UK farms with serious solar capital to deploy face a fundamental choice: rooftop PV on the farm buildings, ground-mount PV on land, or a combination. The right answer varies significantly by farm type, available land, DNO capacity, and strategic priorities. Here's the detailed 2026 comparison.
Rooftop economics
Rooftop PV pays the highest per-kWh value because it self-consumes at grid retail prices (currently 24–28p/kWh for UK commercial supplies). For dairy parlours, intensive livestock houses, grain stores during drying season, poultry sheds, and pig units — all with significant 24/7 or seasonal daytime baseload — rooftop captures 70–95% of generation at retail value. Simple payback typically 4.5–7 years. The constraint: roof area limits maximum capacity. A 2,000 sqm clear-span livestock shed can carry around 320 kW; a 4,500 sqm grain store can reach 700 kW. Beyond what the roof can hold, more capacity requires ground-mount.
Ground-mount economics
Ground-mount PV deploys on agricultural land — typically marginal pasture or low-yield arable land — at large scale (often 500 kW to 10 MW per project). Two operating models: owner-operated (farm builds and owns the array, captures all economics) or third-party developer lease (developer builds and operates, farm captures lease income). Owner-operated economics: capex roughly £600–£800/kW (lower than rooftop due to scale efficiency); generation 950–1,050 kWh/kW (slightly better than rooftop due to optimal orientation and lack of shading); self-consumption depends on adjacent farm load (often low at 10–25%, so most generation exports under SEG at 8–15p/kWh). Simple payback typically 7–9 years on owner-operated ground-mount.
Developer lease economics: zero capex from the farm. Lease rates in 2026 typically £900–£1,300 per acre per year on 25–40 year terms, often with index-linked annual increases. A 10-acre ground-mount lease delivers £9,000–£13,000 per year of farm income for 25–40 years. Significantly better than arable rental income on the same marginal land (typically £100–£200/acre/year).
Planning context
Rooftop: typically Permitted Development under Class A Part 14 GPDO 2015 — no planning permission required for installations on agricultural buildings (size and height limits apply but rarely binding). Listed buildings, AONBs and Conservation Areas may add planning consultation.
Ground-mount: Permitted Development applies only up to 9m × 9m × 4m height per ground-mount (so capped at small-scale only). Above this, full planning permission required, typically 6–12 month timeline including ecological assessment. Above 1 MW, Environmental Impact Assessment required. AONBs and National Parks add significant planning complexity but rarely block — Defra and NFU are increasingly engaged on agrivoltaic schemes in protected landscapes.
Agrivoltaics — ground-mount with continued farming
Agrivoltaic ground-mount installations continue agricultural use of the land beneath the panels. The established UK model is sheep grazing under raised panels (2.5m+ ground clearance, which most modern ground-mount achieves) — fully compatible with sheep welfare, often improving forage availability due to partial shading. SFI 2025 includes biodiversity actions specifically compatible with agrivoltaic schemes (pollinator margins, mixed-species grassland, hedge management around the array). Total farm income from a typical 10-acre agrivoltaic install: £9,000–£13,000 ground-mount lease + £3,000–£6,000 SFI biodiversity stacking + £500–£1,000 continued sheep enterprise income = £12,500–£20,000/year — compared to perhaps £1,500–£4,000/year from the same land in continued arable rotation.
Crop agrivoltaics (translucent panels)
A newer model: agrivoltaics over shade-tolerant crops using translucent or semi-transparent panels. Trials are progressing for soft fruit, salads, ornamentals, and certain berry crops in the UK. Defra and NFU engaged. The economics are still under-evaluated — most UK soft fruit operations should validate against their existing crop agronomist before committing to a major crop-agrivoltaic install. Trials suggest 15–25% yield reduction on most crops (compensated by lease income), with some crops (lettuce, certain berries) showing yield neutral or positive results under partial shading.
DNO capacity constraints
Ground-mount above ~150 kW typically requires significant export capacity from the DNO. Rural networks across the UK are increasingly capacity-constrained — 2026 connection timelines for ground-mount above 250 kW typically run 12–18 months, with some networks effectively closed to new export connection. Rooftop installs sized for self-consumption (no export) can connect in 6–8 weeks regardless. This makes rooftop the faster route to commissioning on capacity-constrained networks.
The hybrid case
For mid-to-large farms (above 800 acres), the hybrid case wins more often than not in 2026: rooftop PV on buildings with strong daytime baseload (high-value self-consumption), plus ground-mount on marginal land with sheep grazing (income diversification, SFI stacking). Both deployed under a single G99 application. Both on a shared monitoring portal. Total programme might be 400 kW rooftop + 800 kW ground-mount across 8 acres = 1.2 MW combined capacity capturing strong rooftop economics on the buildings and meaningful land-lease income on the marginal pasture.
Decision framework
Start with: how much electrical baseload does the farm have during daylight hours? High baseload (dairy, intensive livestock, year-round poultry) → maximise rooftop first. Low baseload (sheep finishing, beef finishing on grass, smallholdings) → ground-mount lease likely better. Then: how much marginal land is available? Substantial marginal land (10+ acres) → hybrid case. Limited land → rooftop only. Finally: what's the DNO capacity? Capacity-available → either route works. Capacity-constrained → rooftop self-consumption only.
We deliver feasibility studies covering rooftop-only, ground-mount-only, and hybrid scenarios for every working farm we work with. The right answer depends on the specifics, but the hybrid case is winning more often than not in 2026.