Solar Land Supply Constraints for Data Centers

Can terrestrial land constraints for solar-powered data centers become a binding factor that increases energy costs beyond our conservative scenario ($0.088–$0.105/kWh)?

Answer

No. Land is not a binding constraint on terrestrial solar energy costs for data centers at any plausible scale through 2040. Three independent lines of evidence converge on this conclusion:

  1. Aggregate land supply vastly exceeds demand. Even the most extreme scenario — 100% of projected 134 GW US data center demand met entirely by dedicated solar+storage — would require ~6.7 million acres (0.30% of US land). The BLM has designated 31 million acres of federal land for solar development. NREL identifies 5,750 GW of solar potential on 44 million acres of federal land alone. The DOE Solar Futures Study explicitly concludes "land availability will not limit solar deployment."

  2. Land cost is a negligible fraction of solar LCOE. Land lease costs contribute 2–5% of utility-scale solar LCOE (~$2–4/MWh out of $38–58/MWh). Land costs would need to increase roughly 10x before materially affecting solar economics — an increase not observed or projected even in premium markets.

  3. Local constraints are real but don't affect energy cost. Competition for land near data center hubs (Northern Virginia at $2–3.75M/acre, Columbus at $150K+/acre) is intense, but this affects data center siting decisions and development cost, not solar generation LCOE. The "power-first" siting trend — where operators locate data centers near cheap power rather than near population centers — provides a structural release valve.

Our conservative scenario values of $0.088–$0.105/kWh already incorporate sufficient margin to absorb any plausible land-related cost increases. No revision is warranted.

Analysis

Physical Land Requirements

Solar land requirements per MW of nameplate capacity are well-characterized empirically:

Configuration Acres/MW_DC Source
Fixed-tilt (median) 2.8 LBNL 2022, >90% of US utility PV lbnl-solar-land-2022.1
Single-axis tracking (median) 4.2 LBNL 2022 lbnl-solar-land-2022.1
Industry rule of thumb 5–10 Including access roads, buffers nrel-land-use-2013.1
Western US empirical (719 projects) ~4.9 Jordaan et al. 2025 jordaan-solar-land-metrics-2025.1

These are nameplate figures. For continuous data center load, solar requires roughly 4x overbuild to account for a ~25% US average capacity factor eia-capacity-factors.1, plus storage. A 1 GW continuous data center would need ~4 GW_DC of solar nameplate — approximately 11,200 acres (fixed-tilt) to 16,800 acres (tracking). Battery storage adds less than 1% to the footprint.

Land requirements are declining: LBNL found median power density improved 52% (fixed-tilt) and 43% (tracking) from 2011–2019 pv-mag-solar-per-acre-2022.1, driven by higher-efficiency modules. Commercial panel efficiency is now 22–25%, with 27–30% expected by 2030 via tandem perovskite-silicon cells nrel-atb-2024-solar.1. Each percentage point of efficiency reduces land per MW by ~4%. By 2035, acres/MW could decline another 20–35% from current levels.

National-Scale Land Demand vs. Supply

Data center solar demand at realistic adoption rates is a tiny fraction of available land:

Scenario DC demand (2030) Solar share Solar nameplate needed Land needed % of US land
Conservative BTM 80 GW 20% ~82 GW_DC 410,000 acres 0.018%
Aggressive BTM 80 GW 40% ~164 GW_DC 820,000 acres 0.036%
Theoretical maximum 80 GW 100% ~411 GW_DC 2,055,000 acres 0.091%
Extreme (134 GW, 2035, 100%) 134 GW 100% ~687 GW_DC 6,700,000 acres 0.30%

For context:

The DOE Solar Futures Study modeled up to 1,570 GW_DC of solar by 2050 requiring ~10.3 million acres (0.5% of contiguous US) and explicitly concluded "land availability will not limit solar deployment" doe-solar-futures-2021.1. Data center solar demand, even at the theoretical maximum, is a fraction of that total.

Land Cost Impact on Solar LCOE

Land is a small fraction of solar LCOE:

Sensitivity to land cost escalation:

Land cost multiplier Additional LCOE impact Solar+storage LCOE
1x (current) ~$57/MWh
2x +$2.2/MWh ~$59/MWh
5x +$9/MWh ~$66/MWh
10x +$18/MWh ~$75/MWh

Even a 5x increase in land costs keeps solar+storage well below our conservative scenario of $0.088–$0.105/kWh ($88–105/MWh). A 10x increase — well beyond anything documented or projected — would raise solar+storage to ~$75/MWh ($0.075/kWh), still below our conservative estimate. Land cost is simply too small a fraction of LCOE to be a binding constraint.

