Climate Change by the Numbers

Global surface temperature has risen approximately 1.1°C above the 1850–1900 baseline, with the rate of warming accelerating to roughly 0.2°C per decade since the early 1980s—more than triple the long-term average of 0.06°C per decade since 1850 [1]. The Intergovernmental Panel on Climate Change (IPCC) states that human activities, principally greenhouse gas emissions, have "unequivocally caused" this warming [2]. These are the opening numbers in one of the most consequential—and most bitterly contested—policy debates of the 21st century.

The contest is not primarily about whether the planet is warming. Thermometers settle that question. The real fights are about pace, cost, tradeoffs, and who pays: How quickly must the world decarbonize? What happens to the billions of people who still lack reliable electricity? Is the cure worse than the disease—or is the disease worse than anyone admits?

This report presents the data from both sides, names the people making each argument, and lets the numbers speak.

The Temperature Record and Human Attribution

The IPCC's Sixth Assessment Report (AR6), published in 2021, concluded that greenhouse gas emissions from human activities are responsible for approximately 1.07°C of the observed 1.1°C warming from 1850–1900 to 2010–2019 [2]. That leaves roughly 0.03°C attributable to natural variability—solar cycles, volcanic eruptions, and internal ocean-atmosphere oscillations. The finding upgraded the panel's confidence language from "extremely likely" (AR5, 2013) to "unequivocal" [2].

Climate scientist Michael Mann of the University of Pennsylvania has argued that these attribution findings understate the urgency, pointing to the acceleration of extreme weather events as evidence that impacts are arriving faster than models predicted [3]. NASA climatologist James Hansen, who first brought climate change to Congressional attention in 1988, has gone further, arguing that equilibrium climate sensitivity—the amount of warming expected from a doubling of CO₂—may be higher than the IPCC's central estimate of 3°C, and that Earth could be headed for warming well above 2°C even under moderate emissions scenarios [3].

On the other side of the attribution debate, some researchers emphasize remaining uncertainties. The IPCC itself acknowledges a wide range for equilibrium climate sensitivity, from 2.5°C to 4°C per doubling of CO₂, reflecting genuine scientific disagreement about cloud feedbacks, aerosol effects, and ocean heat uptake [2]. Judith Curry, a former chair of the School of Earth and Atmospheric Sciences at Georgia Tech, has argued that natural multidecadal ocean oscillations may account for a larger share of recent warming than the IPCC assigns, and that the range of possible outcomes extends further in the direction of lower sensitivity than mainstream summaries suggest [4]. While most published estimates cluster around the IPCC's range, the lower tail of the distribution—the possibility that warming will be less severe than median projections—remains a legitimate part of the scientific literature.

How Accurate Have Climate Models Been?

A 2020 study published in Geophysical Research Letters, led by Zeke Hausfather of Berkeley Earth, evaluated 17 climate model projections made between 1970 and 2007. Fourteen of the 17 were consistent with observed warming once actual greenhouse gas concentrations were used as inputs rather than the emissions scenarios originally assumed [5]. A key finding: many models that appeared to overpredict warming had simply assumed higher emissions than actually occurred. When corrected for actual emissions, they tracked observations closely [5].

NASA's Goddard Institute for Space Studies confirmed these findings, noting that the models "ichever were evaluated showed skill in projecting warming over the subsequent years" [6]. James Hansen's 1988 model, one of the earliest, projected warming of about 0.55°C by 2010—close to the roughly 0.6°C that was observed by that point [5].

This does not mean models are without limitations. The largest sources of uncertainty in current models involve cloud feedbacks (how clouds change as the planet warms, which can either amplify or dampen warming), aerosol forcing (the cooling effect of particulate pollution, which is difficult to measure precisely), and the behavior of ice sheets in Greenland and Antarctica under continued warming [2]. Regional precipitation projections remain far less reliable than global temperature projections. And the possibility of abrupt "tipping points"—such as the collapse of the Atlantic Meridional Overturning Circulation or rapid permafrost thaw—adds non-linear risks that models handle with varying degrees of sophistication [2].

