Executive summary
- Energy is the central weakness of indoor farming. A peer-reviewed desert-site study found hydroponic lettuce used about 82 times more energy per kg than field-grown lettuce,1 and greenhouses remain more energy-efficient than fully closed plant factories with current technology.2
- The biggest lever is physical, not digital: how efficiently electricity becomes useful light, and how efficiently that light becomes saleable biomass. Current LED technology has calculated efficacy limits of about 3.4–4.1 µmol/J.3
- Climate changes the problem. In deserts, heat from lights must be removed at high outdoor temperatures; in cold countries, the same heat is valuable and can be recovered.
- AI adds value on top of good engineering: forecasting-driven scheduling, model-predictive climate control, digital-twin testing and fault detection. We rate each lever’s maturity honestly.
- Every claim of savings should be measured against a metered baseline. InCrop reports kWh per kg monthly from per-zone sub-metering.
1. Why energy decides viability
Indoor farms replace free sunlight and outdoor air with electricity. That substitution buys independence from climate, near-total water recovery and very high yield per square metre — but it moves the dominant cost from land and water to power.
Lettuce in Yuma, Arizona: hydroponic greenhouse 90,000 ± 11,000 kJ/kg/yr versus conventional field 1,100 ± 75 kJ/kg/yr — about 82 times more energy, alongside 11 times higher yield and much lower water use.1
A comparison across the Netherlands, northern Sweden and the United Arab Emirates reached a consistent conclusion: with current technology, greenhouses are more energy-efficient than closed plant factories, and closed systems make most sense where water or land is scarce.2 Any credible indoor-farming proposal therefore starts with an energy model, not a yield brochure.
2. Where the energy goes
In a closed vertical farm, electricity flows to four main end uses:
- Lighting — the largest load. All photosynthetic light is artificial.
- Cooling and dehumidification — almost all lighting energy ends up as heat in the room, and plant transpiration adds humidity that must be removed.
- Air movement — fans to maintain uniform climate across stacked layers.
- Water and auxiliaries — pumps, dosing, RO, controls, automation.
Because lighting energy becomes heat inside the building, lighting and climate loads are coupled: every watt saved on light is also a watt the cooling system no longer has to remove in a hot climate.
3. A kWh-per-kg framework
We decompose energy per kilogram so that each factor can be owned, measured and improved separately:
The decomposition shows four separate improvement routes for lighting alone: better fixtures (µmol/J), better light delivery to leaves rather than walls and aisles, higher light-use efficiency through crop recipes and climate, and a higher saleable fraction through less waste. AI contributes mainly to the last two, and to E_climate.
4. Lighting: photons first
Research on LED physics estimates efficacy limits with current LED technology of about 3.4 µmol/J for white + red fixtures and 4.1 µmol/J for blue + red fixtures, with optical protection from water and humidity reducing these by roughly 10%.3
InCrop practice:
- Specify fixtures on independently tested photosynthetic photon efficacy, not wattage or marketing labels.
- Design layouts so photons land on canopy, not on aisles and walls.
- Use dimmable, zoned fixtures so light intensity can follow crop stage instead of running at full power for the whole cycle.
- Track lumen (photon) maintenance; degraded fixtures quietly raise kWh per kg.
5. Hot deserts vs. very cold climates
| Factor | Hot desert | Very cold climate |
|---|---|---|
| Fate of lighting heat | A burden: must be removed at high outdoor temperatures | An asset: can be recovered to heat the farm or neighbours |
| Dominant climate load | Cooling and dehumidification | Dehumidification; heating only if heat recovery is poor |
| Natural opportunity | Solar PV output peaks when cooling demand peaks; cooler nights | Cold outdoor air for free cooling; often low-carbon hydro or geothermal grids |
| Key design moves | Opaque insulated envelope, night-shifted lighting, chilled-water thermal storage, high-efficiency chillers | Heat pumps for heat recovery, super-insulated envelope, free cooling, district-heat integration |
| Example context | Gulf and wider MENA region | Iceland: 100% of electricity from hydro and geothermal (2010 figures)4 |
Heat pumps move more heat energy than the electrical energy they consume, which is why heat recovery is so effective in cold climates. In deserts, the equivalent move is to shift heat rejection into cooler night hours and to store “cold” when electricity is cheap or solar-rich.
