What We Publish

The effect of land costs on the economic and sustainability performance of solar photovoltaics in China

Shi Chen, Xi Lu, Jiming Hao, Edgar Virgüez, Ken Caldeira, and Steven J. Davis · Proceedings of the National Academy of Sciences 123, e2512930123 · 2026

Key finding. Because tracking systems need substantially more land per unit of capacity, land costs raise their levelized cost of electricity by 20% against 8% for fixed-tilt, making fixed-tilt cheaper in real Chinese conditions despite generating less — so China installs 18 to 26% more panels for the same electricity output.

Six maps of China in two rows, shaded by the difference in solar electricity cost between tracking and fixed-tilt systems, purple where single-axis tracking is cheaper and green where fixed-tilt is cheaper. The top row holds land cost at status quo while technology cost falls from left to right; the bottom row repeats it with reducing land cost. Panel A also overlays the locations of real fixed-tilt and tracking installations as green and pink dots. Green spreads across eastern and northern China in the top row, while the bottom row is dominated by purple across the north and west.
Which technology is cheaper is a map, not a number. Holding land costs at today's levels (top row) leaves fixed-tilt ahead across much of eastern and northern China; letting land costs fall (bottom row) turns most of the country over to tracking. The dots in panel A show that real installations follow the pattern — tracking where land is cheap, fixed-tilt where it is not. Figure 3 from Chen et al. (2026), Proceedings of the National Academy of Sciences 123, e2512930123. Reproduced under CC BY-NC-ND 4.0. Used unmodified, as the licence requires.

What question did this research address?

Single-axis tracking, which turns panels to follow the sun, raises the electricity a solar installation generates by more than 20%. In United States utility-scale solar it is used on over 90% of capacity. In China, the world's largest solar market and largest emitter, it is used on 12%.

That divergence is usually read as a technology lag. This paper asked whether it is instead a rational response to different conditions — specifically, whether the price of land changes which technology is actually cheaper — and what the consequences are for how much silicon and how much land China's solar build-out consumes.

What did we find?

A spatially explicit model comparing the two technologies across China evaluates generation, land use, cost, sustainability, and resilience to policy change together, rather than comparing them on generation alone.

Tracking does generate more. Capacity factors reach 21.4% against 16.6 and 17.4% for fixed-tilt, roughly 20% more electricity, concentrated in the morning and afternoon hours that fixed-tilt captures poorly.

It also occupies much more ground — 24.8% more land than fixed-tilt in the Northwest, Tibet, North China, and the Northeast, and 18.8% more elsewhere. Where land is expensive, that footprint is what decides the comparison.

The result is a reversal. Land costs add 20% to the levelized cost of tracking electricity but only 8% to fixed-tilt, so the technically less efficient option is the cheaper one under real Chinese land prices.

Choosing fixed-tilt trades land for materials. Matching tracking's output takes 18 to 26% more panels, raising the silicon, glass, and aluminium demand of the same amount of electricity.

Scaled to a 6 petawatt-hour solar target for 2060, current land policies drive 59% of solar electricity to fixed-tilt.

Policy can move it. Reducing the soft costs of land would raise tracking's share to 63% and cut required installed capacity by up to 8%, or 219 gigawatts, for the same electricity — while expanding land use by 35%, or 12,900 square kilometres.

Why does it matter?

It reframes a gap that looks like backwardness as a difference in constraints. China is not failing to adopt tracking; tracking is not the cheaper option where land is priced the way it is in China, and a recommendation to adopt it imported from United States conditions would be wrong.

The trade-off it exposes has no free side. Land-efficient solar costs more silicon; material- efficient solar costs more land. Which to prefer depends on which constraint binds locally, so there is no globally correct technology choice — only a context-specific one.

It also identifies a lever that is not technological. Soft land costs are an administrative quantity, and changing them shifts hundreds of gigawatts of capacity and thousands of square kilometres of land use without any change in the underlying hardware.

Citation

Shi Chen, Xi Lu, Jiming Hao, Edgar Virgüez, Ken Caldeira, and Steven J. Davis (2026). The effect of land costs on the economic and sustainability performance of solar photovoltaics in China. Proceedings of the National Academy of Sciences 123, e2512930123.

Related