How I assessed the global potential of solar energy

6 min readApr 1, 2023


This article is part of a series on the DEC Report. The DEC report is a 200+ pages freely accessible report I wrote on climate change and energy. It assesses the world’s potential to tackle climate change by removing GHG emissions in energy and agriculture, and it assesses how we can optimize this transition.

Global solar power potential

Global solar potential mainly refers to how much electricity we can produce from solar panels. Solar energy can be used to produce heat or to produce electricity, in this article and in the DEC report I mainly investigated their use for electricity, that is photovoltaic (PV) panels. Solar panels are a low-carbon source of energy (, A.III.2: median of ~45 gCO2eq/kWh vs > 400 for gas and > 800 for coal).

Solar panels are currently the TOP 4 source of low carbon electricity in the world. It comes after hydroelectricity, nuclear energy and wind power, which themselves produce less electricity than coal and natural gas.

Solar panels still produce a lot less electricity than fossil fuels

As you can see, solar electricity production has continuously increased during the last years and is currently on a kind of exponential curve. However, it is to be considered that solar panels we currently install have a low lifespan of approximately 30 years. When we have a +200 TWh increase one year, we may have a -200 TWh decrease in 30 years, making this exponential probably short term as current increases aren’t affected by a large decrease (we weren’t installing as much solar panels 30 years ago). In order to tackle climate change and reduce energy issues, it is extremely interesting to estimate the full potential of this energy in the world.

In order to do so, I used the Global Solar Atlas.

The Global Solar Atlas is a freely accessible dataset to estimate the potential of solar panels over the world.

What it does:

  • Estimate the output of a solar panel for a specific location on the planet

What it doesn’t do:

  • Consider where solar panels could realistically be installed (based on population density, etc.)
  • Consider that solar panels will produce less electricity with continous degradation
  • Consider where solar panels are currently installed, where future solar panels will be installed, and how we can replace current solar panels when they reach their end of life where they’re installed with newer solar panels

In order to adjust the results from GSA, I evaluate the true capacity factors solar panels had in some places thanks to WRI’s power plants dataset and data from EIA. I also used data on topography and distribution of population on the world.

I evaluate a global potential of 2000 PWh/y for solar panels just with constraints on land (no forests, no crops, above density limit). Just land surface is therefore not a constraint for solar panels.

In the report I optimized the location of solar panels based on where needs are located (i.e. where current energy production is located), but I only did this per country. A global optimization should ignore borders. If solar panels produce more electricity in northern Africa than in northern Europe, they should rather be installed in northern Africa. A detailed example of this strategy can be found in the report:

This is where the model installed solar panels for the configuration used in the report:

As you can see there may be a large potential in some places, but because not enough people actually live in these places, it’s harder to forecast a large installation of solar panels there.

The main forecasted constraint on solar panels seems to be their lifespan. In the evaluated configuration, we can add even more solar panels. We’re currently installing a bit less than 300 GW of solar panels per year. In the model we can reach up to 400 GW per year:

However, as you can see, at some point in the future, solar panels we’re currently installing will reach the end of their life (negative -400 GW). It’ll be a very large challenge to produce enough solar panels to keep replacing these old solar panels. While, in theory, we would have much less fossil fuel available. The current production of solar panels largely profits from the highly CO2-intensive world we’re living in.

The model also doesn’t consider grid constraints. It could be extremely hard to make an efficient grid with a lot of solar panels at it would require large quantities of copper and a large workforce to install enough cables, or it would also require a lot of lithium to build enough batteries. Pumped-storage hydroelectricity is promising but has limits and can’t be done everywhere, while hydrogen also has many limits and isn’t efficient.

Overall, in this configuration of the DEC model, I don’t expect countries to install enough solar panels to produce much more than 15 PWh of electricity, and I expect that it’ll be a peak rather than a continous increase as a lot of solar panels will reach their end of life and be hard to replace. 15 PWh alone is insufficient to replace the > 100 PWh of fossil energy we’re currently consuming. But it could be enough to replace a significant portion of fossils we use for electricity (~15 PWh of coal+gas). Yet, where we consume a lot of coal and gas for electricity, there isn’t necessarily enough land to install solar panels while using this land for carbon storage (aka forests) or food production. The electricity from solar panels could instead be used to replace parts of the oil we use in cars, by using electric cars.

Solar electricity in the global energy mix


It seems plausible that countries like United Arab Emirates, Qatar, Oman, Saudi Arabia and Iran could produce enough electricity from solar panels to be autonomous IF they are able to store this energy. Otherwise, other countries should rather install solar panels if they fit in a global strategy to entirely remove fossil fuels, and they should avoid a strategy where fossil fuels would be kept forever (i.e. in a long-term gas/solar or coal/solar strategy).


You can see how some of these results were used in much more details in the DEC report. Some of the data and models I used are also available on Github.

👏 if you liked this article and I’ll do more!