Solar PV: Cells, Inverters, Rooftops and Fields

A solar panel has no moving parts. Light goes in, current comes out. That simplicity is why solar went from a rounding error to the largest source of new U.S. generating capacity: EIA counted 30.8 GW of utility-scale solar added in 2024, 27.2 GW in 2025, and 43.4 GW planned for 2026, about half of everything developers expected to build that year.

What happens inside a cell

A silicon solar cell is a thin wafer with two layers. One is "doped" with atoms that leave spare electrons; the other with atoms that leave gaps where electrons could go. Where the layers meet, an electric field forms. When a photon of sunlight hits the silicon with enough energy, it knocks an electron loose. The field pushes that electron one way and the gap it left the other. Put metal contacts on both faces, connect a wire, and you have a current.

The physics sets a ceiling. Photons with too little energy pass through or warm the cell. Photons with too much waste the excess as heat. For a single silicon junction the theoretical limit is roughly 30 percent of the sunlight's energy. Commercial panels sold today convert roughly a fifth or a bit more. Heat also hurts: a cell gets less efficient as it warms, so a cold, clear spring day can beat a scorching July afternoon.

From cell to grid

Dozens of cells (commonly 60 or 72, or twice that many half-cut cells) are wired into a panel. Panels are wired into strings. Strings feed an inverter, which turns DC into AC, matches the grid's 60-cycle rhythm and, under modern interconnection rules, can help hold voltage steady. Utility plants use central or string inverters by the hundreds; homes use one string inverter or a small microinverter on each panel.

The DC versus AC trap

Developers often install more panel capacity than inverter capacity. A plant might have 120 MW of panels (DC) feeding 100 MW of inverters (AC), a DC-to-AC ratio of 1.2. On bright middays the inverters cap output and a little energy is "clipped," but in mornings, evenings and cloudy hours the plant runs closer to its AC limit for longer. When you see a solar number, ask which rating it is. A press release that says 120 MW and a utility filing that says 100 MW may describe the same project.

Utility-scale versus rooftop

Small-scale solar is a bigger slice than many people assume. EIA estimated about 74 billion kWh from small-scale PV in 2023, on top of 165 billion kWh from utility-scale solar.

Capacity factor by region

The sun sets the budget. EIA's national utility-scale PV fleet ran at about 23.2 percent in 2023 and 2024 and 24.4 percent in 2025. Idaho's fleet ran at about 25 percent each year from 2022 through 2024. Lazard's 2026 regional analysis used 18 percent for the Midcontinent grid (MISO), 19 percent for the Southwest Power Pool and 27 percent for California.

Desert plant at 27%:   100 x 8,760 x 0.27 = 236,520 MWh per year
Midwest plant at 18%:  100 x 8,760 x 0.18 = 157,680 MWh per year

Difference: 78,840 MWh, the desert plant makes 50% more

Land use

Solar farms take a lot of ground. Exact acreage depends on panel efficiency, spacing between tracker rows, terrain and setbacks, so take the number from the project's own filings rather than a national rule of thumb. Suppose our invented Sunfield project fences 600 acres for 100 MW. That is 6 acres per MW. At 27 percent capacity factor, each acre yields about 394 MWh a year (236,520 / 600).

The tradeoffs are real on both sides. Opponents point to lost cropland, changed views and habitat fragmentation; supporters note that a farm can lease land for steady income, that grazing sheep under panels ("agrivoltaics") is common, and that land can return to cropping after a 30- to 35-year project life if the decommissioning bond is good. In irrigated farm country, a less discussed question is water rights: does the lease retire them, sell them, or hold them?

Net metering: why the arguments never end

When rooftop solar makes more than the house is using, the extra flows onto the grid. The billing rule for that export is a fight in nearly every state.

• Classic net metering: the meter spins backward. Each exported kWh offsets a purchased kWh at the full retail rate, which includes the cost of poles, wires and customer service.
• Net billing: exports are measured in real time and credited at an export rate, usually tied to the utility's avoided cost of energy, which can vary by season and hour.

The utility argument: retail rates bundle in fixed grid costs, so paying full retail for exports lets solar owners skip their share of the grid, shifting costs to neighbors without panels. The solar side: exports have extra value the utility does not count (avoided line losses, reduced peak demand, lower emissions), and changing the rules after people invested is unfair. Both sides commission "value of solar" studies, and their answers depend heavily on what values they include.

In December 2022 the California Public Utilities Commission adopted decision D.22-12-056, a net billing tariff for customers who applied to interconnect on or after April 15, 2023. In Idaho, the Public Utilities Commission issued Order No. 36048 on December 29, 2023, in case IPC-E-23-14, approving Idaho Power's move to real-time net billing with an export credit rate. The commission said on-site generation should offset a customer's own use, avoid cost shifting, and pay a fair value for exports. News coverage at the time counted 846 public comments, about half opposing the change; systems installed before December 20, 2019 were grandfathered.

The federal incentive changed too. The 2025 tax law, Public Law 119-21, ended the 30 percent residential clean energy credit (section 25D) for expenditures after December 31, 2025. The cost lesson of this module covers the commercial credits.