the need for electricity outstrips supply and other reliable options are tapped out, nuclear power is seeing a resurgence worldwide. Nuclear power plants provide the type of around the clock electricity that the technology industry needs to power internet data centres, whilst also being emissions-free.

In the US, Meta, Google, Amazon and Microsoft have all signed agreements with companies developing nuclear technologies as they seek to power their AI data centres at the same time as meeting internal emissions reduction targets. Big Tech’s support is critical to the development of small modular reactors, a new type of reactor that typically generates a third or less of the power of standard models.

Meanwhile, surging power demand is giving new life to existing large-scale nuclear plants and prompting a construction boom in the sector. In 2024 Microsoft announced a power supply deal with US utility Constellation Energy, which is scheduled to reopen the mothballed Three Mile Island, a nuclear plant in Pennsylvania.

Later this year Holtec International, an US-based nuclear company, plans to reopen the Palisades nuclear plant in Michigan, which was closed for decommissioning in 2022. US utilities are also extending the lifespan of several nuclear plants due for closure in coming years, including Diablo Canyon in California.

The Trump administration has called for development work to begin on constructing an additional 10 large-scale reactors before 2030, as it prioritises generating power to enable the US to win the AI race. It has set a goal of quadrupling nuclear capacity by 2050.

The explosion of activity in the US, which has the world’s largest nuclear fleet of 93 operating reactors, represents a turnaround for the sector. Competition from low-cost gas power plants during the country’s shale revolution and the Fukushima nuclear accident in Japan in 2011 had caused the industry to contract. In 2024 nuclear energy’s share of overall electricity generation in the US fell below 18 per cent, its lowest level in the 21st century.

The sector’s return to favour is not limited to America. The nuclear industry is expanding rapidly in Asia and growing again in Europe after a lull following Fukushima. China has deployed 58 reactors and has an additional 33 under construction, as it seeks to meet its climate targets and reduce air pollution that is caused by coal power plants.

Nick Smith, managing director of Global Decarbonisation Advisory, says nuclear power is one of few existing ways available to much of Asia to achieve decarbonisation goals. Many Asian countries are “energy resource poor and they don’t have the land and renewable resource availability that somewhere like Australia does”. In July, Japan’s Kansai Electric Power began to investigate whether it might build the country’s first new reactor since the Fukushima disaster. Only 14 of the 54 reactors that were shut down after 2011 have reopened.

While the medium-term incentive may be to meet a generation shortfall, tackling climate change relies on successful nuclear rollout. Nuclear Energy Agency and OECD research from 2024 says that 90 pathways considered in 2018 to meet net zero required nuclear energy generation capacity to treble by 2050.

But significant challenges to the rapid and cost effective deployment of nuclear energy remain, particularly in western nations. Large-scale nuclear construction projects are vulnerable to delays and cost overruns, due to their one-of-a-kind nature. In July the UK government confirmed the projected cost of the Sizewell C nuclear plant had jumped to £38bn, up from a previous estimate of £20bn.

Meanwhile, the first commercial deployment of SMRs in the US is not expected until later this decade, highlighting ongoing technology and financing risks for the sector.

The hydrogen economy stalls

Hydrogen is a zero emission fuel at the point of use and for decades advocates have suggested it can play a crucial role in decarbonising sectors where there are barriers to using electricity, such as heavy industry, long-haul transport and power storage.

The hydrogen currently used by heavy industry is produced through a process involving natural gas, which is relatively cheap but creates significant carbon emissions. There is an alternative method where the fuel can be made via electrolysis of water, using renewable energy to power the energy intensive process. This creates so-called green hydrogen, which is emissions-free but typically costs more than double the hydrogen produced from gas.

Adding to the expense of hydrogen generally is the cost and relative low-efficiency of hydrogen electrolysers, according to Smith at Global Decarbonisation Advisory.

Hydrogen also poses challenges for infrastructure providers due to its low density, which makes it prone to leaking from pipelines and other storage tanks.

“It’s going to be difficult to export in the short term as liquid hydrogen . . . We still need to understand liquid hydrogen properties a lot more before we can safely and cost-effectively introduce that to the economy longer-term.”

Another issue, Smith says, is that many projects around the world are struggling to find an offtaker who will commit to a premium priced product. Governments “need to take a lead role in terms of underwriting some of the risk associated with it”.

The US, Japan and the EU have backed taxpayer funded incentive schemes to encourage hydrogen production. But the collapse of tens of billions of dollars of hydrogen projects in 2025 in the US, UK, Australia and elsewhere highlight the challenges facing the industry.

