Energy, utilities and sustainabilityEnergy, utilities and sustainability

Purpose and scope

Indoor cannabis is one of the most energy-intensive ways humans make anything. Growing one kilogram of flower in a conventional indoor room takes thousands of kilowatt-hours, the classic model puts it around 6,000 kWh, about six units of electricity per gram^[1]. At industry scale that added up to roughly 1% of US national electricity use back in 2012, and the 2025 update puts the industry's climate footprint at 44 million tonnes of CO₂e a year, the emissions of about 10 million cars, with a US energy bill near US$11 billion^[3].

Almost none of that intensity is compulsory. The spread between an efficient room and a wasteful one growing the same flower is enormous, and the spread between indoor, greenhouse and outdoor is two orders of magnitude^[4]. Energy is typically one of the top operating costs of a cultivation business, estimates run from 20% to as much as 50% of production cost^[5], and unlike rent or wages, you can engineer it down without giving anything up. This paper is the map: where the kilowatt-hours go, what the benchmark studies actually measured, and the order to attack the bill in.

Definitions

Energy talk collapses into a handful of units. Get these straight and every spec sheet, tariff and benchmark in this paper reads easily.

One more habit: convert everything to the same basis before comparing. US sources quote kWh per square foot and kWh per pound; multiply by 10.8 to get kWh/m² and by 2.2 to get kWh/kg.

Evidence and limitations

Energy-cost fundamentals

1. Lighting is the biggest single line, climate is the biggest family. In the classic end-use model, lighting takes about a third of total energy, and ventilation, dehumidification and air conditioning together take almost half^[1]. A Pacific Northwest utility survey of licensed producers found lighting closer to three quarters of indoor electric load in that mild climate^[4]. In greenhouse-gas terms, the big 2021 life-cycle study found HVAC the largest contributor in every US location it modelled, with lights second everywhere^[2]. Whichever way you slice it: lights plus climate is 80–90% of the story, everything else is noise.

2. Every lighting watt is paid for twice. Essentially all fixture input power ends up as heat in the room, which your cooling then removes at a cost of roughly one extra watt for every three of heat. Cut lighting power and the cooling bill falls with it, the double dividend (Section 7).

3. You are billed for kW as well as kWh. Commercial tariffs commonly charge for your peak demand (kW) on top of your energy (kWh)^[11]. Three flower rooms flipping on at the same second can set a peak that costs you all month, for zero extra yield.

4. You cannot manage what you do not meter. The facilities that benchmark (kWh per square metre of canopy, grams per kWh) are the ones that improve, the benchmarking bodies estimate most growers can cut at least 30% of energy spend with existing measures^[10].

Measured energy use by system

Mills (2012) built the first transparent end-use model of a standard indoor production module and it is still the reference skeleton: lighting about 33% of energy, ventilation and dehumidification 27%, air conditioning 19%, with CO₂ injection, water handling, space heat and drying making up the rest^[1][6]. Power density in a flowering room runs near 2,150 W/m² (200 W/ft²), the same order as a data centre^[5].

Real facilities scatter around that skeleton. The Northwest Power and Conservation Council surveyed licensed producers in Oregon and Washington and measured indoor operations at about 1,380 kWh/m²·yr (128 kWh/ft²·yr) of canopy — roughly 1,075 kWh/m² (100 kWh/ft²) lighting and 300 kWh/m² (28 kWh/ft²) HVAC and pumping — against 130 kWh/m² (12 kWh/ft²) for greenhouses and about 11 kWh/m² (1 kWh/ft²) for outdoor^[4]. Mild-climate rooms lean harder on lighting; hot, humid or freezing climates push the HVAC share up. The 2021 national life-cycle study modelled a facility in every US county and found climate control the largest greenhouse-gas contributor in all of them, driven partly by ventilation rates of 12–60 air changes per hour in surveyed grows, versus 0.35 for a house and a 15-ACH minimum for a hospital operating theatre^[2].

Energy-intensity benchmarks

How much electricity does a gram of flower take? The studies disagree, usefully. The spread tells you how much of the number is physics and how much is choices.

