The global average surface temperature between January and September 2026 was approximately 1.47°C above pre-industrial levels, according to the World Meteorological Organization’s (WMO) latest State of the Climate Update 2026. The past 12 years, from 2015 to 2026, are on track to be the warmest on record, as rising greenhouse gas concentrations continue to increase the amount of heat trapped in the Earth’s climate system. Reliable climate prediction is becoming need of the hour as scientists work to understand how continued warming could affect future weather patterns, ecosystems and human populations.

Understanding how different emissions pathways could influence future temperatures, rainfall and other climate conditions requires extensive computer modelling. However, running comprehensive Earth system models can take months for a single climate scenario, limiting how many possible futures scientists can examine.

Massachusetts Institute of Technology (MIT) have now developed a new artificial intelligence training technique that could reduce the computational effort involved in this process. Their method uses carefully generated synthetic data to train climate emulators, potentially reducing the number of full-scale climate model runs required from six or eight to just one. The researchers also report improved performance when these AI models are tested against emissions scenarios they have not encountered during training.

Climate Prediction: Why Climate Scientists Need Faster Models

Earth system models are among the most comprehensive tools used to project long-term changes in the climate. They simulate interactions between different components of the Earth system and help scientists estimate how temperature, rainfall and other climate variables could respond to changing greenhouse gas emissions. However, their computational demands are considerable. Running an ESM on a supercomputer can take months to produce a single future climate projection.

Climate emulators offer a faster alternative. They are trained to replicate the outputs of full-scale climate models and can produce simulations in seconds. This allows researchers to test thousands of emissions scenarios instead of relying on a limited number of computationally expensive runs. However, the reliability of these AI models depends heavily on the data used to train them.

Scientists commonly train climate emulators using six to eight future emissions scenarios based on realistic projections. Each scenario must first be processed through a full-scale Earth system model, making the preparation of training data both expensive and time-consuming. The MIT researchers found that this approach could also leave emulators struggling to predict scenarios that differ from the ones they encountered during training.

Training AI With Synthetic Climate Scenarios

To address this limitation, the researchers developed an algorithm that generates synthetic emissions scenarios specifically designed to improve an emulator’s predictive ability. Unlike conventional training datasets, these synthetic scenarios do not have to follow realistic patterns of future emissions. Instead, the algorithm identifies variations that help the AI model learn more about how the climate system responds to changing emissions.

The process begins with an initial climate scenario generated using a relatively fast and simple climate model. The researchers then compare the emulator’s predictions with the results of the Earth system model to measure its errors. The algorithm repeatedly adjusts the training data to reduce these errors. For example, it may introduce a sharp increase in carbon dioxide emissions around 2050 or a rapid reduction in methane emissions in 2037.

These changes produce emissions trajectories with rapid fluctuations, rather than the smoother patterns generally seen in conventional future projections. According to the researchers, these variations help the emulator capture more information about the overall climate system. An important advantage is that the method requires just one full-scale Earth system model run to generate the training data, compared with the six to eight runs typically needed for conventional approaches.

Why Adaptability Matters for Climate Policy

The ability to predict unfamiliar scenarios is particularly important when scientists assess the effects of specific climate policies. For instance, a government may introduce regulations to reduce one pollutant while emissions of other greenhouse gases continue to increase. Understanding how the climate responds to such an intervention requires models that can account for changes that may not follow conventional emissions projections.

The MIT researchers found that emulators trained using their technique could capture the effects of local pollution-control policies that reduce one pollutant while other emissions continue rising. The traditional emulators tested by the team failed to reproduce these outcomes. This could make the technique useful for evaluating targeted interventions and examining how different policy decisions influence future climate conditions.

Accurate climate projections also inform decisions beyond emissions policy, including planning for resilient power grids, protecting public health and safeguarding agricultural production. The research was led by Christopher Womack, a PhD graduate of MIT who is now a postdoctoral researcher at ETH Zurich. The team also included researchers from MIT’s Department of Earth, Atmospheric and Planetary Sciences, its Center for Sustainability Science and Strategy, and Imperial College London. Noelle E. Selin, professor at MIT and director of its Center for Sustainability Science and Strategy, was the senior author.

The researchers plan to test the technique using emissions scenarios generated by a full-scale Earth system model. They also intend to examine its application to other climate models and explore whether the approach could be useful in broader machine learning applications. The work was supported by Schmidt Sciences through the MIT Grand Challenges initiative and was part of the Bringing Computation to the Climate Challenge project.