Who Emits What: The Global Emissions Landscape

Source: EDGAR / European Commission Joint Research Centre

Data as of Mar 21, 2026CSV

In 2024, global greenhouse gas emissions reached approximately 53.2 billion metric tons of CO₂ equivalent (GtCO₂e) [7]. The distribution is starkly unequal:

• China: 29% of global emissions, the largest emitter by a wide margin [7]
• United States: approximately 11%, having declined from its historical peak [7]
• India: approximately 8%, with emissions rising 5.3% in 2024 alone—the fastest rate among major economies [7]
• EU-27: 5.9%, down from 6.1% in 2023 [7]
• Russia: approximately 5% [7]
• Indonesia: growing rapidly, with a 5.0% increase in 2024 [7]

Together, the top six emitters account for 61.8% of global emissions, 51.4% of global population, and 62.5% of global GDP [7].

The shift over three decades is striking. In 1990, the United States and Europe collectively dominated global emissions. China's share has roughly tripled since then, driven by industrialization that lifted hundreds of millions out of poverty. India's trajectory today mirrors China's from two decades ago [7].

Per capita, the picture reverses: Americans emit roughly 14 tons of CO₂ per person annually, compared to about 8 tons for Chinese citizens and under 2 tons for Indians [8]. This gap fuels the central equity argument in climate negotiations—developing nations contend that per-capita emissions, not national totals, should govern responsibility.

What Would 1.5°C or 2°C Require?

Limiting warming to 1.5°C above pre-industrial levels would require reducing global emissions by 40–50% below 2010 levels by 2030 and reaching net-zero CO₂ emissions by approximately 2050 [9]. For 2°C, the required reduction is roughly 20% below 2010 levels by 2030, with net-zero by around 2070 [9]. Neither trajectory is currently on pace. As of 2024, emissions were still rising.

The Human Toll: Deaths, Displacement, and the Adaptation Paradox

Climate-Related Deaths

Between 2015 and 2025, climate-related disasters caused approximately 305,156 deaths globally, according to the EM-DAT disaster database—down from 354,428 in the prior decade [10]. Heat waves have become the deadliest category, with an estimated 66,825 heat-related deaths in 2024 alone [10]. The Lancet Countdown estimates that average annual heat-related mortality reached 546,000 deaths between 2012 and 2021, a 63% increase from 1990–1999 levels [10].

Storm-related deaths fell sharply—from 184,237 in 2005–2014 to 36,652 in 2015–2025—largely due to improved early warning systems and disaster preparedness [10].

Alex Epstein, author of Fossil Future and founder of the Center for Industrial Progress, uses a different frame for these statistics. He argues that climate-related disaster death rates have fallen by 98% over the past century, and that this decline is itself a product of fossil-fuel-powered infrastructure: air conditioning, sturdy buildings, flood barriers, mechanized agriculture, and transportation networks that enable evacuation [11]. In Epstein's analysis, the roughly 3 billion people worldwide who still consume less electricity than an American refrigerator face greater immediate danger from energy poverty than from gradual warming [11].

The World Economic Forum projects that climate change could cause 14.5 million additional deaths by 2050 through its effects on food security, infectious disease, and extreme heat [12]. Epstein and Bjorn Lomborg, president of the Copenhagen Consensus Center, both argue that these projections must be weighed against the deaths caused by energy poverty itself. The World Health Organization estimates that household air pollution from burning biomass (wood, dung, crop waste) for cooking—a direct consequence of lacking access to modern energy—kills approximately 3.2 million people per year [13]. Unsafe water and sanitation, problems solvable with reliable energy infrastructure, kill another 1.4 million annually [13].

This is the core of the conservative energy argument: the world's poorest people need cheap, reliable energy now, and restricting fossil fuel access in the name of emissions reduction imposes costs on those least able to bear them.

The counterargument, advanced by researchers such as Marshall Burke of Stanford University, is that climate change itself disproportionately harms the poorest populations—through crop failures, water stress, and heat exposure—and that failing to limit warming condemns these same populations to compounding risks [14].

Displacement

An unprecedented 83.4 million people were living in internal displacement at the end of 2024, according to the Internal Displacement Monitoring Centre (IDMC) [15]. Of the 46.9 million new internal displacements recorded in 2023, 56% were triggered by disasters, many of them weather-related [15]. Sub-Saharan Africa, hosting 46% of the world's internally displaced persons, was the most affected region [15].

