Farming Upwards

Thesis

Vertical farming is not an agriculture business. It is an electricity conversion business with a produce distribution channel attached, and the great majority of failures in the sector followed from management teams and investors who did not price it that way.

This is not a dismissal. The model works in defined circumstances, and those circumstances can be specified precisely. What it will not do is what the sector spent a decade claiming it would do, and the gap between those two statements destroyed several billion dollars of capital between 2021 and 2024.

The substitution being made

Begin with what the technology actually replaces.

A field crop receives sunlight at no cost. Solar radiation arriving at the surface is free, delivered without infrastructure, and available across the entire cultivated area of the planet. It is also uncontrolled: it varies by season, latitude and weather, and it arrives together with pests, frost, drought and hail.

Controlled-environment agriculture trades that free, unreliable input for a purchased, reliable one. Everything else follows from that single substitution — the yield gains, the water savings, the year-round production, and the cost structure that has repeatedly proved fatal.

Where the money goes

The unit economics of a fully enclosed vertical facility divide into three parts, and their relative weights are consistently misjudged.

Energy is the dominant recurring cost and does not fall with scale. Lighting is the largest component; the balance goes to dehumidification, because plants transpire nearly all the water they take up and it must be condensed out continuously, and to cooling, because every watt of light becomes heat inside a sealed building. Published figures for leafy greens cluster in the range of thirty to forty kilowatt-hours per kilogram of saleable product. At industrial electricity prices in most developed markets, that alone approaches or exceeds the wholesale price the crop commands.

Capital is heavy and long-lived. Racking, lighting, climate plant, irrigation, controls and building shell commonly total between fifteen hundred and three thousand units of currency per square metre of growing area — an order of magnitude above a glasshouse and two above open field. Depreciation on that base is a fixed charge in every quarter regardless of sales.

Labour is the cost most consistently underestimated. Seeding, transplanting, harvesting and packing remain substantially manual, and stacked growing tiers are ergonomically hostile. Several operators found labour exceeding energy in their cost base, which was not in any of their original models.

The lever that is nearly exhausted

The historical argument for the sector rested on lighting efficiency. Between roughly 2010 and 2020, the efficacy of horticultural light-emitting diodes improved dramatically, and the projection of that trend underwrote most of the business plans written in the period.

That projection no longer holds, and the reason is physical rather than commercial. Efficacy is measured in photosynthetically active photons delivered per joule of electricity. The theoretical ceiling for the relevant wavelengths is a little under five; commercial fixtures now reach roughly three to three and a half. The remaining headroom is therefore around thirty to forty per cent in total, not the successive doublings that the previous decade delivered.

Any business case requiring lighting cost to fall by a further factor of two is requiring something that cannot occur. This single observation invalidates a substantial portion of the sector's published projections, and it was available to anyone who looked it up.

Which crops can and cannot work

The viable set is determined by revenue per unit of light energy, and it is narrow.

Crop class	Wholesale value	Growth cycle	Verdict
Microgreens, herbs	Very high	7–21 days	Clearly viable
Leafy salads	Moderate	25–35 days	Marginal; location-dependent
Strawberries, vine crops	High	Months	Viable in glasshouse; rarely stacked
Staple grains, pulses	Very low	Months	Structurally impossible

The final row deserves emphasis, because it is the claim most often made in general coverage and it is not close. Wheat trades at a small fraction of the price of salad by weight, requires a long season, and devotes most of its captured energy to inedible structure. No conceivable improvement in lighting or automation brings it within range. Vertical farming will not feed populations in the caloric sense, and proposals that assume otherwise should be read as marketing.

The middle option that is usually ignored

The strongest performer in controlled-environment agriculture is not the stacked warehouse. It is the modern glasshouse, and it has been quietly profitable for forty years.

A glasshouse takes light for free and buys only the environmental control: heat, carbon dioxide enrichment, screening, humidity management and hydroponic feeding. Yields in advanced operations reach an order of magnitude above field production for the same crops, on a fraction of the water, with a capital base far below an enclosed facility and an energy bill dominated by heat rather than light — heat that can frequently be sourced as waste from an adjacent industrial process.

The sector's error was framing the choice as warehouse versus field. The actual competitor, in nearly every market, was the glasshouse, and it wins on cost in every location where light is adequate.

Where the enclosed model does win

Four conditions, and a project should satisfy at least two.

1. Expensive or unreliable imports. High latitudes, islands and remote regions where fresh produce arrives by air, arrives degraded, or does not arrive.
2. Very high local produce prices relative to industrial electricity — the ratio, not either figure alone.
3. Contamination-critical output. Sealed production offers verifiable freedom from pathogens and residues, which commands a genuine premium in some supply agreements and is not obtainable in a field.
4. Non-food output. Pharmaceutical precursors, seed multiplication, plant propagation and research all attach value per square metre that dwarfs any edible crop, and they are where much of the sector's surviving capital has quietly relocated.

Verdict

The technology functions exactly as described. The failures were not technical; every bankrupt operator grew the crop successfully.

They failed because a business whose principal input is electricity and whose principal output is a low-value commodity has a margin determined by a spread it does not control, and because expansion was financed on an efficiency trend that had a physical ceiling.

Underwrite it as a power-intensive manufacturing business with a perishable product and a three-week inventory cycle. On that basis a small number of projects are attractive. On the basis the sector used, almost none were.