Global climate finance reached an unprecedented annual average of $1.9 trillion in 2023–2024, increasing by 32 per cent compared with the previous two-year period. However, despite the growth in investment, developing countries continue to face significant barriers in accessing the resources needed to adapt to climate change, according to assessments by the United Nations Framework Convention on Climate Change (UNFCCC) and the Organisation for Economic Co-operation and Development (OECD).

The findings come as countries meet for the 31st United Nations Climate Change Conference (COP31) in Antalya, Türkiye, from November 9 to 20. The conference takes place during the transition from the previous $100 billion annual climate-finance commitment to a new target of $300 billion by 2035.

Developed countries exceed $100 billion target, but questions remain

Developed countries had pledged to mobilise $100 billion annually by 2020 to support climate action in developing countries. The commitment was subsequently extended through 2025. According to the OECD’s 2026 assessment, developed countries provided and mobilised $132.8 billion in 2023 and $136.7 billion in 2024, exceeding the target for the third consecutive year.

However, the reported totals do not necessarily represent grants or money directly disbursed to developing countries. They include different forms of public finance, multilateral development bank contributions and private finance mobilised through public interventions.

Differences in accounting methods have also complicated assessments of whether countries are meeting their commitments. The UNFCCC’s 2026 assessment highlights variations in how countries report loans, grants, commitments and disbursements, making comparisons between datasets difficult.

Total global climate finance reached an annual average of USD 1.9 trillion in 2023–2024.

The earlier target is now being replaced by the New Collective Quantified Goal (NCQG), adopted at COP29 in 2024. It calls for developed countries to take the lead in mobilising at least $300 billion annually for developing countries by 2035. The wider ambition is to scale up climate finance from all sources to at least $1.3 trillion annually by the same year.

Adaptation remains dependent on public finance

The expansion of climate finance has not translated into comparable support for adaptation. Investments in mitigation, including renewable energy and low-emission transport, have attracted substantial private capital. Adaptation projects such as flood protection, resilient agriculture, water management and coastal infrastructure, however, often generate fewer direct commercial returns.

The OECD reported that developed countries’ adaptation finance reached $33.6 billion in 2023 and $34.7 billion in 2024. Public finance accounted for more than 90 per cent of the tracked adaptation finance in both years, highlighting the continued dependence of adaptation initiatives on government and multilateral funding.

The broader financing gap is considerably larger. The United Nations Environment Programme’s Adaptation Gap Report 2026 estimates that developing countries will require approximately $305 billion annually by 2035 to meet their adaptation needs. This is substantially higher than the current international public adaptation finance flows.

LDCs and island nations receive a limited share

The financing imbalance is particularly significant for Least Developed Countries (LDCs) and Small Island Developing States (SIDS), which face high exposure to extreme weather, rising sea levels and other climate-related risks.

According to the OECD, LDCs received an annual average of $24.5 billion in climate finance in 2023–2024. SIDS received $3.8 billion in each of those years. The figures underline the limited scale of financial support reaching countries with substantial adaptation requirements.

A report published by the United Nations Capital Development Fund ahead of COP31 found that LDCs, which are home to almost 20 per cent of the global population, received only 1.9 per cent of global climate finance in 2024.

The terms on which finance is provided are also important. The UNFCCC’s assessment found that loans accounted for 91 per cent of multilateral development bank climate finance in its 2023–2024 channel-level breakdown. While loans can support large infrastructure projects, debt-based finance may place additional pressure on countries already facing economic constraints.

COP31 must move from targets to delivery

The latest figures reveal a gap between the scale of global climate investment and the resources reaching countries most vulnerable to climate impacts. Although the new $300 billion target represents an increase in the international climate-finance commitment, its effectiveness will depend on how much funding is accessible, predictable and directed towards adaptation.

For developing countries, particularly LDCs and SIDS, increasing the grant component, improving transparency in financial reporting and expanding concessional funding will be essential.

COP31 offers an opportunity to address these concerns as countries begin implementing the new climate-finance framework. The challenge is no longer only to mobilise larger sums, but to ensure that finance reaches communities facing the greatest climate risks.

India has reached a major milestone in renewable energy capacity, with non-fossil fuel sources accounting for half of its installed electricity capacity in June 2025. Yet coal continued to generate 71% of the country’s electricity that year. This gap between capacity additions and the actual electricity mix is central to India’s performance in the Climate Change Performance Index (CCPI) 2027, where the country has slipped five places to 28th.

The index, published by Germanwatch and the NewClimate Institute, gives India an overall medium rating. Its assessment points to a mixed record: India has made substantial progress in expanding renewable energy and has performed relatively well on climate policy, but its energy use, emissions and renewable energy performance remain areas of concern.