An average of 124 million people were affected by disasters annually between 2014 and 2023, a 75% increase over the preceding decade [15]. Projections for coastal populations are particularly stark: the World Bank has estimated that without action, 216 million people could be forced to migrate within their own countries by 2050 due to climate impacts [16].

The Economics: What Climate Change Costs

Swiss Re Institute's stress-test analysis projects that under a 3.2°C warming scenario, the global economy could lose up to 18% of GDP by 2050—equivalent to roughly $23 trillion annually at current output levels [17]. Under a 2°C scenario, losses would be approximately 11% of GDP. Meeting the Paris Agreement targets (below 2°C) would limit losses to roughly 4% of GDP [17].

Regional impacts vary dramatically. China stands to lose up to 24% of GDP under the severe scenario, while the United States, Canada, and the UK would each see approximately 10% losses. Europe would suffer slightly more at 11%, though northern economies like Finland and Switzerland face lower exposure at around 6% [17].

Current annual economic losses from four major weather perils—floods, tropical cyclones, winter storms, and severe thunderstorms—already total approximately $200 billion globally [17].

How Do Actual Costs Compare to Past Projections?

A 2024 study published in Nature found that the economic costs of climate change are "substantially higher" than projected by earlier integrated assessment models. The study estimated that global income will be reduced by approximately 19% by 2049 compared to a scenario without climate change, with costs six times larger than the price of limiting warming to 2°C [18]. This suggests that models produced 10–20 years ago systematically underestimated climate damages, particularly from extreme events and agricultural disruption.

Lomborg has contested these higher-end damage estimates, arguing that they rely on assumptions about limited human adaptation. In his framework, societies historically adapt to changing conditions—farmers switch crops, engineers build sea walls, populations migrate—and models that hold adaptation constant overstate future damages [19]. Lomborg's Copenhagen Consensus Center advocates for spending $100 billion annually on energy R&D rather than on emissions reductions, arguing this approach would produce larger returns per dollar [19].

The Grantham Research Institute at the London School of Economics has published detailed critiques of Lomborg's cost-benefit methodology, arguing that his estimates undercount extreme event damages and apply discount rates that devalue future harms [20].

The Energy Transition: Costs, Sources, and Tradeoffs

Source: U.S. Energy Information Administration (EIA)

Data as of Mar 21, 2026CSV

U.S. Electricity Generation by Source

The composition of U.S. electricity generation has shifted substantially. In 2025, natural gas generated 1,807 terawatt-hours (TWh), accounting for roughly 41% of total generation. Coal produced 737 TWh (17%), down from 1,240 TWh (30%) in 2016. Nuclear contributed 785 TWh (18%), while wind reached 464 TWh (10%) and solar hit 296 TWh (7%)—the latter nearly quadrupling since 2020 [21].

Coal's decline has been driven primarily by market forces: cheap natural gas from the shale revolution undercut coal on price before any climate policy intervened. The retirement of coal plants accelerated after 2015, with over 100 GW of coal capacity shuttered in the United States in the past decade [21].

What Do Renewables Actually Cost?

The levelized cost of energy (LCOE)—the total lifecycle cost per unit of electricity, including construction, fuel, and maintenance—has plummeted for renewables. According to Lazard's 2025 LCOE+ report, utility-scale solar averages $66/MWh globally (range: $28–$117/MWh), while onshore wind ranges from $37–$86/MWh [22]. In 2024, 91% of newly commissioned utility-scale renewable capacity delivered power at a lower cost than the cheapest new fossil fuel alternative [22].

These figures represent unsubsidized costs. The Inflation Reduction Act's clean energy tax credits further reduce effective costs for U.S. projects, though the potential rollback of these credits under the current administration introduces investment uncertainty [22].

Battery storage costs have dropped 93% between 2010 and 2024, from $2,571/kWh to $192/kWh for fully installed systems [23]. The U.S. added a record 57.6 GWh of new battery storage capacity in 2025, a 30% increase over 2024 [23]. Long-duration energy storage deployments rose 49% in 2025 [23].

Grid Reliability: The Intermittency Problem

The central technical challenge for renewables remains intermittency: solar panels produce no electricity at night, and wind turbines stop when the wind dies. Vaclav Smil, Distinguished Professor Emeritus at the University of Manitoba, has written extensively about the physical constraints of energy transitions, emphasizing that no major energy transition in history has occurred in less than 50–70 years and that electricity represents only about 20% of final energy consumption—meaning that decarbonizing the grid, while necessary, addresses only a fraction of total emissions [24].

