Economics and Profitability

A vertical farm's losses are not the industry's fate—they are decided by how you build it

A project you are considering, or a factory you are already running. Many of you, I suspect, opened this page without having been able to reach a verdict on whether it pays.

What usually dulls that judgment is the fixed label that “a vertical farm is a loss-making industry.” If the whole industry runs in the red, then surely my own will too—and you lay the industry-wide rate straight over your single project. But the substance of a loss differs completely from one project to the next. Even within the same “red,” there are losses that vanish if you wait and losses that do not vanish however long you wait, mixed together.

Over the past ten-plus years I have been involved in the production and start-up of more than ten vertical farms. I took part in the start-up of factories among the largest in Japan. What became visible to me there is that the real dividing line is not whether it is “an industry” but “how you build it.” This article is about telling which side your own project is on, through three questions: sales channels, utilization, and cost.

The reversal: the more impressive the facility, the bigger the loss

“60% of vertical farms run in the red.” In the news and in trade journals, you see headlines of this kind constantly. See it enough times and you come to take it, vaguely, as “that’s just how this industry is.” But when you actually look at the factories one by one, there is profit and there is loss, and where that difference comes from—this is the genuinely interesting part.

When I think back over the factories I have been involved with, one thing nags at me. Among the factories running in the red, somehow “impressive facilities” are mixed in. Places that brought in the latest equipment, are large in scale, and got splashy coverage in the news. Conversely, the ones that quietly keep going often have older or smaller equipment. By ordinary logic, better equipment should be the advantage, yet on the floor it can look the opposite. “Huh, why is that?”—don’t you wonder?

This is worth teasing apart carefully. Unexpectedly, when you look at the field-survey numbers, there is no link between larger area and profitability. If anything, in PFAL the facilities of 1,000 m2 or more in cultivation area show profit or break-even at 50%, lower than the 65% for those under 1,000 m2. In Greenhouses there is almost no difference between large and small (72% versus 71%), and the survey reads the large facilities running in the red as “the burden of depreciation, given the large capital investment, having an effect” (ref: 19). What is linked to profitability is not area but high yield and productivity—the same field survey shows a tendency for facilities with higher yield to have a higher proportion in the black (ref: 19). The unit price of construction itself does fall as you raise the scale (one estimate says raising the scale 100-fold lowers the per-unit cost by about 55% / ref: 1), and yet profitability once you are operating is not decided by area. This discrepancy is the hint. Install the latest equipment first, without working backward from sales channels, and that investment bears down as a fixed cost before any revenue comes in. LED, HVAC, automation lines—the monthly depreciation accrues even at zero utilization. In other words, the part of “impressiveness” that tips you into the red is not scale itself but the part where depreciation was front-loaded without working backward from sales channels. That the field survey explains large-scale losses by depreciation is precisely this, as I read it. The plain, old factories have either long since taken that weight off, or never shouldered it in the first place.

What is more, among the factories that get splashy coverage, there are projects where the equipment came first. We can do this kind of cultivation, this kind of automation—the equipment is there first, and where, at what price, and how steadily it will be sold is put off. What does one head of lettuce fetch on the market, and does that price cover even the depreciation? Make the equipment big first while that backward calculation is thin, and depreciation hits first and the loss grows bigger too.

So whether something is impressive has almost nothing to do with the loss in itself. The real dividing line is “whether the three—cost structure, sales channels, and utilization design—mesh together.” The factories that quietly keep going are not running old equipment; within a fixed cost scaled to their means, they have locked in their buyers and their utilization first. The reason impressive facilities look the reverse is, I think, that the result of letting the equipment run ahead while leaving those three behind comes out into the open in a conspicuous form.

