Solar Farms
The cost of solar electricity fell ninety percent in fourteen years. Nothing in the history of energy has ever done that. Most people have not updated.
The effect compounds within years. Put it in place and it keeps working.
Origins
The photovoltaic effect was noticed by a nineteen-year-old Frenchman, Edmond Becquerel, in 1839. It sat as a curiosity for over a century.
The first practical silicon solar cell was made at Bell Labs in 1954, and it was reported at the time as a marvel: electricity, directly from sunlight, with no moving parts, no fuel, and no noise. It was also preposterously expensive, and its first serious application was the only place where cost did not matter and fuel could not be delivered — space. Satellites have run on solar panels since Vanguard 1 in 1958, which means the technology has been working continuously, unattended, in the harshest environment there is, for nearly seventy years.
On Earth it remained a niche for decades: calculators, remote telemetry, a few idealists off-grid. It was the technology of the future and everyone was confident it would remain so.
What nobody adequately anticipated was the learning curve. Every doubling of cumulative solar production has driven the cost down by roughly twenty percent, relentlessly, for four decades. This is not a subsidy effect or a temporary glut; it is the ordinary behaviour of a manufactured product, and it has been extraordinarily consistent.
The result is the steepest cost decline of any energy technology in recorded history. Between 2010 and 2024, utility-scale solar electricity got roughly 90% cheaper. Almost every forecast made about solar in the last thirty years, including by the most serious institutions in the world, was wrong in the same direction: they all underestimated it.
What it actually is
A solar farm is an unusually boring object, and that is its great virtue.
There are no moving parts. Nothing burns. Nothing is delivered. A panel sits in a field, in the sun, and produces direct current, and an inverter converts it to alternating current, and it does this for twenty-five to thirty years with almost no maintenance beyond occasional cleaning. The failure mode of a solar farm is that it very slowly gets slightly less good.
This is why the cost fell so far. Solar is not a power plant in the traditional sense; it is a manufactured product, and manufactured products follow learning curves. Every doubling of production makes the next batch cheaper. Coal plants are construction projects, and construction projects do not get cheaper — they get more expensive over time. That divergence is the entire story of the last fifteen years and it was structural, not lucky.
The honest problem is that the sun sets. Solar generates during the day, peaks at noon, and produces nothing at night. In grids with a lot of solar this produces the famous “duck curve”: a mid-day glut of cheap electricity and a sharp evening peak when the sun goes and everyone comes home and turns things on.
Which is precisely why the storage and grid flexibility pages exist, and why the interesting question is no longer “is solar cheap” — it is — but “can we build a system around it.”
The numbers
The collapse. Utility-scale solar LCOE fell approximately 90% between 2010 and 2024, reaching a global weighted average of US$0.043/kWh (IRENA, 2025). Total installed cost dropped to around US$691/kW — an 87% decline from 2010.
Against fossil. Solar PV was on average 41% cheaper than the cheapest fossil fuel alternative in 2024. In China and India, utility-scale solar LCOE is now among the lowest electricity costs ever recorded anywhere.
The build. Global renewable capacity additions reached a record 582 GW in 2024, dominated by solar. Renewables saved an estimated US$467 billion in avoided fossil fuel costs that year.
Firm solar is now competitive too. IRENA’s 2026 analysis found solar-plus-storage delivering round-the-clock electricity at US$54–82/MWh in high-resource regions, down from over US$100/MWh in 2020, with further declines of roughly 30% expected by 2030. The intermittency argument is being priced out of existence.
The learning curve. Roughly a 20% cost reduction per doubling of cumulative production, sustained for four decades. This is why every institutional forecast underestimated solar, consistently, for thirty years.
The remaining costs are not the panels. Balance-of-system — racking, inverters, wiring, labour, permitting, grid connection — now accounts for roughly 60–65% of installed cost. The panel is nearly free. Everything around it is not.
Why it matters
A solar panel is a piece of glass that sits in a field and turns sunlight into electricity for thirty years without anybody touching it, and we have made it so cheap that the panel is now the cheapest part of the installation.
Sit with that. For the whole of human history, energy meant burning something — wood, coal, oil, gas — and burning something means extracting it, transporting it, and putting the exhaust somewhere. Every fire we have ever lit has needed a supply chain and produced a waste stream. It has been the fundamental structure of civilisation.
And now, quietly, we have a way of getting energy that requires neither, and it is the cheapest option available, and it took about fifteen years to happen.
The reason this matters beyond the carbon is what it does for people who have never had electricity. A village in the Rift Valley does not need to wait for a national grid to arrive across a thousand kilometres of nothing. It needs panels, a battery, and an afternoon. Solar is the first energy technology in history that is cheaper at small scale in remote places than the alternative — which turns the entire twentieth-century model of energy development inside out.
The sun has been arriving, free, every morning, for four and a half billion years. It landed on your grandparents and it will land on your grandchildren. We have finally built something that catches it, and the extraordinary thing is not that it works. It is that it is cheap.
What it actually takes
Land, and being careful with it. Utility-scale solar takes space, and where it goes matters. On degraded land, brownfield, rooftops, car parks, reservoirs and alongside crops (agrivoltaics, which can raise yields in hot climates by shading the plants), it is close to free of ecological cost. Bulldozed into intact desert or grassland, it is not, and desert ecosystems in particular are far more fragile and far slower to recover than they look.
Grid connection, which is now the bottleneck. In many markets the queue to connect a project to the grid is years long. Built, financed solar sits waiting for a wire. This is the single largest brake on deployment in the developed world.