Current battery storage, while growing rapidly, provides hours of backup, not the days or weeks needed to handle extended periods of low wind and solar output (known as "Dunkelflaute" in German energy planning). Natural gas plants currently serve as the primary backup for renewable intermittency, which is why natural gas consumption has grown even as coal declines [21].

Grid operators in Texas (ERCOT) and California (CAISO) have experienced reliability challenges during extreme weather events, raising questions about whether a grid heavily dependent on variable renewables can maintain the 99.97% reliability standard that modern economies require. The 2021 Texas grid failure, which caused over 200 deaths, involved failures across multiple generation sources—gas, wind, and nuclear—but highlighted the systemic risks of insufficient dispatchable capacity [25].

Nuclear: The Unloved Solution

Nuclear power is the single largest source of zero-carbon electricity in the United States, generating 785 TWh in 2025—more than wind and solar combined [21]. Globally, nuclear provides about 10% of electricity generation and roughly 25% of all low-carbon electricity.

Nuclear faces opposition from two directions. Environmental groups have historically opposed it due to concerns about radioactive waste storage (no permanent U.S. repository exists), accident risk (Chernobyl, Fukushima), and weapons proliferation. Market participants oppose it because new nuclear plants have suffered catastrophic cost overruns: the Vogtle Units 3 and 4 in Georgia, the only new U.S. reactors completed in decades, came in at roughly $35 billion—more than double the original estimate—and seven years behind schedule [26].

Advocates such as the Breakthrough Institute argue that these cost overruns reflect regulatory and institutional failures rather than inherent technological limitations, and point to South Korea and China, where standardized reactor designs are built on time and on budget [26]. Small modular reactors (SMRs) are proposed as a lower-cost alternative, though none have yet achieved commercial-scale deployment in the United States.

Germany's Energiewende: Cautionary Tale or Misread Lesson?

Germany's energy transition (Energiewende) offers a contested case study. Germany shut down its last nuclear reactors in April 2023, completing a phase-out motivated by the Fukushima accident in 2011 [27]. The results, depending on who is interpreting them:

Critics point out that Germany's residential electricity prices are among the highest in Europe, that the nuclear exit increased reliance on Russian natural gas (a dependency exposed by the 2022 invasion of Ukraine), and that overall energy costs contributed to Germany's 2023–2024 economic recession. The estimated cost of the Energiewende through the late 2030s could reach €1 trillion, with feed-in tariff subsidies alone accounting for €680 billion [27].

Defenders note that in the year following the nuclear exit, Germany achieved record renewable output, coal usage fell to its lowest level in 60 years, and overall emissions dropped. Wholesale electricity prices also declined in 2024 as renewable supply grew [28]. They argue that the nuclear phase-out, while debatable, did not produce the grid collapse or emissions spike that critics predicted.

The honest assessment: Germany demonstrated that rapid renewable deployment is physically possible, but also that prematurely closing reliable zero-carbon capacity (nuclear) while maintaining fossil fuel backup imposes real costs and contradictions.

China's Paradox: Leading in Both Renewables and Emissions

China installed 360 GW of new wind and solar capacity in 2024 alone—more than any other country has ever added in a single year. Solar capacity grew 45.2%, adding 277 GW [29]. By September 2025, renewables accounted for 59.1% of China's installed power capacity [29].

Simultaneously, construction began on 94.5 GW of new coal-fired power plants in 2024, a 10-year high. In 2025, another 83 GW of new coal construction started [29]. China accounts for 29% of global greenhouse gas emissions and continues to increase them, even as it builds more clean energy capacity than the rest of the world combined [7].

How can both be true? China's electricity demand is growing so rapidly—driven by industrialization, urbanization, air conditioning adoption, and electric vehicle charging—that even massive renewable additions cannot cover all new demand. Coal plants are increasingly being built as backup capacity rather than baseload, with utilization rates hovering around 51% since 2025 [29]. Many analysts expect China's power sector emissions to peak before 2030, as renewable growth overtakes demand growth [29].

The situation illustrates Smil's broader point about energy transitions: they are additive before they are substitutive. New energy sources layer on top of old ones before eventually displacing them [24].