There is a big backer that could make a difference: China laid out a long-term plan for the industry’s development in 2022. Smith believes that China “will effectively do to the hydrogen electrolyser and fuel cell industry what they did to the solar panel industry in terms of making it very, very cost effective. So I think there’s a faster pathway there in the future.”

Not everyone agrees that hydrogen can have an expanded role in the transition. “The time it takes and the expense involved is often played down,” says Rosenow at the Environmental Change Institute, with progress so far too slow and the scale too small to do much beyond replacing current industrial uses of hydrogen. However, given that the existing emissions from this sector are larger than those of all of Germany, that is still a significant contribution to reduced emissions.

Carbon capture solutions

While all major international organisations such as the EU, IPCC and IEA consider carbon capture a fundamental element of a decarbonisation strategy, it is not without issue. As the MIT research cited above states: relying on carbon capture is a less certain approach than not emitting GHGs in the first place. Even more recently, in September 2025, a study found that the risks of captured carbon leaking back into the atmosphere makes such technologies less of a silver bullet for resolving emissions than previously hoped. Not only would the high levels of carbon sequestration required to make a difference to emissions fail to reduce warming as much as hoped, but they would also lead to risks of earthquakes and engineering failures.

Still, a plethora of companies have evolved to find a solution — an add-on technology that works would have a huge market among incumbent fossil fuel energy providers and users resistant to change. For some industries, CO₂ emissions are inherent to the process and hard if not impossible to avoid, making CCUS a more straightforward solution than an entire redesign of such sectors.

Interconnectivity and storage solutions

Modernised and interconnected grids are a critical aspect for progress, as the spring blackout in Europe illustrated. In April, large parts of Spain, Portugal and France lost power as a power surge from a solar plant knocked out supply, revealing weaknesses in the grid. While Spain has been well ahead of the region with its rollout of solar capacity, the rest of its infrastructure has not kept up — notably, its grid is not optimised for the more variable supply that comes from renewables.

In fact, while Europe’s interconnected grid is the largest in the world there has been under-investment across the bloc, with 40 per cent of it more than 40 years old. The collapse of the system showed that investment in renewable generation capacity must be matched by investment to optimise the grid for its use if stable supply is to be assured.

Since the blackout, France, which had been resisting more interconnectors with Spain, is coming around to the idea of enhancing integration. Better interconnectivity more broadly would allow nations with more renewable resources to transfer power to those with less. It might also help Spain, the bloc’s poster child for renewable generation, stabilise energy prices — which frequently trade below zero.

Improved battery storage could also help by smoothing out lumpy renewables generation to match better with demand. Battery storage also offers a more rapid solution versus the lengthy approvals process required for transmission lines. Smith at Global Decarbonisation Advisory says: “We’re seeing that in Australia in particular that’s a real challenge to build out the transmission that the grid requires for renewables, but at the same time we’re seeing battery costs fall incredibly fast. So the ability to store energy and shift that into periods from the middle of the day is something that will continue to occur at a pretty rapid rate. And that’s fantastic in terms of being able to capture wind and sun and move it.”

Financing and the private sector

Dedicated green finance is still very small-scale. The entire market reached just $8tn in 2024, according to the UN Conference on Trade and Development, a fraction of the size of global credit and equity markets, which LSE Group estimates at about $250tn. Although the green finance market grew by 17 per cent from 2023, Unctad noted “intensified headwinds and growing investor caution”.

For the transition to have any hope of success, it needs a lot more funding. The Energy Transitions Commission, a coalition of energy leaders, said in 2023 that $3.5tn in average annual capital investment would be needed up to 2050 to build a net zero economy. Some $500bn of this, it says, can come from a reduction in fossil fuel funding, a target that looks increasingly unlikely. More money will be needed from almost all economies, with smaller ones needing proportionately greater increases, including grants to support the end of deforestation, among other things.

A Climate Policy Initiative looks at the mechanisms by which green funding can be provided. For any hope of improvement, green banks need clearer mandates with more supportive government policy and strategy, aligning green finance initiatives with central banks and economic planning as well as the existing financial ecosystem. More favourable pricing, lower interest rates and longer debt tenures as well as repayment guarantees and currency risk hedging would also encourage more risk-averse capital flows.

Private sector involvement is also crucial, with closer public-private collaboration. The International Renewable Energy Agency has called active private sector engagement “vital”, particularly in offering low-cost capital for financing energy transition projects.

Underpinning from public funds can encourage this private sector involvement. The Connecticut Green Bank uses a combination of ratepayer funds and private capital to provide low-cost, long-term capital. In August, it partnered with GoodLeap to provide virtual power plants, localised collaborations that combine domestic solar generation with battery storage to reduce strain on commercial grids.