The research side, too, repeatedly backs up this structure where “the high initial investment and operating cost bite first.” High initial construction cost and high operating cost—this is reported, across countries and types of research, as the primary barrier blocking the commercial deployment of vertical farms (ref: 1, 2, 16). In indoor vertical farms in particular, one estimate has lighting alone accounting for about 80% of total power consumption (ref: 3), and another puts the energy per unit yield on the order of hundreds of times that of open-field farming (ref: 4). “The more impressive the equipment, the bigger the loss” is not an impressionistic claim but a story about the structure in which the weight of fixed costs and energy bites first.

Funding only covers the equipment

It is not that impressiveness itself is the problem; it is that the equipment runs ahead while the three fail to mesh. So far, so good. But one snag remains. Why does that phenomenon of “the equipment coming first” happen so repeatedly?

Once or twice, and it can be filed away as an individual management misjudgment. But this is happening as a pattern, in the same shape. Even though both the weight of depreciation and the importance of working backward from buyers seem like things you’d grasp readily enough. Is there a force at work, from the very start, that makes the people who start vertical farms and the people who put up the money “want to build something impressive”? Or is it simply that cultivation itself is interesting and the sales channels get put off?

Both, probably, but what is really at work is “where the money comes from,” I think. Cultivation is interesting and the sales channels get put off—that surely happens. But that is a matter of individual temperament, and temperament ought to vary from person to person. If the same failure repeats as a pattern despite that, you do better to suspect a force at work beyond individual differences—that is, the source of the money.

What is going on? A vertical farm’s initial investment tends to reach a scale that self-funding cannot meet, and it often rides on subsidies or on outside capital coming in as a new venture. In fact, in Greenhouses 58% of operators use energy-related subsidies, while there are also facilities operated without any subsidy—31% overall, and 46% for PFAL, by one survey (ref: 19)—and the source of funds splits widely by project. Here the direction gets twisted. What I feel, having been involved in many start-ups, is that what the people putting up the money most want to see is the picture of a state-of-the-art vertical farm built in their region. When that is so, the rational behavior for the side going after the funding tilts not toward locking in a sales channel that covers the depreciation first, but toward getting an impressive proposal through. Less a theory than a gut read, from watching things proceed in that order over and over on the floor.

This is not limited to subsidies. The same thing happens with outside business capital. Companies that assembled the world’s most advanced equipment on abundant fundraising, only to stall because they could not make the per-unit balance work at commercial scale—such cases have been reported repeatedly overseas, too; names like AeroFarms, Bowery, and Infarm come up. What I myself have seen on the floor in Japan is rather the same tendency: whether the source of funds is a subsidy or business capital, the more capital there is to build something impressive, the more easily the backward calculation of the sales channel gets put off. So even when serious, capable people act rationally, they can fall the same way. This is the crux. Of the three—cost structure, sales channels, utilization design—what funding most easily covers is the equipment, that is, the part of the cost structure that grows heavy; the sales channels and the utilization design are not something capital locks in on your behalf.

That “putting money in does not, on the whole, make things move easily” shows up in the numbers, too. Figures like the “60% in the red” you see in the news come from sources and years all over the place; an old trade-journal column says 75% in the red, another survey says 49% in the red—they swing widely from report to report (ref: 5, 6). In the newest authoritative field survey, operators in the black or at break-even in their most recent accounts are 64% overall—meaning the red is roughly a third (ref: 19). That the numbers move this much from report to report is itself the best proof that you must not apply this rate straight to your own project. What you should watch out for, rather, is newness. In the same field survey, facilities that began cultivation in 2021 or later are still only 20% in the black, with the red at 52%. The survey sees the reasons as the large burden of depreciation against the investment, and production not yet being fully stabilized (ref: 19). The newer a facility started up on abundant capital, the more squarely it shoulders the weight of the ramp-up period.

Tell your own project apart with three questions

So far this has been the whole-picture story of “why so many factories topple into the red.” But what the person reading this most wants to know is, probably, somewhere else. “So which side is the project I’m considering, or running right now, on?” Taking the industry-wide figure of “60% in the red” and applying it straight to your one project—then either giving up or, conversely, feeling reassured—is the single worst thing to do. The industry-wide loss rate and the outlook for the one project in front of you are different things. The industry-wide rate is no more than a figure that averages together factories with different ways of meshing; it does not apply equally to a project where the three are aligned and one where they are missing. So, on which factor is your own project likely to stumble, and what should you look at first?