Source: U.S. Energy Information Administration (EIA)

Data as of Mar 21, 2026CSV

Energy Poverty and the Equity Argument

Approximately 685 million people worldwide lacked access to electricity as of 2023, and 2.1 billion lacked access to clean cooking fuels—relying instead on wood, charcoal, and animal dung, with severe health consequences [13]. The question of how these populations gain energy access sits at the center of the climate policy divide.

Lomborg's argument, drawing on the work of economist William Nordhaus, is that the optimal climate policy balances mitigation costs against adaptation and damages, and that aggressive near-term decarbonization is not cost-effective [19]. He points to the Copenhagen Consensus analysis showing that each dollar spent on climate mitigation yields approximately $0.02–$0.11 in avoided damages, while investments in nutrition, disease prevention, and education yield returns of $15–$40 per dollar [19].

The counterargument, made by economists Nicholas Stern (London School of Economics) and Joseph Stiglitz (Columbia University), is that Nordhaus's framework uses discount rates that effectively devalue the welfare of future generations, and that once catastrophic risk is properly accounted for, aggressive mitigation becomes the economically rational choice [20]. The 2006 Stern Review estimated that the costs of inaction would exceed the costs of mitigation by a factor of 5–20 [20].

Carbon Intensity of Poverty Reduction

A critical data point in this debate: the carbon intensity of lifting people out of poverty varies enormously by context. China's industrialization, powered largely by coal, produced approximately 6–8 tons of CO₂ per capita during its phase of rapid poverty reduction. India's current trajectory is somewhat less carbon-intensive, at roughly 2 tons per capita, reflecting both different industrial structures and the falling cost of renewables [8]. In Sub-Saharan Africa, where energy poverty is most severe, leapfrogging to solar microgrids and distributed renewables could theoretically enable poverty reduction at far lower carbon intensity—but this depends on technology transfer and financing that has not yet materialized at scale.

Over the past decade, more than 300 million people gained electricity access globally, predominantly in South and Southeast Asia [30]. The countries where these gains occurred have, in most cases, increased their total emissions. As of 2024, approximately 67% of the global population lives in countries where total emissions rose over the prior decade—a category that includes China, India, Indonesia, and most of Sub-Saharan Africa [7]. The remaining 33% lives primarily in the United States, Europe, and Japan, where emissions have declined [7].

The Cost-Benefit Ledger: Act Now or Adapt Later?

The central economic question is whether it is cheaper to prevent warming or to cope with its consequences. The answer depends heavily on assumptions about discount rates, adaptation capacity, and the probability of catastrophic outcomes.

The case for aggressive intervention: A 2024 study in Nature estimated that the global economy faces income losses of approximately 19% by 2049 under current trajectories, with costs six times larger than the investment required to limit warming to 2°C [18]. The IPCC's Special Report on 1.5°C found that limiting warming to 1.5°C rather than 2°C would reduce climate-related risks to food security, water supply, health, and economic growth, and would lower adaptation costs from approximately $128 billion per year (at 3°C) to $63 billion per year (at 1.5°C) for global agriculture alone [9].

The case for adaptation and gradual transition: Lomborg argues that the Paris Agreement, even if fully implemented, would reduce warming by less than 0.05°C by 2100—a negligible climate benefit at enormous economic cost [19]. Epstein points out that global life expectancy has roughly doubled over the past century, extreme poverty has fallen from 42% to under 10%, and climate-related disaster deaths have plummeted by 98%, all during a period of rising emissions and fossil fuel use [11]. In this framing, cheap energy is the most powerful adaptation tool ever devised, and restricting it to reduce emissions is a policy that privileges speculative future harms over concrete present benefits.

Economist Richard Tol of the University of Sussex, one of the lead authors of the IPCC's Working Group II, has published analyses suggesting that moderate warming (1–2°C) produces net economic benefits in some regions through longer growing seasons, reduced heating costs, and CO₂ fertilization of crops, with net costs emerging only at higher warming levels [31]. This finding is contested by other economists who argue Tol's analysis underweights extreme event damages and non-market costs such as biodiversity loss.

The Paris Agreement: Does Participation Matter?

President Trump signed an executive order withdrawing the United States from the Paris Agreement on January 20, 2025, with the withdrawal taking effect on January 27, 2026 [32]. The U.S. had committed to a 61–66% reduction in net greenhouse gas emissions below 2005 levels by 2035 [32].