Supportive financing has clearly helped in developed markets, where subsidies and rebates have encouraged installation of renewables such as solar panels at the homeowner level in countries that have temperate climates. The Environmental Change Institute’s Rosenow says that far-more suitable countries for this technology such as Egypt, which hosted COP27, have practically no solar despite huge potential. “I think it’s a combination of policy but also the cost of capital — and risk that’s driving the cost of capital — that explains why we see this disparity.” Figuring out how to attract investment into more challenging countries from a risk perspective is critical to helping them advance their energy transitions.

While subsidies and support are rising in aggregate and reached record highs in 2024, according to the Institute for Energy Research, sentiment has been damped by the new US administration, whose attitude is more pro-fossil fuel. This has encouraged more companies to give up on climate pledges, with financial institutions abandoning a co-operative stance to transition funding. In August, Barclays became just the latest to exit the Net-Zero Banking Alliance, alongside HSBC and US banks including JPMorgan and Bank of America. In April 2025, following the departure of several Japanese banks and Australia’s Macquarie, the Institute for Energy Economics and Financial Analysis said that those that had exited accounted for nearly 40 per cent of the alliance’s total assets.

These backward steps hamper co-ordinated progress and risk destroying hope of staying within Paris Agreement goals. A 2024 analysis from MIT Sloan School of Management, Columbia University and European Central Bank academics found that European banks’ so-called commitments to helping stall climate change had had little impact on their practical lending habits. Amounts loaned varied little between financial institutions with climate commitments and those without, while interest rates charged to companies did not differentiate between high and low emitters.

The optimistic view is that with the unravelling of voluntary targets and alliances, governments that have legally binding net zero goals might step in to legislate more firmly.

Where next?

The energy transition is a global challenge but the strategies to tackle it vary from region to region, never mind country to country. Carrots have been effective but sticks also need to be better deployed. Rosenow says that governments must make it a lot harder to continue using fossil fuels. “There’s a lot of inertia in the system that we built . . . a lot of jobs, a lot of institutional capital was invested in our fossil fuel infrastructure. So rolling that back . . . requires a whole set of reforms far beyond just supporting clean energy rollout.” At this stage, withdrawing support from fossil fuels is possibly even more important than promoting renewables.

China has been responsible for a huge proportion of the switch into renewable energy sources and yet its ever expanding economy needs more power than these can supply. It continues to build new coal-fired plants.

But it can still offer a road map for other countries, according to Joseph Jacobelli of Bourne Impact Capital and an author of two books on the energy transition. He says geopolitics often prevent China from getting credit in the US for its progress, but Beijing has driven “massive” increases in clean energy generation and electric vehicle uptake. The country also leads in green finance, and not only domestically. Leading state-owned bank ICBC has “issued green bonds globally as well. It’s a leader in green lending . . . it’s getting into carbon credits.”

China is also leading the way in new technologies, backing developments in hydrogen and piloting “vehicle-to-grid” technology. Jacobelli says: “The objective is to establish by the end of 2025 comprehensive V2G infrastructure and policies in over 9 ‘demonstration’ cities at over 30 pilot projects, also setting up large-scale technical and commercial viability.” He says this will translate into huge energy storage capacity.

Europe’s carbon border pricing mechanism is incentivising manufacturing countries such as Vietnam to find cleaner alternatives. “If you want to be able to sell products into Europe then you actually need to find a way to decarbonise,” says Smith at Global Decarbonisation Advisory.

Some smaller countries are making the switch even without a centralised policy. In 2024 Pakistan was the largest importer of solar equipment from China, as individuals opted to decouple from the grid.

Rosenow says cheap equipment and an inefficient grid were both motivating factors. “People were discovering solar could be a much cheaper way of getting electricity, and that’s why they started just buying lots of solar panels — not because politicians gave them lots of money and subsidised it or required it. It was really very much driven from the bottom up.”

The Trump administration’s reversal of climate friendly policies, meanwhile, need not necessarily be a bad thing. Smith says that the Inflation Reduction Act had pulled a lot of capital into the US, limiting funding for projects elsewhere. Jacobelli says that the abandonment of the IRA, while risking the US falling behind, may be a “blessing in disguise” for the two fastest-growing clean energy regions, Asia and the EU. “Global infrastructure funds will still have to invest. If the US is not an option or is a ‘tough’ option, then other regions will benefit from capital availability,” he says.

While overall progress towards the energy transition has been patchy and non-linear, Jacobelli remains optimistic. “[The energy transition] is a massive undertaking, with different countries moving at different speeds.Broadly, though, they are moving in one direction: global carbon neutrality. Just look at the massive growth so far.”