Order matters. First, look at sales channels. Of the three, the sales channel is the one most easily put off and the one least recoverable. The distinction comes through fairly well with a single question: “Before you build, is it decided who buys, at what price, and how much every week?” A project that cannot answer this with numbers is dangerous however impressive the equipment. Conversely, if this is filled in, both the cost structure and the utilization design can be worked out backward from it. Only once the sales channel is set do the necessary scale and the ceiling on allowable fixed cost emerge. If the order is reversed—the equipment there first, hunting for buyers—that itself is a sign of the loss side.

Next is utilization. Because a vertical farm’s depreciation runs even when it is stopped, you should ask whether the planned utilization rate is too high—that is, “whether the proposal is built on full utilization as the premise of profitability.” In my experience, in the ramp-up period some racks sit empty for months, and the yield, too, does not reach the planned value at first. So a plan premised on profitability at full utilization topples into the red the moment that premise breaks. What you should look at here is not the single threshold of “at what utilization rate do we turn a profit.” The necessary utilization rate changes by crop and by type. What you should look at is whether the ramp-up curve—how fast the utilization rate realistically rises, and when it reaches profitability—fits inside the span your funding lasts. Last is cost structure, and here it is whether you can lay out “what share of revenue electricity and depreciation eat” on a single sheet. A project that cannot is still stuck at the impressiveness stage. So do not apply the industry’s 60% figure to your own project. Instead, judge which side you are on by these three: whether the sales channel is filled in with numbers, whether the ramp-up curve reaches profitability while the funding lasts, and whether you can lay out the cost on a single sheet.

That profitability moves by orders of magnitude with sales channel and crop comes out clearly in model estimates, too. Even with the same closed-type (artificial-light) vertical multi-tier equipment, a leafy crop like lettuce can become profitable from a small scale of 17–38 m2, whereas strawberries, by the estimate, require upward of 16,000–110,000 m2 (ref: 1). That said, this strawberry figure is a rough estimate as of 2022. At the time there were almost no actual cases of commercially producing strawberries in a closed type (domestic efforts began in 2021), and the value was extrapolated from other crops based on the highest yield obtained in the lab. It is a figure that reflects the situation at the time—that the yield improvements made in leafy greens had not yet caught up for fruiting vegetables—and with advances in cultivars and cultivation technology, or conversely with the recent rise in energy and material costs, this line still moves today. It is more accurate to take it not as a settled required scale but as a direction: “even with the same closed-type equipment, leafy greens and fruiting vegetables differ by an order of magnitude in how easily profitability is reached.” Moreover, this 17–38 m2 is a desktop minimum for the case where the assumed unit price and yield stand up as is; even a small downward shift in the selling price swells the area needed to make it work considerably (ref: 1). In a phase where costs are rising, a footprint that small is not easily achieved. There is even an estimate that growing a grain like wheat indoors makes the cost of lighting (light energy) about 100 times the market wholesale price (ref: 7)—what you grow and where and at what price you sell it bite first as the precondition of profitability.

