Vertical Farming: The Economics Behind Growing Up

The pitch for vertical farming is compelling: stack crops in a warehouse under LEDs, control every variable, use 95% less water and a fraction of the land, grow year-round next to the city that eats it. Between 2020 and 2022 investors put billions behind it. Then several of the largest and best-funded operators — AeroFarms, AppHarvest, Infarm and others — went bankrupt or retrenched. What happened is a lesson in the difference between a technology that works and a business that does, and it clarifies where vertical farming genuinely belongs.

How it works

Vertical farming is a subset of controlled-environment agriculture. Plants grow in stacked trays or towers indoors, fed by hydroponics (roots in nutrient solution), aeroponics (roots misted) or occasionally aquaponics (fish waste as fertiliser). LEDs tuned to the wavelengths plants use most replace sunlight; HVAC systems control temperature, humidity and CO₂; sensors and software adjust everything continuously. Because the environment is sealed, there are no pests, no pesticides, no weather and no seasons. Yields per square metre of floor space can be hundreds of times those of a field, and water use is a small fraction because nothing evaporates or drains away.

The physics of the problem

Everything a field gets free, a vertical farm buys. Sunlight becomes electricity: producing the light for one kilogram of lettuce takes roughly 10 kWh or more, and cooling the heat those LEDs produce costs more on top. Energy is typically 25–40% of operating cost, and it scales with every tray you add. That's tolerable for high-value, fast-growing crops with little inedible mass — leafy greens, herbs, microgreens — and hopeless for staples: wheat, rice or potatoes grown under LEDs would cost many times their market price. The technology is not "the future of agriculture"; it's the future of a specific slice of it.

Why the big operators failed

• Energy costs rose sharply in 2022 just as the industry scaled, wiping out thin margins.
• Capital intensity. Purpose-built facilities cost tens of millions of dollars before the first harvest, and cheap venture capital dried up as interest rates rose.
• Competing with a cheap incumbent. Field-grown lettuce trucked across a continent is still cheaper than lettuce grown a mile away under lights. Consumers pay a premium for "local" and "pesticide-free" only up to a point.
• Operational complexity. Running a sealed biological system at scale is hard; a single pathogen outbreak or HVAC failure can destroy a crop cycle. Several operators had automation ambitions that outran what worked.
• Overexpansion. Building multiple large sites before unit economics were proven at one.

What's working

The survivors and the new entrants look different from the boom-era companies:

• Right-sized and crop-focused. Facilities matched to a specific regional demand for greens, herbs and berries, rather than continental ambitions.
• Hybrid greenhouses. High-tech glasshouses that use sunlight and supplement with LEDs capture most of the control benefits at a fraction of the energy — this is where the Dutch, the world leaders in controlled agriculture, have concentrated.
• Cheap or captive power. Farms sited next to renewable generation, using off-peak electricity, or co-located with waste heat and CO₂ from industry. Where electricity is cheap and land or water is scarce — the Gulf, Singapore, parts of the US West — the economics flip.
• Seed and plant production. Growing seedlings, tissue-culture plantlets and seed for conventional farms — a high-value, low-volume use where control matters more than cost per kilo.
• Pharmaceutical and research crops. Plants grown for medicinal compounds, and research facilities studying crop genetics under perfectly repeatable conditions.
• Food security contexts. Island nations, cities with fragile supply chains, and — in a very different setting — space missions, where the alternative to growing indoors is not growing at all.

Technology still improving the odds

LED efficiency continues to climb, cutting the biggest cost. Breeding programmes are producing crop varieties optimised for indoor conditions — compact, fast, tuned to LED spectra — rather than adapting field varieties. Better automation for seeding, transplanting and harvesting is arriving from adjacent industries. Digital twins and machine learning tune light and nutrient recipes crop by crop. And energy storage paired with on-site renewables can make the electricity problem smaller. None of these changes the fundamental physics; all of them shift where the break-even lies.

The honest assessment

Vertical farming is a real, useful technology for high-value crops in places where land, water, climate or supply chains make conventional farming hard, and a poor one for feeding the world cheaply. Its lesson for every "future of X" technology is the same: engineering that works is necessary, and unit economics against the incumbent decide the rest.