The duck curve, which needs a system answer. Mid-day solar gluts and evening peaks require storage, demand shifting and transmission. Solar alone is not a grid.
Supply chain and its ugly corners. Polysilicon production is concentrated, and a significant share has been linked to forced labour in Xinjiang. This is not a reason to abandon solar; it is a reason to demand traceable supply chains, and the industry has been slow and defensive about it.
End of life. Panels last around thirty years and then they are waste. Recycling exists, is improving, and is not yet at the scale that is coming. Better to build the infrastructure now than to discover the problem in 2045.
Where it matters most
The deserts and drylands have the best resource on Earth — the Sonoran, the Mojave, the Sahara, the Gulf, the Atacama. They are also ecosystems, not empty, and desert soils and cryptobiotic crusts take centuries to recover from a bulldozer. Siting on already-disturbed land matters enormously here and is frequently ignored.
The East African Rift and rural South Asia are where solar is genuinely transformative, because it leapfrogs the grid entirely. A village gets electricity in an afternoon rather than waiting thirty years for a transmission line.
The North China Plain and India are where the largest absolute deployment is happening, at the lowest costs ever recorded.
The Mediterranean is where agrivoltaics matters most: panels shading crops that are increasingly struggling in the heat, generating power and reducing water loss simultaneously. In a warming Iberia this stops being a trade-off and becomes a synergy.
How to tell it’s being done well
What was the land doing? Brownfield, degraded ground, rooftops, car parks and reservoirs are close to free. Intact desert, grassland or forest is not, and desert ecosystems recover far more slowly than they look.
Is the supply chain traceable? Polysilicon has documented links to forced labour. A developer who cannot tell you where the silicon came from has not looked.
Is there a plan for the panels in 2055? Thirty-year panels are a waste stream arriving on a schedule we can already see.
Is it paired with anything? Solar without storage or flexibility contributes to the duck curve as much as it solves it. The good projects are thinking about the system.
What you can do
Anyone
- Solar electricity got about 90% cheaper between 2010 and 2024. If your mental model of solar is from 2010, it is wrong by an order of magnitude.
- Ask where a solar farm is being sited before you support or oppose it. Rooftops, brownfield and car parks are close to free. Intact desert is not.
Communities
- Community solar lets people who cannot put panels on their own roof own a share of a local array. It is the most underused ownership model in the field.
- Car parks, reservoirs and rooftops are the least contentious sites in existence and they are massively underused.
Policymakers
- Fix the grid connection queue. Financed, permitted solar is sitting idle waiting for a wire, and that is now the largest brake on deployment in the developed world.
- Prioritise built environment and degraded land over intact ecosystems. Desert is not empty.
- Support agrivoltaics. In hot climates, panels over crops can raise yields and cut water use while generating power.
Business and investors
- The panel is now the cheapest part. Balance-of-system, permitting and grid connection are 60 to 65% of cost and that is where the returns are.
- Demand traceable polysilicon. The forced labour exposure in this supply chain is real and it is a growing liability.
Who is working on this
We are researching which organizations in our directory of 8,493 actively work on this solution, and we only list an organization once we have verified it. That research is ongoing. In the meantime, search the directory yourself:
Questions
How much has solar actually fallen in cost?
Utility-scale solar electricity fell approximately 90% between 2010 and 2024, reaching a global weighted average of US$0.043 per kilowatt-hour. Installed cost dropped 87% to around US$691/kW. It is the steepest cost decline of any energy technology in recorded history, and nearly every institutional forecast for thirty years underestimated it.
Why did it get so cheap?
Because solar is a manufactured product, not a construction project. Every doubling of cumulative production has driven costs down roughly 20%, sustained for four decades. Coal and gas plants are construction projects, and construction projects do not get cheaper with volume; they get more expensive. That divergence is structural, not lucky.
What about when the sun goes down?
That is the real problem and it is a system question. Solar peaks at midday and produces nothing at night, creating the duck curve: a mid-day glut and a sharp evening peak. The answer is storage, demand shifting, transmission and pairing with wind, which blows harder at night and in winter. IRENA now finds solar-plus-storage delivering round-the-clock power at US$54-82/MWh in good regions.
Does solar take up too much land?
It depends entirely where you put it. On rooftops, car parks, brownfield, reservoirs and degraded land, the ecological cost is close to zero. Bulldozed into intact desert or grassland it is not, and desert ecosystems in particular are far more fragile and slower to recover than they appear. Siting is the whole question.
What is agrivoltaics?
Solar panels installed above crops. In hot, dry climates the shade can reduce heat stress and water loss enough to raise yields while the panels generate electricity. In a warming Mediterranean this stops being a trade-off between food and power and becomes a synergy.
Are there problems with solar supply chains?
Yes, and they should be stated plainly. Polysilicon production is highly concentrated and a significant share has been linked to forced labour in Xinjiang. That is not a reason to abandon solar; it is a reason to demand traceable supply chains, and the industry has been slower and more defensive about this than it should have been.
Sources
- Project Drawdown - Deploy Utility-Scale Solar PV (Drawdown Explorer) Framework and classification. Cited, not reproduced.
- IRENA (2025) - Renewable Power Generation Costs in 2024
- IRENA (2026) - 24/7 Renewables: The economics of firm solar and wind
- Our World in Data - Why did renewables become so cheap so fast?
- IPCC (2022), AR6 Working Group III - Energy Systems
The solution taxonomy follows the framework popularised by Project Drawdown. The analysis above is our own; for their carbon modeling and rankings, visit them directly.