Climate Change by the Numbers: The Data Both Sides Don't Want You to See

The climate debate generates more heat than light. One side treats every weather event as proof of imminent apocalypse; the other treats every cold snap as a refutation of physics. Meanwhile, the data tells a story more uncomfortable than either camp prefers: the planet is warming faster than most people realize, the proposed solutions are more expensive and less reliable than advocates admit, and the costs of inaction are more severe than skeptics acknowledge.

This is what the numbers actually show.

The Temperature Record: Where We Stand

The planet is unambiguously warming. Berkeley Earth's analysis places the 2025 global mean temperature at 1.44 ± 0.09°C above the 1850–1900 pre-industrial baseline, making it the third warmest year on record behind 2024 and 2023 [1]. NASA GISS reported 2025 at 1.19°C above the 1951–1980 average [2]. The top three warmest years in recorded history have now occurred consecutively.

The warming rate since 1970 has proceeded at roughly 0.20°C per decade [1]. To put this in geological context, the rate of warming over the past 40 years is at least ten times faster than any comparable period during the last deglaciation—the most rapid natural warming event in the recent geological record.

There is structural uncertainty of roughly ±0.1°C between major datasets (NASA GISS, NOAA GlobalTemp, HadCRUT, Berkeley Earth), driven primarily by different methods of correcting historical ocean temperature measurements—from wooden buckets to engine intakes to modern buoys [1]. This uncertainty is real but small relative to the overall signal. No major dataset disputes that the planet has warmed more than 1°C since the pre-industrial era.

The UAH satellite record, maintained by Roy Spencer and John Christy at the University of Alabama, shows less warming in the lower troposphere—0.30°C anomaly for December 2025 [3]. This discrepancy between surface and satellite records has been a persistent point of debate, though the divergence has narrowed as both datasets have undergone corrections over the decades.

Who Emits What: The Equity Question

The United States has contributed roughly 21% of cumulative CO2 emissions since 1850—approximately 560 GtCO2—making it the largest historical emitter by far [4]. China is second at roughly 11%, followed by Russia (7%), Brazil (5%), and Indonesia (4%) [4]. When land-use emissions from deforestation are included, Brazil and Indonesia crack the top ten—a fact often omitted from fossil-fuel-only analyses.

But cumulative emissions and current emissions tell very different stories. China now emits more CO2 annually than any other country and accounted for roughly 32% of global fossil fuel emissions in 2024 [5]. The United States is second at approximately 13%, followed by India at 8% and the EU at 7%.

Per capita, the picture shifts again. The US emitted approximately 13.8 tonnes of CO2 per person in 2023; China emitted 9.2 tonnes; India emitted approximately 2 tonnes [6]. When adjusted for consumption—accounting for the carbon embedded in traded goods—US per-capita emissions increase slightly (Americans import carbon-intensive goods), while China's decrease modestly (China exports carbon-intensive manufactured goods). India's consumption-based emissions remain among the lowest of any major economy.

This is the core equity dilemma: the countries that contributed least to the problem historically are the ones least able to afford the transition, and the ones whose citizens still need vastly more energy to escape poverty.

Source: Global Carbon Budget 2025

Data as of Mar 20, 2026CSV

How Accurate Are the Models?

Climate models have performed better than critics allege and worse than advocates imply.

The IPCC's First Assessment Report in 1990 projected warming of approximately 0.3°C per decade under a "business as usual" scenario. Observed warming has been approximately 0.2°C per decade—within the uncertainty range but at the lower end [7]. The 2001 Third Assessment projected 1.4–5.8°C of warming by 2100; the 2021 AR6 narrowed this to 1.0–5.7°C depending on emissions pathway, with a "best estimate" equilibrium climate sensitivity of 3°C (likely range 2.5–4.0°C) [7].

The track record suggests models capture the central tendency of warming reasonably well but have persistent issues. McKitrick and Christy (2020) documented a warm bias in CMIP6 models throughout the tropical troposphere—the models predict a "hot spot" at roughly 10 km altitude that observational data does not clearly show [8]. Judith Curry has argued that the IPCC's reliance on models that cannot reproduce the Medieval Warm Period undermines confidence in long-range projections [8].

On the other hand, the recent narrowing of the equilibrium climate sensitivity range to 2.5–4.0°C (from the longstanding 1.5–4.5°C) represents genuine progress [7]. And the basic physics—CO2 absorbs infrared radiation, more CO2 means more retained heat—is not model-dependent. It is laboratory-verified.

