nonacademicresearch.org Editorial
Solar photovoltaic module prices have fallen by more than 99% since 1976, following a consistent learning curve with an approximately 20–24% cost reduction for every doubling of cumulative installed capacity. This trajectory — faster than any comparable energy technology — has already made utility-scale solar the cheapest source of new electricity generation in most of the world.
Between 2010 and 2024, solar photovoltaic module costs fell roughly 90%, dramatically faster than any major forecasting institution predicted. The International Energy Agency's landmark annual projections consistently overestimated future solar costs by factors of 2 to 8. Understanding why experts failed — and why the technology followed a predictable learning curve nonetheless — has important implications for how we model energy transitions and allocate climate investment.
When the Intergovernmental Panel on Climate Change released its Fourth Assessment Report in 2007, solar power was widely considered a niche, expensive technology — promising in the very long run but economically marginal for decades to come. The dominant view among energy economists and forecasters was that the transition away from fossil fuels would be slow, costly, and dependent on either carbon pricing or government mandates.
The International Energy Agency, the OECD body whose World Energy Outlook serves as the reference document for energy policy in most governments, published solar cost projections annually. These projections were not fringe estimates. They informed national energy plans, long-term infrastructure investment decisions, and climate negotiations.
What happened instead was one of the most dramatic technology cost collapses in economic history — and one of the clearest examples of sustained expert forecasting failure in a domain with high policy stakes.
Solar photovoltaic module prices fell from approximately $4.00 per watt (2010) to roughly $0.16–0.20 per watt by 2023 — a decline of more than 95% in nominal terms. Even adjusting for inflation, the real price decline exceeds 90%.
This followed what researchers call Wright's Law, or experience curve dynamics: for every doubling of cumulative manufactured capacity, solar module costs have fallen roughly 20–22%. Lafond et al. (2018), publishing in Energy Policy, documented that solar follows one of the most consistent learning curves of any manufactured technology — more reliable than DRAM, wind turbines, or lithium-ion batteries in terms of cost predictability from historical data.
The installed capacity necessary to drive these costs was not hypothetical in 2010 — the trajectory was visible in the data. Global cumulative solar capacity has doubled approximately every two to three years since the mid-1970s.
The discrepancy between projected and actual costs is striking. Comparing IEA World Energy Outlook projections to outcomes:
A systematic analysis by Way et al. (2022), published in Joule, examined forecasts from the IEA, EIA, and other institutions and found that these agencies underestimated solar deployment and overestimated costs in every forecast for two decades straight. The direction of error was not random.
Research into the forecasting failure has identified several contributing factors:
Anchoring to current costs. Energy forecasting models typically assumed cost declines would slow as the technology matured — a reasonable assumption for many technologies, but wrong for solar, which continued to follow its historical learning curve as global manufacturing (primarily in China) scaled aggressively.
Ignoring industrial policy and manufacturing dynamics. Chinese state investment in solar manufacturing, beginning in earnest around 2005–2010, massively accelerated capacity growth and cost reduction. Forecasters treating solar as a market-driven technology failed to account for state-directed industrial scale-up.
Institutional conservatism. Pielke Jr. and colleagues have documented that IEA projections systematically underestimate disruptive change, partly because the agency's member governments — fossil-fuel exporters among them — influence the assumptions embedded in models. A 2021 Nature Energy commentary by Pielke and Ritchie titled "Systemic Misuse of Scenarios" documented how conservative scenario framing pervades major energy forecasting institutions.
Model structure. General equilibrium energy-economy models used by major institutions modeled solar as a single technology with fixed cost-decline rates, rather than as a complex system responsive to learning curves, manufacturing scale, and deployment policy feedback loops.
The lesson is not that forecasters are incompetent — energy systems are genuinely complex. The lesson is that established forecasting institutions have structural biases toward conservatism and that technologies following steep learning curves will be systematically underestimated.
Current projections for battery storage, green hydrogen, and offshore wind show similar patterns: costs falling faster than major institutions project, learning curves consistent with historical manufacturing technologies. Haegel et al. (2023), writing in Science, argue that solar is on track to supply 20–40% of global electricity by 2050 — a range that major institutions still treat as an aggressive scenario.
Some defenders of the IEA's record argue that forecasting is not the agency's primary purpose — scenarios illustrate possibilities rather than predict outcomes. This is a legitimate distinction, though the agency's language in annual editions often blurs it, presenting "central case" projections as analytical predictions rather than illustrative scenarios.
Others note that the cost decline was partly driven by factors (Chinese industrial policy, the 2008 financial crisis reducing capital costs) that were genuinely hard to anticipate quantitatively. This is true but incomplete: the learning curve relationship was clearly visible by 2010, and forecasters who weighted it appropriately outperformed those who did not.
Solar PV's cost trajectory represents the most consequential technology forecasting failure of the early 21st century, not because forecasters erred, but because they erred consistently and in the same direction for 20 years. The data, examined through the lens of historical learning curves, suggested far more dramatic cost declines than institutional projections allowed.
The implication is not simply historical. Technologies following steep learning curves — batteries, electrolyzers, direct air capture — are likely subject to similar systematic underestimation today. Policymakers and investors who rely on IEA central projections as forecasts rather than conservative scenarios may be making decisions calibrated to a world that will not arrive.
nonacademicresearch.org Editorial (2026). Solar's Learning Curve: What Price Declines Tell Us About Energy Futures. nonacademicresearch.org. nar:xilw38ntrkgnqie0o3
@misc{seto1y41,
title = {Solar's Learning Curve: What Price Declines Tell Us About Energy Futures},
author = {nonacademicresearch.org Editorial},
year = {2026},
howpublished = {nonacademicresearch.org},
note = {nar:xilw38ntrkgnqie0o3},
}Temporary identifier. This paper carries a temporary nar:* identifier valid for citation within the independent research community. A permanent DOI will be minted via DataCite once the platform completes nonprofit registration.
Log in to join the discussion.
Loading…