To make the units concrete: 4,400 kWh/kg is 4.4 kWh per gram, about the electricity a typical fridge uses in three days, per gram. Flip it into the benchmarking platform's preferred metric and the same number is ~0.23 g/kWh; the 2021 model's best case (~1,700 kWh/kg) is ~0.6 g/kWh. Well-run LED rooms report their progress in exactly these units^[10].

Lighting efficacy

A grow light's job is photons. Efficacy, photosynthetic photons out per joule of electricity in, written µmol/J, is the number that decides your lighting bill at any given light level. A 1,000 W double-ended HPS delivers about 1.72 µmol/J^[7]. By 2020, commercial LED fixtures had reached 2.5–2.8 µmol/J for white+red spectra and 3.0 for blue+red, with practical limits near 3.4 and 4.1 respectively^[7]. The DesignLights Consortium's horticultural qualification floor, the minimum to get on the QPL that rebate programmes reference, is 2.5 µmol/J from April 2025, which the DLC notes is more than 45% above the best non-LED option^[8].

The arithmetic is worth doing once by hand. Delivering the same photon flux at 3.0 µmol/J instead of 1.72 cuts lighting power by about 43% (1 − 1.72/3.0). That is what the field data shows too: in a benchmarking analysis of 84 indoor facilities, the ones flowering under LED averaged 34% better facility energy efficiency and 80% better production efficiency (grams per unit energy) than double-ended HPS facilities^[9], and a documented Oregon HPS→LED retrofit improved grams per kWh by 68%^[10].

Two honesty notes. First, efficacy differences within the LED market are now bigger than the brand stories suggest, read the tested µmol/J on the QPL listing, not the marketing page^[8]. Second, an LED swap changes the room's physics: less radiant heat means cooler leaf surfaces at the same air temperature, so target air temperature typically rises a degree or two and humidity control gets harder, not easier, the lighting paper covers the plant side; Section 8 covers the machine side.

Lighting heat and HVAC load

Here is the physics that makes lighting efficiency the anchor measure. Essentially every watt you feed a fixture ends up as heat in the room. The photons themselves are absorbed by leaves and surfaces and become heat too. While lights are on, your cooling must remove that heat, and a vapour-compression system spends roughly one watt of compressor power per three watts of heat moved (the ACEEE industry analysis assumed 1.2 kW per ton, a COP of about 2.9, with cooling required year-round during lights-on)^[5].

Run the numbers over a year and the dividend is most of the payback case. The ACEEE worked example: a flowering fixture stepping from 1,000 W HID to 600 W LED saves about 2,300 kWh a year once avoided cooling is counted (12-hour days), and a veg fixture from 600 W to 300 W saves about 2,600 kWh (18-hour days), simple paybacks of 2–4 years at US$0.12/kWh, before any rebate^[5]. At higher power prices the payback shortens proportionally.

HVAC and dehumidification efficiency

After lighting, the climate plant is where the remaining kilowatt-hours live^[1][2]. You cannot opt out of the work. The room's heat must leave and the transpired water must leave, but the same work can be done at wildly different efficiencies. Four levers, roughly in order of how much they save:

1. Buy COP and stage it. Higher-COP equipment does the same job for fewer watts, but only if it can run in its happy zone. A single oversized unit short-cycles at part load: it never runs long enough to dehumidify properly, cycles its compressor to death, and wastes energy on every restart. Multiple smaller staged units, or variable-capacity compressors, track the room's actual load, which in a grow swings hugely between lights-on and lights-off. Size for the real post-LED heat load, not the HPS-era one.

2. Reuse heat you already paid for. Dehumidification by sub-cooling needs the cold air reheated to supply temperature; doing that with electric resistance is a COP-1 penalty you pay twice. Hot-gas reheat uses the refrigeration circuit's own rejected heat for free; air-to-air heat-recovery dehumidifiers pre-cool incoming air against the cold outgoing stream, UC Davis testing of such a system measured 30–50% savings over conventional units without recovery^[5]. Desiccant systems can likewise be recharged with waste heat where it exists^[5].

3. Use free cooling where the climate offers it. When outside air is cooler and drier than the room, an economiser can dump heat for the cost of a fan instead of a compressor. The surveyed industry runs ventilation anywhere from 12 to 60 air changes per hour^[2], sealed CO₂-enriched rooms rightly sit at the low end, but lights-off hours and shoulder seasons still offer free-cooling windows if the controls and filtration are designed for it. The trade-offs (CO₂ loss, humidity ingress, filtration, biosecurity) are covered in airflow design and CO₂ enrichment.