What does this mean for 2050 projections? The honest answer is that a range of 1.5–3°C of total warming by mid-century is plausible depending on emissions trajectory, with the spread driven less by physics uncertainty and more by whether humanity actually reduces emissions.

Attribution: How Much Is Us?

The IPCC AR6 concluded that human activities have caused approximately 1.07°C of the observed 1.1°C warming from 1850–1900 to 2010–2019 [9]. The confidence language has escalated dramatically across reports: AR4 (2007) said warming was "very likely" (>90%) due to human activity; AR5 (2013) said "extremely likely" (>95%); AR6 (2021) said it is "unequivocal" that human influence has warmed the atmosphere, ocean, and land [9].

Attribution science for individual extreme weather events has matured rapidly. A meta-analysis found that 70% of 405 studied extreme weather events were made more likely or more intense by human-caused climate change [10]. But this statistic cuts both ways: 30% showed no detectable human influence or were actually made less likely—a finding rarely mentioned in media coverage.

Natural variability remains significant on shorter timescales. El Niño and La Niña cycles, volcanic eruptions, and solar variability can produce year-to-year temperature swings of 0.1–0.3°C. The 2023–2024 temperature spike, partly driven by a strong El Niño, "appears to have deviated significantly from the previous trend," with Berkeley Earth assigning less than a 1-in-100 probability of occurring from natural variation alone [1]. The subsequent cooling in late 2025 and into 2026 is consistent with a shift to La Niña conditions.

The Cost of Action vs. Inaction

McKinsey estimates the net-zero transition requires $9.2 trillion in annual spending on physical assets—$3.5 trillion more per year than current levels—totaling roughly $275 trillion between 2021 and 2050 [11]. The IEA puts the figure at around $4.8 trillion per year in energy investment alone over the next decade, up from $3.3 trillion today [12].

Are these costs worth it? The answer depends on assumptions about discount rates—how much you value future damages versus present costs—and climate sensitivity. William Nordhaus, whose work earned a Nobel Prize, calculated that the optimal carbon tax starts at roughly $31–53 per tonne and rises over time, reflecting substantial net benefits from reducing emissions now rather than waiting [13]. His models estimate the cost of delaying action by 50 years at approximately $4 trillion in today's dollars.

Bjørn Lomborg counters that Nordhaus's own models show aggressive near-term emissions cuts are less cost-effective than moderate carbon pricing combined with heavy investment in R&D to make clean energy cheaper. Lomborg argues the Paris Agreement targets would cost $1–2 trillion per year while reducing temperatures by only 0.05°C by 2100 [14]. The Grantham Research Institute has challenged Lomborg's cost figures as inflated, arguing he systematically doubles policy cost estimates [14].

The UK's Climate Change Committee reported in March 2026 that the total net cost of reaching net zero by 2050 is less than the cost of a single fossil fuel price shock of the magnitude seen in 2022 [15]. This framing highlights an often-overlooked point: fossil fuel dependence carries its own massive economic risks.

Under a 3°C warming scenario, estimated annual climate damages by 2100 range from $5 to $25 trillion depending on the model—far exceeding transition costs [11]. But these estimates carry enormous uncertainty. The damage functions in economic models are their least reliable component, often extrapolating from limited historical data.

Tipping Points and Timelines

At current emissions trajectories, the 1.5°C threshold will likely be crossed between 2026 and 2042 [16]. The remaining carbon budget for a 50% chance of staying below 1.5°C is approximately 170 billion tonnes of CO2—roughly four years at current emission rates [5].

The tipping points that worry scientists most:

Greenland and West Antarctic ice sheets: Committed to partial disintegration above approximately 1.5–2°C, though full collapse would take centuries. The contribution to sea-level rise is the primary concern [16].

Amazon rainforest dieback: Climate science alone suggests risk at 3–5°C, but when deforestation and biodiversity loss are factored in, the threshold drops to 1.5–2°C [16].

AMOC (Atlantic Meridional Overturning Circulation) collapse: The most contested tipping point. A 2025 study suggested collapse could begin as early as the 2060s, but the IPCC's assessment is that collapse in the 21st century remains unlikely under most scenarios [16].

At 2°C of peak warming, tipping risks are estimated at roughly 15%. At 3°C, there is a 66% probability that at least one major tipping element loses stability [16]. Half of identified tipping points could be triggered at 2°C or less.

