On-Site Solar + Battery Storage for Grow Facilities

Why the roof does not cover the load

Start with the two numbers that decide everything. The Department of Energy puts the country's rooftop technical potential at "over 1 terawatt" across more than 8 billion square meters of roof [doe-solar-rooftop-potential]. Divide those and you get roughly 125 watts per square meter, or about 11.6 watts per square foot of suitable roof. A 40,000 square foot building with 60 percent of its roof usable, so 24,000 square feet, can hold around 280 kW of panels. That is the ceiling, before shading, HVAC units, skylights, and setbacks.

Now the load side. An indoor cultivation facility is one of the densest electrical loads in commercial real estate; lighting alone was 66 percent of use in the Northwest Power and Conservation Council's survey of licensed producers, with cooling and dehumidification most of the remainder [nwpcc-cannabis]. Benchmarks on the kWh per square foot page put many indoor facilities well above 100 kWh per square foot of building per year.

A worked comparison, with assumed inputs:

Item	Value	Basis
Building	40,000 sq ft, 20,000 sq ft flowering canopy	Assumed
Annual usage	2,400,000 kWh (60 kWh per sq ft of building)	Assumed; conservative for indoor
Usable roof	24,000 sq ft	Assumed 60 percent
Array size	280 kW DC	11.6 W per sq ft derived from DOE's national figures [doe-solar-rooftop-potential]
Capacity factor	17 percent	Assumed for a fixed-tilt rooftop in a northern state. EIA's utility-scale average, which includes tracking systems in sunny states, was 24.4 percent in 2025 [eia-epm-6-07-b]
Annual production	280 × 8,760 × 0.17 = 417,000 kWh	Arithmetic
Share of load	417,000 / 2,400,000 = 17 percent	Arithmetic

Seventeen percent, from a full roof, on a facility we deliberately made less energy-dense than most. A denser facility at 120 kWh per square foot gets 9 percent. And the solar arrives between roughly 9 a.m. and 4 p.m. while a staggered facility runs half its rooms lit around the clock, so the array offsets daytime kWh only. Production also swings by season; EIA's monthly capacity factors ran from 15.7 percent in January to 31.9 percent in June 2026 [eia-epm-6-07-b], so the winter contribution is roughly half the summer one.

None of that makes solar a bad idea. It makes it a partial idea. Ground mount on adjacent land, carport canopies, and greenhouse operations change the arithmetic. For an urban warehouse grow, the roof is a supplement.

What batteries can do for a grow, and what they cannot

Storage attacks a different line on the bill. Where solar reduces kWh, a battery is aimed at kW, the demand charge. NREL's survey of more than 10,000 tariffs found nearly 5 million commercial customers face demand charges above 15 dollars per kW, the level at which behind-the-meter storage begins to make economic sense, and demand can be 30 to 70 percent of a commercial bill [nrel-demand-charge-survey].

The shape of the peak decides whether a battery works. A grow's profile has two features:

• A spike at lights-on. Several hundred kW of fixtures close on a contactor, and HVAC and dehumidifiers start within the same interval. This spike can last 15 to 60 minutes before the room settles. A battery can absorb it.
• A plateau. After the spike, the lit rooms draw a steady load for 12 hours. A battery cannot carry that. To shave 200 kW for 12 hours you would need 2,400 kWh of storage, which is a utility-scale asset, not a behind-the-meter one.

So the realistic storage strategy is to shave the edge, not the plateau. With assumed numbers: interval data shows a 550 kW monthly peak that occurs for 30 minutes at lights-on, then settles to 450 kW. A battery rated at 100 kW with 100 kWh of usable energy discharges through that half hour, and the meter sees 450 kW. At an assumed 15 dollars per kW-month delivery demand charge, that is 100 × 15 = 1,500 dollars a month, 18,000 dollars a year. If the tariff also has a ratchet, the battery has to perform every month for the saving to hold.

Two cheaper alternatives usually come first. Staggering photoperiods across rooms removes most of the spike at no cost, and ramping LED fixtures over 15 to 30 minutes spreads the edge across intervals. A battery makes the most sense after those are done, when the remaining spike is from equipment that cannot be staged, or where the tariff has a coincident-peak capacity charge that a few hours of discharge can dodge. The peak demand page explains that second case.

