Issue 050 - Solar manufacturing - Material-cost scaling
How much would expensive U.S. polysilicon add to the cost of solar power?
A Wacker Chemie polysilicon plant in Tennessee is at risk of closing after new U.S. trade measures affecting semiconductor and solar supply chains. Reuters reports that U.S.-made polysilicon can cost about four times as much as competing foreign material.
The problem
Estimate how much polysilicon is required to manufacture a typical modern solar panel.
Then estimate how much more the silicon in one panel would cost if it were made from U.S. polysilicon costing roughly four times the global benchmark.
Finally, scale that premium to enough panels to build a 1-gigawatt solar installation.
You'll need to estimate the physical dimensions and thickness of the silicon cells, silicon density, how much of a panel's area is actually silicon, and the electrical output of a typical panel.
Because Fermi problems target an order of magnitude, I normally use no more than two significant digits and write most calculations in scientific notation; the Fermi reference explains both conventions.
Before checking sources
Matt's first pass
I assumed a 1 GW solar installation is labeled according to its maximum expected production.
If the maximum sunlight power reaching Earth's surface is about 1,400 W/m2, then after accounting for atmospheric interference and cloud cover, I used about 1,200 W/m2. If solar panels are about 20% efficient, they can generate a maximum of about 240 W/m2.
I guessed the average panel is about half a square meter, so two panels would generate a maximum of about 240 W.
panel power ~= 120 W/panel
panels per kW ~= 1,000 W / 120 W/panel
~= 8 panels
panels per MW ~= 8 x 10^3
panels per GW ~= 8 x 10^6Then I assumed a panel weighs about 25 kg, and polysilicon is maybe 20% of that by mass, or 5 kg.
global polysilicon price ~= $5/kg
U.S. polysilicon price ~= 4 x $5/kg
~= $20/kg
global silicon cost/panel ~= 5 kg x $5/kg
~= $25
U.S. silicon cost/panel ~= 5 kg x $20/kg
~= $100For the full project:
global silicon cost ~= 8 x 10^6 panels x $25/panel
~= $2 x 10^8
U.S. silicon cost ~= 8 x 10^6 panels x $100/panel
~= $8 x 10^8
premium ~= $6 x 10^8That means the more expensive polysilicon would add about $600 million to the material cost of enough panels for a 1 GW solar installation.
Calibration Score
Matt's Calibration Score: 50 / 100
Higher is better: earn points for accurate pegs, sound models, correct math, and a result close to the sourced answer. The image shows percent full of it: 100 minus the Calibration Score.
Pegs: 0/30. Solar efficiency and panel mass were useful, but panel area, panel count, and polysilicon mass per panel were far enough off to dominate the miss.
Model: 30/30. Panels times silicon per panel times price gap is the right model for separating a raw-material premium from the whole project cost.
Math: 10/10. The arithmetic was clean once the assumptions were chosen.
Result: 10/30. The first pass landed a little more than one order of magnitude high: about $600 million versus a sourced range near $30 million to $40 million.
Grounding facts
Bernreuter Research listed the global polysilicon spot price at about $5.45/kg on September 2, 2026.
The U.S. Department of Energy's 2025 solar cost benchmark describes a representative utility-scale PV module as 600 W with 21.9% efficiency. A 1 GW direct-current installation therefore needs roughly 1.7 million 600 W modules.
Recent industry and renewable-cost references put modern polysilicon consumption around 2 grams per watt, far below old solar-industry levels. That means a 600 W module contains on the order of 1 to 1.5 kg of polysilicon input, not 5 kg.
DOE's recent benchmark also puts U.S. utility-scale PV system pricing around $1/W, so a 1 GW project is already roughly a billion-dollar construction problem before this one material premium is considered.
After checking sources
Check and recalibrate
Use a 600 W module and a polysilicon intensity of about 2 grams per watt:
module power ~= 600 W
polysilicon intensity ~= 2 g/W
polysilicon per module
~= 600 W x 2 g/W
~= 1.2 x 10^3 g
~= 1.2 kgAt a global spot price around $5/kg, the polysilicon in one panel is cheap:
global price ~= $5/kg
U.S. price ~= 4 x $5/kg ~= $20/kg
price premium ~= $15/kg
premium per module
~= 1.2 kg x $15/kg
~= $18/moduleIf you use Bernreuter's $5.45/kg current price, the premium is closer to $20 per 600 W module. Scaling to 1 GW:
1 GW ~= 1 x 10^9 W
module power ~= 600 W
modules needed ~= 1 x 10^9 W / 600 W/module
~= 1.7 x 10^6 modules
project premium
~= 1.7 x 10^6 modules x $20/module
~= $3.4 x 10^7The better estimate is therefore about $20 extra per panel and roughly $30 million to $40 million for a 1 GW solar project.
That is real money. But it is not a fourfold increase in the cost of the panel or solar plant. On a utility-scale project that may cost around $1 billion, this premium is more like a few percent of total installed cost, while it can be a more noticeable share of the module-only manufacturing cost.
Post-check reflection
Matt's reflection
I was about right with the sun power at the surface for this calculation, and modern solar-panel efficiency.
I was pretty low on typical panel surface area, by about 5x. I got the average panel mass right, but was way too high on the estimated percent of the panel mass that is polysilicon: 20% instead of closer to 5%. That put me an order of magnitude above the actual cost for that material.
Ultimately, while that individual material cost would be 4x higher, and that does create a meaningful increase in project cost, building a 1 GW solar installation would not become 4x more expensive, which might be one possible misreading of the headline.
Recommended memory peg
For solar-panel material problems, remember: a modern utility panel is roughly 600 W, a 1 GW solar project needs about 1.5 to 2 million panels, and current polysilicon use is only about 2 g/W. A $15/kg polysilicon premium therefore adds only about $0.03/W.
Reader results
Bars show how submitted estimates sort into the answer choices from the gut-check prompt.