Local Constraints: Real but Misframed

Local land constraints are genuine and intensifying — but they affect data center siting, not solar LCOE:

Data center land price escalation:

Data centers outcompete solar for the same land. Amazon paid $700M for a single Prince William County site credaily-dc-land-market-2026.1. Solar economics support only $500–$2,000/acre/year in leases. In Virginia, data center companies have "shoved aside solar developers" for farmland lancaster-farming-dc-solar-va-2025.1.

Local opposition is intensifying: 459 counties across 44 states have adopted severe renewable energy restrictions pv-mag-opposition-zoning-2025.1. Virginia counties rejected more MW of solar than they approved in 2024 for the first time virginia-mercury-solar-rejected-2024.1. $64 billion worth of DC projects have been blocked ($18B) or delayed ($46B) datacenterwatch-64b-blocked-2025.1.

However, these constraints push operators to relocate, not to accept higher solar LCOE. The "power-first" siting trend moves data centers toward cheaper land and better solar resources: 64% of new data center capacity is going to frontier markets (West Texas, Tennessee, Wisconsin) credaily-dc-land-market-2026.1. Co-locating data centers near cheap desert solar (hundreds of thousands of acres in the US Southwest at minimal cost) sidesteps local land scarcity entirely exowatt-dispatchable-solar.1.

Transmission costs for remote solar are modest: LBNL research shows transmission capital costs add $1–10/MWh depending on distance — meaningful but not transformative.

The Actual Bottleneck

Multiple independent sources confirm that grid interconnection, not land, is the binding constraint on both data center and solar expansion. 222 GW of announced DC capacity faces only 147 GW of deliverable power — a 75 GW gap driven by interconnection queues, not land scarcity enr-grid-not-land-bottleneck.1. BTM solar partially avoids this (18–24 months deployment vs. 3–15 years for grid), but still requires grid connections for export or backup.

Mitigating Factors

Several trends work against any future land constraint:

  1. Improving panel efficiency nrel-atb-2024-solar.1: 20–35% reduction in land per MW by 2035.
  2. Agrivoltaics: 14 GW installed globally by end 2024. Higher LCOE (4–148% premium) pv-magazine-agrivoltaics-lcoe-2026.1 — a land-conservation tool, not a cost-reduction tool.
  3. Contaminated/disturbed land: 20 million suitable acres — more than enough for all projected solar doe-solar-futures-2021.1.
  4. Desert colocation: Hundreds of thousands of acres in the Southwest with exceptional solar resources and minimal competing uses exowatt-dispatchable-solar.1.

Sources

lbnl-solar-land-2022

nrel-land-use-2013

jordaan-solar-land-metrics-2025

doe-solar-futures-2021

nrel-federal-lands-solar-2025

blm-western-solar-plan

breakthrough-solar-land-2024

smartenergyusa-solar-land-lease-2026

bisnow-dc-land-prices-2024

datacenters-com-land-prices-2025

credaily-dc-land-market-2026

lancaster-farming-dc-solar-va-2025

pv-mag-opposition-zoning-2025

virginia-mercury-solar-rejected-2024

datacenterwatch-64b-blocked-2025

enr-grid-not-land-bottleneck

exowatt-dispatchable-solar

pv-magazine-agrivoltaics-lcoe-2026

seia-wood-mackenzie-2026

nrel-atb-2024-solar

pv-mag-solar-per-acre-2022

eia-capacity-factors

Evidence

1. LBNL empirical study (2022) of >90% of US utility-scale PV plants: median power density 2.8 acres/MW_DC (fixed-tilt), 4.2 acres/MW_DC (tracking). Power density improved 43–52% from 2011–2019. — lbnl-solar-land-2022

2. NREL baseline study (2013): total capacity-weighted land use 7.3 acres/MW_AC (direct), 8.9 acres/MW_AC (total). Higher than LBNL 2022 due to older, less efficient panels. — nrel-land-use-2013

3. Jordaan et al. (2025, Nature Communications Earth & Environment): empirical analysis of 719 solar projects in Western US found average capacity-based land-use efficiency of 24.7 W/m² (~4.9 acres/MW). — jordaan-solar-land-metrics-2025

4. DOE Solar Futures Study (2021): highest scenario (1,570 GW_DC by 2050) requires ~10.3 million acres (0.5% of contiguous US). Explicitly concludes "land availability will not limit solar deployment." Less than 10% of suitable disturbed land could meet all projected needs. — doe-solar-futures-2021