4. Mind the dehumidifier's own efficiency. Dehumidifiers are rated in litres of water removed per kWh; integrated and heat-recovery designs beat fleets of portable domestic units decisively, and drying rooms deserve the same scrutiny as flower rooms. When the lights-off period arrives, the room still transpires for hours, lights-off dehu capacity is a mould-risk control as much as an energy line (see mould risk).

Demand charges, energy use and photoperiod

Your bill has two different products on it. Energy (kWh) is the total you used. Demand (kW) is the fastest you used it, typically the highest 15- or 30-minute average in the month, and commercial tariffs commonly price it separately. In a US survey, nearly five million commercial customers were on tariffs with demand charges above US$15 per kW-month^[11]; at that rate a 190 kW peak costs ~US$2,850 a month before you have paid for a single kWh.

This is the cheapest retrofit in the whole paper: it is a scheduling change. Flower rooms run 12/12 regardless, so staggering their on-times costs no photons, and it also smooths the load on your own transformer, wiring and HVAC plant, the climate system sees one room's worth of step change at a time instead of three. If your veg room runs 18/6, park its 6 dark hours over the flower rooms' overlap window.

Tariff thinking, kept generic. On a flat tariff, only efficiency and demand management matter. On a time-of-use tariff, a kWh at peak can cost several times an off-peak kWh, and a 12/12 room is free to put its entire lights-on block wherever the cheap hours are, typically overnight. Running lights-on through the night also puts your biggest cooling load into the coolest outdoor air, which raises the real-world COP of every unit rejecting heat outside. The costs are human: night-shifted rooms need night checks, and any light-leak discipline problems get harder to spot. Read your actual tariff sheet (fixed charge, energy rates by period, demand charge, power-factor clause) before optimising anything.

Water-system energy use

Indoor cannabis is thirsty in a specific, recoverable way. Reported irrigation demand runs around 9–11 L/plant/day (2.4–2.9 gal/plant/day) for mature indoor plants in peak season, and about 22.7 L/day (6.0 gal/day) for outdoor plants at the height of summer^[6], though per-plant numbers vary so much with pot size, plant size and stage that the benchmarking bodies deliberately measure water per unit of canopy instead^[10]. In drain-to-waste systems another 10–30% is pushed through deliberately as runoff to control salts (see the irrigation papers for why).

Here is the part beginners miss: in a sealed room, almost every litre you irrigate ends up in the air, because the plant transpires the overwhelming majority of what it drinks. Your dehumidifier and AC coils then condense it back to liquid, a large flower-room dehumidifier can yield around 270 L/day (71 gal/day), roughly 1,900 L/week (502 gal/week), of near-distilled condensate^[13]. That is water you already paid to pump, treat, and then remove from the air at real electrical cost. Sending it down the drain is paying full price for a product and binning it at the door.

Rules of thumb for the loop. Treat condensate before reuse, carbon/sediment filtration plus UV is the common stack, then blend or re-mineralise since near-zero-EC water is aggressive on plumbing and useless as a calcium source^[13]. Collect runoff separately: it carries nutrients and root-zone microbes, so it needs proper treatment or disposal, not quiet blending. Keep collection tanks dark, cool and cycled to stop biofilm. And check your local rules first, jurisdictions differ on whether reclaimed condensate may touch a consumable crop, on backflow prevention, and on trade-waste discharge of runoff. Water quality itself (what is in it before nutrients) is covered in source water, RO and alkalinity.

Carbon intensity and greenhouse emissions

Carbon follows the same arithmetic as the bill: kWh consumed × the grid's emissions per kWh. The 2021 national study put cradle-to-gate emissions of indoor production at 2,283–5,184 kg CO₂e per kg of dried flower depending on location (median 3,658), with US grid intensities spanning roughly 245–766 g CO₂e/kWh^[2]. Mills' earlier central estimate was 4,600 kg CO₂e/kg, enough that a single joint carried about 1.5 kg of CO₂^[1]. The 2025 industry-wide update lands at 44 Mt CO₂e per year, about 1% of total US emissions, with roughly 90% of it attributable to indoor production^[3].