The Paris Agreement's commitments (Nationally Determined Contributions, or NDCs) are not legally binding—there are no penalties for failure to meet targets [32]. This is by design: the United States and China both insisted on this structure during negotiations. Modeling by the Climate Action Tracker suggests that U.S. non-participation could increase projected warming from 2.1°C to 2.2°C and reduce the probability of staying below 2°C from 34% to 27% [32].

The practical question is whether the agreement's "naming and shaming" mechanism—public comparison of commitments and performance—actually drives emissions reductions. The evidence is mixed. EU emissions have declined consistently, but much of that decline reflects deindustrialization and outsourcing of manufacturing to countries like China, which then bear the emissions burden [7]. When emissions are measured on a consumption basis (attributing emissions to the country that consumes the goods, not the one that produces them), the picture shifts significantly: developed countries' reductions are smaller, and developing countries' increases are larger.

The 194 countries remaining in the agreement represent approximately 90% of global emissions [32]. Whether the framework survives the withdrawal of its second-largest emitter—and whether non-binding commitments can produce binding results—remains an open question.

Uncertainty and Honest Accounting

Several areas of genuine scientific and economic uncertainty deserve explicit acknowledgment:

Climate sensitivity range: The IPCC's "likely" range of 2.5–4°C for equilibrium climate sensitivity means that outcomes at both the lower and upper bounds remain plausible. A world that warms 2.5°C by 2100 looks qualitatively different from one that warms 4°C [2].

Tipping point thresholds: The temperature thresholds at which major Earth system tipping points activate—such as Amazon rainforest dieback, West Antarctic ice sheet collapse, or permafrost carbon release—are not precisely known and may not follow linear trajectories [2].

Discount rates: The appropriate rate at which to discount future climate damages against present costs is not a scientific question but a philosophical one, and reasonable people can disagree. Nordhaus's preferred rate of roughly 4% implies that damages decades from now are relatively unimportant today. Stern's rate of 1.4% treats future generations as roughly equal in moral weight to the present. This single parameter drives enormous differences in recommended policy [20].

Adaptation capacity: How effectively societies will adapt to warming—through technology, migration, agricultural innovation, and infrastructure—is inherently uncertain and varies enormously by wealth level. Rich countries will adapt far more effectively than poor ones, which means that the distributional consequences of climate change may matter more than the aggregate costs.

Degrowth versus green growth: A minority of climate advocates, particularly in European academic circles, argue that no amount of technological substitution can decouple economic growth from emissions, and that only deliberate reduction in material consumption can achieve climate targets. This "degrowth" position is rejected by most mainstream economists and policymakers but draws on real data about the correlation between GDP growth and emissions growth that "green growth" advocates have not yet fully resolved at global scale [33].

Source: U.S. Energy Information Administration (EIA)

Data as of Mar 21, 2026CSV

Source: Swiss Re Institute

Data as of Mar 21, 2026CSV

Where the Debate Actually Stands

The political framing of climate change—one side treating it as an existential emergency requiring immediate economic transformation, the other treating it as an overblown threat used to justify economic self-harm—obscures the genuine tradeoffs that the data reveals.

The temperature data is not in serious dispute. The planet is warming, humans are the primary cause, and the trend will continue without intervention. The economic damage estimates, while uncertain in magnitude, consistently show that higher warming produces higher costs. The IPCC's attribution science is robust, and models from decades ago have largely held up against observations.

At the same time, climate-related disaster deaths have fallen dramatically over a century, cheap energy has been the single most powerful tool for reducing poverty and extending human life, energy transitions take decades even under the most optimistic assumptions, and the world's poorest populations have legitimate claims to affordable energy that aggressive decarbonization timelines may conflict with. Grid reliability is not a hypothetical concern—it is an engineering constraint that current storage technology has not yet fully solved. And the Paris Agreement's non-binding structure raises genuine questions about whether international frameworks can deliver results at the pace that climate science suggests is necessary.

The data does not resolve these tensions. It sharpens them. The honest position is that both the costs of inaction and the costs of action are real, that they fall on different populations and different timescales, and that the policy question—how fast, how much, who pays—does not have a single answer that the numbers compel.