India’s fall in the rankings should not be read as evidence that its renewable energy expansion has stalled. The country reached the 50% non-fossil capacity milestone five years ahead of its earlier 2030 target. However, installed capacity measures the potential to generate electricity, not how much power sources actually supply. The milestone also includes nuclear energy.

Data from Ember’s Global Electricity Review 2026 show that coal accounted for 71% of India’s electricity generation in 2025, against a global average of 33%. India’s coal generation did fall by 2.9% that year, while renewable generation rose sharply. But the continued dominance of coal in the electricity mix shows how much of the country’s power demand remains tied to fossil fuels.

Grid Constraints Slow the Transition

The challenge is becoming visible in the way India’s electricity grid handles solar power. An Ember analysis published in January 2026 estimated that the country curtailed 2.3 terawatt hours (TWh) of solar generation between late May and December 2025. Grid operators had to reduce solar output as an emergency measure to maintain grid security.

The problem is linked partly to the limited flexibility of conventional power plants. Solar generation rises during daylight hours, while electricity demand continues into the evening, when solar output falls. Coal-fired plants that remain online to meet demand can have limited room to reduce their output further. Without sufficient storage and other flexible sources of electricity, the grid cannot always accommodate the full amount of renewable power available.

This creates a practical constraint for India’s energy transition. Adding solar and wind capacity is essential, but the electricity system also needs stronger transmission networks, battery storage and power plants capable of adjusting output to changes in demand and renewable generation. Otherwise, more installed capacity will not translate into an equivalent reduction in coal-fired electricity.

The CCPI’s category rankings reflect the uneven progress. India ranks 13th in climate policy, but 31st in greenhouse gas emissions, 33rd in energy use and 40th in renewable energy. Its renewable energy category receives a low overall rating, despite a high trend indicator. This suggests that recent progress is significant, while the overall level of renewable energy deployment and its contribution to the energy system still have considerable room to improve.

India’s targets: Stronger Implementation

India’s updated climate commitments set out targets for 2035. These include reducing the emissions intensity of its gross domestic product by 47% from 2005 levels, raising the share of non-fossil sources to around 60% of cumulative installed electricity capacity, and creating a carbon sink of 3.5–4 billion tonnes of carbon dioxide equivalent through forest and tree cover.

The country has also begun implementing its Carbon Credit Trading Scheme, with emissions-intensity targets assigned to 490 industrial entities from January 2026. The effectiveness of the scheme will depend on the ambition of those targets, reliable monitoring and verification, and evidence that participating industries are actually reducing emissions.

The index also draws attention to gaps in energy efficiency. Building energy codes and appliance standards exist, but compliance remains uneven, particularly in informal markets. In transport, urban rail connectivity had expanded to 26 cities by August 2026. However, public transport improvements will also require affordable and reliable bus services, along with safer infrastructure for walking and cycling.

India’s dependence on imported fossil fuels adds an economic and energy-security dimension to the transition. Disruptions to shipping through the Strait of Hormuz have highlighted the risks faced by countries that rely on international supplies of oil and gas. Reducing fossil fuel consumption would help limit exposure to such external shocks, alongside its climate benefits.

Climate Change Performance Index: How India Compares With Other Major Economies

The CCPI leaves its first three positions vacant because no country meets the highest performance standards across all assessed categories. Denmark ranks highest at fourth place, followed by Lithuania and Luxembourg. Denmark has remained the best-performing country in the index for six consecutive years.

The contrast with the United Kingdom is also instructive. The UK, ranked seventh, is the only G20 economy to receive a high overall rating. Its electricity system became coal-free in 2025, according to Ember. Ten G20 members, meanwhile, received very low ratings in the CCPI, underscoring the difficulty major economies face in aligning their energy systems and climate commitments. India’s 28th position places it ahead of several developing economies, including Brazil at 21st, the Philippines at 16th and Nigeria at 17th. But its ranking also shows that early gains in clean-energy capacity do not, on their own, secure a stronger overall climate performance.

India has set a net-zero emissions target for 2070. The CCPI assessment points to the need for clearer, time-bound pathways for reducing the use of coal, oil and gas, with sector-specific milestones and plans to support workers and regions dependent on fossil fuel industries. The immediate test is visible in the electricity system: India must make its growing renewable capacity more usable, while reducing the share of power generated from coal. Until that happens, the country’s clean-energy expansion will continue to coexist with a power mix heavily dependent on fossil fuels.