The Energy Transition: What's Actually Happening

Source: US Energy Information Administration

Data as of Mar 20, 2026CSV

The US electricity grid is transforming, but unevenly. In 2025, natural gas generated 1,807 TWh (41% of US electricity), coal produced 737 TWh (17%—down from 27% in 2018), nuclear provided 785 TWh (18%), wind contributed 464 TWh (10%), and solar generated 296 TWh (7%) [17].

Solar generation has quadrupled since 2019 (from 72 TWh to 296 TWh). Wind has grown 57% over the same period. These are genuine achievements. But coal's decline has been primarily replaced by natural gas, not renewables—a fact that complicates the decarbonization narrative. Natural gas emits roughly half the CO2 of coal per unit of electricity, so the switch has reduced emissions, but it has not eliminated them.

Nuclear power—the largest source of zero-carbon electricity in the US—has remained essentially flat at around 780 TWh for over a decade [17]. It produces more carbon-free electricity than wind and solar combined. Yet it faces opposition from environmentalists (waste, accident risk) and markets (massive cost overruns at projects like Vogtle, which came in $17 billion over budget). The tension is stark: if you are serious about decarbonization, dismissing nuclear is difficult to justify on the data.

Germany's Cautionary Tale

Germany's Energiewende offers a real-world case study in what happens when you shut down nuclear before alternatives are ready. After Fukushima in 2011, Germany began closing its 17 reactors, completing the process in April 2023 [18]. The resulting gap was filled partly by renewables—but also by Russian natural gas, creating a strategic vulnerability that exploded when Russia invaded Ukraine.

German wholesale electricity prices spiked to €469 per megawatt-hour in August 2022 [18]. They have since fallen but remain elevated compared to pre-crisis levels. Germany now has among the highest electricity prices in Europe. The lesson is not that renewables failed—Germany's solar and wind have grown substantially—but that eliminating a reliable zero-carbon source before replacements were in place imposed enormous costs and created geopolitical dependencies.

China's Paradox

China installed 360 GW of wind and solar capacity in 2024 alone—more than the entire existing US renewable capacity [19]. Total installed wind and solar now exceeds 1.4 TW. Yet China simultaneously accounted for 93% of new global coal plant construction in 2024, bringing 78 GW of new coal capacity online [19].

This is not cognitive dissonance. China is pursuing a "build before breaking" strategy: coal provides dispatchable backup while renewables scale up. The result is that China's emissions intensity per unit of electricity fell 5% year-over-year in 2024, and total emissions growth slowed to 0.4% in 2025—the slowest pace in years [5][19]. China may be approaching peak emissions, but it has not reached it yet. Both claims—that China leads the world in renewables and that China leads the world in emissions growth—are simultaneously true.

Grid Reliability: The Intermittency Problem

The capacity factor question is central and underreported. In 2024, solar's average annual capacity factor was 23% and wind's was 34% [20]. Natural gas plants exceed 90% availability. Nuclear typically runs at 90%+ capacity factors. This means you need roughly four times the nameplate solar capacity to match the actual output of an equivalent gas plant.

Battery storage is growing fast—24 GW projected for installation in 2026 alone, with total US capacity approaching 40 GW by year-end [20]. Large-scale storage can reduce renewable output variability by up to 80%. But current battery technology provides hours of storage, not days or weeks. The "Dunkelflaute" problem—extended periods of low wind and cloud cover, common in northern winters—remains unsolved at scale.

The honest assessment: renewables plus storage can handle most grid conditions most of the time. But the last 5–10% of reliability—the extreme peaks, the prolonged weather events—still requires dispatchable generation. Today, that means gas or nuclear. Pretending otherwise is not a climate strategy; it is a wish.

Source: US Energy Information Administration

Data as of Mar 20, 2026CSV

Climate Displacement and Human Cost

The IPCC projects that climate change could displace up to 216 million people within their own countries by 2050 across six world regions, under a high-emissions scenario [21]. More than one billion people could face coastal-specific climate hazards by mid-century.

But isolating climate as a migration driver is fiendishly difficult. Current displacement is overwhelmingly driven by conflict, economic factors, and governance failure—with climate acting as a "threat multiplier" rather than a standalone cause. Approximately half the global population already experiences severe water scarcity for at least one month per year [21], driven by a combination of climate change, population growth, agricultural practices, and infrastructure failures. Attributing precise percentages to each factor is beyond current science.