Batteries also pair with solar in an obvious way: charge from the array at midday, discharge into the evening lights-on for the second photoperiod group. That turns some of the solar into avoided kW as well as avoided kWh, which is the combination that makes the two technologies better together than either alone.

The tax credit, and why 280E changes the answer

For most businesses, the federal Clean Electricity Investment Credit under Section 48E is the reason solar and storage pencil. The base credit is 6 percent of qualified investment, rising to 30 percent when prevailing wage and apprenticeship requirements are met, with additional 10 point bonuses for domestic content and for energy communities, and it covers both qualified facilities and energy storage technology placed in service after December 31, 2024 [irs-48e].

Two things complicate that for a cannabis company.

Section 280E. The IRS states that "Internal Revenue Code section 280E disallows all deductions or credits for any amount paid or incurred in carrying on any trade businesses that consist of illegally trafficking in a Schedule I or II controlled substance," and that it applies to marijuana businesses because trafficking remains illegal under the Controlled Substances Act regardless of state law; only properly calculated cost of goods sold survives [irs-marijuana-faq]. Note the word credits. A plant-touching entity that owns a solar array cannot, on the plain reading of that guidance, claim 48E. Operators sometimes address this by having a separate entity, such as the property owner, own the system and sell power or lease it to the operating company. Whether that works depends on facts, structure, and current IRS positions, which is exactly why the disclaimer at the top of this page exists. The 280E and energy cost page covers why energy is one of the few costs a cultivator can push into COGS.

The 2025 deadline. The One Big Beautiful Bill Act, signed July 4, 2025, terminates the 48E credit for solar and wind facilities "placed in service after December 31, 2027," and that termination applies to facilities whose construction begins after July 4, 2026 [irs-notice-2025-42]. Notice 2025-42 also removed the five percent safe harbor as a way to establish beginning of construction, except for solar facilities of 1.5 MW AC or less [irs-notice-2025-42]. Energy storage technology is expressly carved out of the wind and solar termination [irs-notice-2025-42]. As of this page's date, the July 4, 2026 begin-construction date has passed. A new solar project that wants the credit must be placed in service by the end of 2027; a storage-only project does not face that deadline.

The upshot for a payback model: build one version with no credit at all, because that is the case for a plant-touching owner. If your advisor finds a structure that qualifies, treat the credit as upside.

Interconnection: the step developers gloss over

Any system that can push power onto the utility's wires goes through an interconnection review. The utility checks whether your transformer, service, and feeder can accept the export, and whether protective equipment is needed. The bigger the system and the more it exports, the more study it triggers, and the study can end with a required upgrade billed to you. The IRS itself lists "interconnection-related delays," including distribution upgrades, among the recognized reasons a project slips [irs-notice-2025-42], which tells you how routine those delays are.

Grid operators have their own thresholds too. In ERCOT, distribution-connected generators are treated as settlement-only distribution generators up to 10 MW, must register with ERCOT if they exceed 1 MW and inject to the grid, and may register optionally below 1 MW [ercot-dr-overview-2023]. State commissions set the review tiers for smaller systems in the other 16 jurisdictions, generally with a fast track for small non-exporting systems and a full study for larger ones.

Practical points for a grow:

1. Design for non-export or limited export if you can. A facility that consumes everything it produces has an easier review, and a grow almost always can.
2. Ask for the pre-application report. Most utilities will tell you the feeder's hosting capacity and your transformer's rating for a small fee before you spend on engineering.
3. Check the service size. A 280 kW array behind a 2,000 A service is one conversation; behind a service that is already at capacity it is a different one.
4. Batteries interconnect too. A storage system that can discharge to the grid goes through the same review; a system configured to never export is simpler.
5. Timeline. Budget months, not weeks, and put the interconnection approval date, not the installation date, on your project schedule.

How to decide

Run three separate calculations, then add them.

• Avoided energy. Solar kWh times your blended supply-plus-delivery energy rate. Use the array size the roof supports, not the size the developer proposes.
• Avoided demand. The kW a battery can reliably shave every month, times your demand rate, after scheduling changes have already removed the cheap part of the spike.
• Resilience. What a lights-on outage during flower costs you, and whether a battery, a generator, or both is the cheaper insurance.

Then apply the tax reality for your entity, the interconnection cost the utility quotes, and any state incentive your PUC or state energy office actually offers. If the number still works without the federal credit, it is a good project. If it only works with the credit, it is a project for a different owner. The solar payback estimator lets you try the inputs.