5. NREL (2025): 5,750 GW of solar PV potential on 44 million acres of federal land. Central scenarios deploy only 51–84 GW on 325,000–2 million acres by 2035 (<0.5% of federal land). — nrel-federal-lands-solar-2025

6. BLM Western Solar Plan: 31 million acres designated for solar across 11 western states (up from 19 million previously). Requirement: within 15 miles of existing or planned high-voltage transmission. — blm-western-solar-plan

7. Breakthrough Institute: solar occupied 336,090 acres in 2020, less than 0.04% of 897 million US farmland acres. Even if all 10.3M acres of 2050 solar came from farmland, less than 1.2% affected. — breakthrough-solar-land-2024

8. PV Magazine (2022): documented 52% improvement in solar power density (fixed-tilt) from 2011–2019, driven by higher-efficiency modules and denser configurations. Tracking systems showed 43% improvement. — pv-mag-solar-per-acre-2022

9. NREL ATB 2024: projects panel efficiency reaching 28% by 2050 via tandem cells. Capacity factor improvements of 7–15% by 2035 from bifaciality and reduced losses. — nrel-atb-2024-solar

10. EIA: US average solar capacity factor ~25%. Arizona: 29.1%. Southwest: 26–28%. Northeast: 14–16%. High-insolation siting reduces land needs by 40–50% vs poor locations. — eia-capacity-factors

11. SmartEnergyUSA: national solar land lease rates $250–$1,000/acre/year. Northeast $600–$1,000+. Midwest $300–$600. Southwest $500–$1,000+. Plains $250–$500. Typical escalation: 1.5–2.5%/year. — smartenergyusa-solar-land-lease-2026

12. EIA/Sargent & Lundy (AEO2025): Owner's Services (land, permitting, legal, engineering) represent $15.3M of $326.3M total capital for a 150 MW solar+storage plant — 4.7% of total capital cost. — eia-capital-cost-aeo2025

13. Bisnow (2024): Microsoft paid $3.75M/acre in Prince William County, VA. Loudoun County values up 60% YoY. Silicon Valley: $5–6M/acre. Land historically <10% of DC development cost but rising. — bisnow-dc-land-prices-2024

14. DataCenters.com: average US DC land $244,000/acre (2024, 50+ acre parcels), up 23% YoY. Ohio farmland repriced from $30K to $150K+ when rezoned for data centers. — datacenters-com-land-prices-2025

15. CRE Daily: Amazon paid $700M for Prince William County site. 64% of new DC capacity in frontier markets (West Texas, Tennessee, Wisconsin). Northern Virginia faces 75,000+ home shortage partly due to DC land competition. — credaily-dc-land-market-2026

16. Lancaster Farming: data center developers have displaced solar developers in Virginia farmland competition. Solar developers lease at up to 10x agricultural rents but cannot match data center outright purchase prices. — lancaster-farming-dc-solar-va-2025

17. PV Magazine: 459 counties/municipalities in 44 states have severe renewable energy restrictions (16% increase). 262 solar projects contested in 2024; 31 canceled including 600 MW Blackwater Solar in Virginia. — pv-mag-opposition-zoning-2025

18. Virginia Mercury: Virginia counties rejected more MW of solar than approved in 2024 for the first time, while data centers were approved in the same communities. — virginia-mercury-solar-rejected-2024

19. DataCenterWatch: $18B data center projects blocked, $46B delayed (2024–2025). 142 activist groups in 24 states. Bipartisan opposition. Virginia: ~$900M blocked, $45.8B delayed. — datacenterwatch-64b-blocked-2025

20. ENR: 222 GW announced DC capacity vs. 147 GW deliverable = 75 GW gap. Grid access, not land, is the primary bottleneck for data center construction. — enr-grid-not-land-bottleneck

21. Exowatt: hundreds of thousands of acres of low-cost desert land in the US Southwest with exceptional solar resources could support 1,200+ GW of data center capacity. — exowatt-dispatchable-solar

22. PV Magazine (2026): German study finds agrivoltaic LCOE is 4–148% higher than conventional ground-mount. Agricultural value creation too small to offset system cost premium. — pv-magazine-agrivoltaics-lcoe-2026

23. SEIA/Wood Mackenzie: US solar installations reached 43 GW in 2025 (279 GW cumulative). Projected 769 GW by 2036. Solar was 54% of all new US electricity-generating capacity in 2025. — seia-wood-mackenzie-2026