The populations most vulnerable by 2040 are concentrated in sub-Saharan Africa, South Asia, and small island developing states—regions characterized by poverty, governance challenges, and climate-sensitive livelihoods [21]. The cruel irony: these populations contributed least to the problem and have the least capacity to adapt.

Alex Epstein argues in Fossil Future that energy poverty is a more immediate killer than climate change, and that restricting fossil fuel access in developing nations condemns billions to preventable deaths from indoor air pollution, lack of refrigeration, and inability to heat or cool homes. The data supports the premise: approximately 2.3 billion people still cook with solid fuels, causing nearly 4 million premature deaths annually from household air pollution. The question is whether the solution is more fossil fuels—with their climate consequences—or accelerated deployment of clean energy in developing nations.

What Actually Works: Cost-Effectiveness

The historical record of energy transitions is humbling. Vaclav Smil has documented that previous transitions—wood to coal, coal to oil, oil to gas—each took 50–70 years [22]. The proposed fossil-to-clean transition is being attempted in 25–30 years, a pace without historical precedent. This does not make it impossible, but it should temper expectations.

On cost-effectiveness, the data is clearer than the politics:

Carbon taxes are broadly considered by economists as the most efficient tool. At $50/tonne, a carbon tax is estimated to reduce emissions 20–30% within a decade while generating revenue that can offset regressive impacts. The challenge is political viability—carbon taxes are unpopular with voters and industries alike.

Renewable subsidies have driven dramatic cost reductions: solar module costs have fallen 99% since 1976 and roughly 90% since 2010. But subsidies become less cost-effective as deployment scales—the marginal ton of CO2 reduced becomes more expensive as the easiest substitutions are completed.

Adaptation spending is chronically underfunded relative to its returns. Every dollar spent on climate adaptation in developing countries yields $4–10 in avoided damages, according to the Global Commission on Adaptation. Yet adaptation receives roughly one-tenth the funding of mitigation.

R&D investment in breakthrough technologies—advanced nuclear, long-duration storage, direct air capture, enhanced geothermal—may offer the highest long-term returns but with the least certainty. Lomborg's argument that prioritizing R&D over near-term deployment is more cost-effective has theoretical support but assumes future breakthroughs that are not guaranteed.

The Paris Agreement: Does It Matter?

The Paris Agreement's commitments are non-binding, and no major emitter is on track to meet them. Global emissions reached a record 38.1 GtCO2 from fossil fuels in 2025, with atmospheric CO2 at 425.7 ppm—52% above pre-industrial levels [5]. The gap between pledged reductions and actual trajectories continues to widen.

US participation has toggled on and off across administrations, raising the question of whether the agreement's primary value is atmospheric or diplomatic. The strongest case for Paris is that it creates a framework for ratcheting up ambition and provides transparency mechanisms. The strongest case against is that it creates an illusion of action that substitutes for actual emissions reductions.

What the data shows is that national emissions trajectories are driven far more by energy economics—the relative cost of gas versus coal versus renewables, the availability of capital for infrastructure—than by international agreements. The countries actually reducing emissions are doing so primarily because clean energy has become economically competitive, not because of treaty obligations.

The Bottom Line

The planet has warmed 1.44°C above pre-industrial levels. Human activity is unequivocally the dominant cause. Fossil CO2 emissions hit a new record of 38.1 billion tonnes in 2025 with no peak in sight. The remaining carbon budget for 1.5°C is approximately four years of current emissions.

The transition to clean energy is real but incomplete. Solar and wind are growing rapidly; nuclear remains indispensable but politically orphaned; battery storage is improving but cannot yet solve multi-day intermittency; natural gas is doing most of the actual decarbonization work that coal retirement creates.

The costs of transition are enormous—$275 trillion through 2050 by one estimate. The costs of inaction are likely larger. Neither figure is precise enough to build policy on with high confidence. The honest position is that this is a problem of risk management under deep uncertainty, and the data supports neither panic nor complacency.

What the numbers demand is a seriousness that the political debate has not yet achieved: acknowledging that decarbonization is necessary, that it will be expensive and disruptive, that developing nations need more energy not less, that nuclear must be part of the portfolio, that grid reliability cannot be hand-waved away, and that the timeline for this transition will be measured in decades, not years.

The data is clear. The tradeoffs are hard. Anyone who tells you otherwise is selling something.