For licensed indoor growers the route is usually fixed by regulation, security and quality requirements, so the honest carbon plan is: cut kWh (everything above), then buy cleaner kWh. Grid mix matters enormously: the same room emits several times less in a hydro-heavy region than a coal-heavy one at identical efficiency^[2]. On a renewables-heavy grid like New Zealand's, 85.5% renewable electricity in 2024^[14], the carbon per kWh story softens dramatically. The bill does not: every efficiency argument in this paper is written in dollars first, and those survive any grid.

Energy retrofit priorities

Efficiency projects fail by sequencing, not by technology. The order below exists because each step changes the sizing maths of the next, do them backwards and you buy equipment twice. It also front-loads the cheap wins so the early savings fund the later capital.

1. 1
   Measure first, two weeks minimum of baseline

   Submeter lighting, HVAC, dehumidification and 'everything else' separately (Section 13), and log room conditions alongside. You are buying two things: the before-photo that proves every later saving, and the load numbers that size every later purchase. Benchmarking your kWh/ft² and g/kWh against the industry dataset tells you how much headroom you have^[10].
2. 2
   LED retrofit, the anchor measure

   Biggest single lever: ~40% off lighting kWh at equal photons plus the cooling dividend (Sections 6–7), with documented 2–4 year paybacks before rebates^[5] and 34–80% efficiency gaps between LED and HPS facilities in the field^[9]. Ramp intensity per the acclimation paper, an efficiency project that bleaches a crop pays for nothing.
3. 3
   Controls and deadbands, near-zero capital

   Stagger lights-on across rooms (Section 9). Widen deadbands, fix fighting setpoints, stage equipment, and relax lights-off targets to what mould risk actually requires rather than flower-time cosmetics. This step routinely finds 5–15% of the bill for the price of commissioning time. And it is where an automation stack earns its keep.
4. 4
   Dehumidification and HVAC upgrade, now sized correctly

   With the LED heat load and staggered schedule known, right-size the climate plant: staged or variable capacity, hot-gas reheat instead of resistance, heat-recovery dehumidification (30–50% savings in testing^[5]). This is the step the LED swap quietly made necessary, the latent share grew (Section 7).
5. 5
   Envelope, seal and insulate last

   Vapour-seal penetrations, insulate against the climate you fight most, fix door discipline and strip-curtain the traffic paths. It is last because it is the slowest payback per dollar in most retrofits, but a leaky envelope quietly taxes every kWh the other four steps saved, and in humid climates infiltration is a permanent latent load on the dehu you just bought.

Energy metering

The utility meter tells you one number a month about a building full of separate machines. Everything in this paper becomes manageable the day you split that number: current-transformer (CT) submeters on the lighting circuits, the HVAC plant, the dehumidifiers and the pump/controls remainder cost little per circuit and connect to any modern monitoring or home-automation platform. From there, three habits do the work:

This is also where the honest sustainability numbers come from (Section 11): a metered kWh/g trend, cycle over cycle, is the entire audit trail. Facilities that submitted benchmarking data found on average that at least 30% savings were available with existing measures^[10], but every one of those savings was discovered by a meter, not a hunch.

Common energy-project failures

These are the recurring wrecks. Every one of them is avoidable at design time and expensive afterwards.

Troubleshooting

Energy faults announce themselves on the meter before the crop. Work this table with your interval data open.

Energy-management principles

Strip the whole subject to one sentence: an indoor grow converts electricity into photons, and then spends more electricity cleaning up the heat and humidity the first spend created. Every kilowatt-hour on your bill is one or the other. That is why efficacy compounds (Section 6), why the double dividend exists (Section 7), and why the retrofit order runs lighting before HVAC (Section 12).

Set your expectations honestly. You will not reach a greenhouse's numbers in a windowless room. Sunlight is a subsidy the building cannot match^[4][12]. But the gap between an unmanaged room and a measured, LED-lit, staged, heat-recovering one is the difference between energy as your scariest cost line and energy as a controlled, boring input, the studies put 30%+ of the bill within reach of existing, proven measures^[10][5]. Start with the meter.