6. AI levers and their maturity
AI does not change the physics above; it helps a farm operate closer to its physical best, every hour, as conditions change. We rate maturity as: Established (widely used in buildings or industry), Emerging (proven in pilots, adoption growing) and Research (promising, not yet routine).
| Lever | What it does | Maturity |
|---|---|---|
| Load forecasting & tariff-aware scheduling | Moves lighting photoperiods and pre-cooling to cheaper, cleaner or solar-rich hours within crop constraints | Established |
| Fault detection & diagnostics | Spots fouled coils, failing fans, drifting sensors and degraded LEDs from meter and sensor patterns | Established |
| Model-predictive control (MPC) | Optimises temperature, humidity, CO₂ and light set-points over a forecast horizon instead of reacting | Emerging in horticulture |
| Dynamic light recipes | Adjusts daily light integral and spectrum by crop stage using growth feedback (images, weights) | Emerging |
| Digital twin | Simulates the farm’s energy and climate so changes are tested virtually before deployment | Emerging |
| Reinforcement-learning control | Learns control policies directly from operation | Research |
Combining tariff-aware scheduling with MPC and dynamic light recipes will reduce kWh per kg relative to a fixed-schedule baseline on the same farm. We do not quote a percentage until it has been measured in a pilot against that baseline.
Safety rule: every AI controller operates within hard limits set by growers and engineers, and falls back to proven rule-based control if data or models fail.
7. Measurement and verification
Energy savings are only real if they are measured against a fair baseline. InCrop projects use:
- Per-zone sub-metering of lighting, HVAC, pumps and auxiliaries.
- Harvest weighing linked to zones, so energy can be divided by saleable kg.
- Baseline periods under fixed, documented control strategies before AI optimisation is switched on, or parallel control zones where the farm layout allows.
- Normalisation for outdoor weather and crop mix, so savings are not confused with a mild month.
- Monthly reporting of kWh/kg, light-use efficiency (g/mol) and their trends.
8. Roadmap
- Energy model at feasibility — climate data, tariffs, grid carbon intensity, building options, fixture options.
- Pilot with full metering — establish baseline kWh/kg under fixed control.
- Introduce AI levers one at a time — scheduling first, then fault detection, then MPC and light recipes — measuring each.
- Scale and operate — carry proven strategies into the production farm; keep monitoring drift.
9. Risks and limitations
- Energy prices can change faster than a farm’s payback period; models should be stress-tested against price scenarios.
- Gains from AI are bounded by the physical design — poor insulation or inefficient fixtures cannot be optimised away.
- Aggressive energy saving can reduce yield or quality; optimisation targets kWh per saleable kg, not raw kWh.
- Control models depend on reliable sensors; calibration and maintenance are part of the system, not an afterthought.
10. Conclusion
Indoor farming in deserts and cold climates is an energy problem first. The most powerful savings come from physics — efficient fixtures, well-placed photons, a good building envelope and climate-appropriate heat handling. AI then keeps the farm operating near that physical optimum as weather, tariffs and crops change. InCrop’s commitment is to measure kWh per kg from day one and to claim only savings we can show on a meter.
References
- Barbosa G.L. et al. (2015). Comparison of Land, Water, and Energy Requirements of Lettuce Grown Using Hydroponic vs. Conventional Agricultural Methods. Int. J. Environ. Res. Public Health 12(6), 6879–6891. https://www.mdpi.com/1660-4601/12/6/6879
- Graamans L. et al. (2018). Plant factories versus greenhouses: Comparison of resource use efficiency. Agricultural Systems 160, 31–43. https://doi.org/10.1016/j.agsy.2017.11.003; summary: HortiDaily, Dec 2017
- Kusuma P., Pattison P.M., Bugbee B. (2020). From physics to fixtures to food: current and potential LED efficacy. Horticulture Research 7. https://doi.org/10.1038/s41438-020-0283-7
- Askja Energy (June 2012). Hydro and geothermal country. https://askjaenergy.com/2012/06/11/hydro-and-geothermal-country/
This white paper is provided for information. Figures labelled “Established” or “Peer-reviewed” are taken from the cited sources; figures labelled “Industry claim” have not been independently verified by LIDAVI; anything labelled “Design target” or “Hypothesis” is LIDAVI’s own and must be confirmed in a site-specific feasibility study. Nothing here is a performance guarantee.