The sense that the probability of profitability splits by crop and buyer is backed up by overseas cases, too. But when you draw on them, you need to look at what they target. In a model estimate of small farms in London, a scenario selling high-value-added leafy greens at a high unit price has a profitability probability of nearly 60%, while a scenario moving many crop varieties at low prices drops to the low 30s% (ref: 8). This targets not indoor vertical farms but soil-based small farms, and under conditions that leave depreciation and financing out of the calculation, so it is not a figure from which the profitability of a vertical farm can be read as straightforwardly “achievable.” Better to take only the direction—that the distribution of outcomes changes with a single way of placing the sales channel. To begin with, the very crops that hold up commercially under environmental control concentrate almost entirely on leafy greens, herbs, and some fruiting vegetables (tomatoes and the like), and in terms of the world’s farmland area they amount to only a few percent (ref: 9, 10). But here the line splits clearly by how you build it. What runs commercially in closed-type (artificial-light) vertical farms is almost all leafy greens—lettuce above all; in domestic PFAL, over 90% is leafy greens, and over 90% of that is lettuce (ref: 1). Fruiting vegetables like tomatoes turn a profit mainly on the Greenhouse and glasshouse side, and strawberries in a closed type are still at the trial stage. So apply “fruiting vegetables, too, hold up commercially” straight to a closed-type LED factory, and it parts ways with reality. Indeed the field survey, too, shows the proportion in profit or break-even varying by type: Greenhouse and combined types are over 70%, whereas PFAL stays at about 50% (ref: 19). Even within the same vertical farm, profitability comes out more easily or less easily by type.

Two cards you can still play in a factory already built

Telling them apart with those three is, certainly, easy to grasp. From sales channels, to the ramp-up curve of utilization, to laying out the cost on a single sheet—you can check in that order for a project you are about to build. But here something starts to bother you. This is a story about a “not yet built” project. You may be told the sales channel being out of order is dangerous, but the equipment is already built, the depreciation runs every month, and the losses continue. Factories like that are, in reality, not few. In that case, do the three from earlier become a “too-late diagnosis”? Or is there still room left to move, even in a factory that is already running?

It is not necessarily too late. If anything, a factory already running has one strength that a project about to be built does not: “the actual numbers are already in hand.” Not the forecast of a proposal, but knowing at what price how many heads sold, what the utilization rate was, and what the electricity bill came to. So the diagnosis can in fact be made accurately. Here, you think separately about what is locked in and what can be moved. What is locked in is the equipment and the depreciation tied to it, plus conditions hard to move once built, like the unit price of power and the location. Depreciation accrues whether you sell or stop, so it is not the first line item to go cutting at. That said, there are situations where room remains to move the cash-flow weight through moves like converting to a lease, selling off some equipment, a write-down from retiring assets, or transferring the business. Even so, what daily operation first puts within reach is the remaining two—sales channels and utilization. And the order is the reverse of a project you are about to build.

First, look at utilization. The reason is simple: since the depreciation is already running, dividing that fixed cost over as many heads of revenue as possible comes first. If you are at 50% utilization now, how do you fill the equipment you are leaving empty? Change the crop, lean toward fast-turning leafy greens, get the stopped racks moving. Since the equipment is already there, there are cases where room remains to push increased production itself ahead without additional investment. This tends to be the growth headroom of a factory that has already been built.

Next, sales channels—but before “looking for new buyers,” look at the price of your current buyers. Pricing that puts product out too cheap just to move it is something I often saw in stalled factories. With the same number of plants, just holding one fixed-unit-price exit by contract changes the profit-and-loss considerably. Can you shift even a part toward an exit where volume and unit price are promised—a weekly fixed contract with a processor or a food-service operator, rather than selling heads one by one into the supermarket spot market? So rather than lamenting the locked-in cost structure, you move the question to how far you can thin that fixed cost by “utilization x fixed unit price.” The equipment cannot be changed, but how much you make that equipment work and at what price you put it out are still in your hands. It is not a too-late diagnosis but a story about how you play the two remaining cards.

That there is room on the utilization side is shown at the research level, too. But the target requires care. In Greenhouses, there are several reports that rearranging the cultivation schedule to fit the season, or combining heat-pump operation and energy-saving equipment, can improve operating energy and annual revenue (ref: 11, 12). These are stories about Greenhouses and tomatoes premised on the seasonal variation of outside air temperature and solar radiation, and they cannot be transferred straight to closed-type, PFAL factories with no seasonality. There are also studies of energy-saving equipment targeting the closed type, but they center on simulation and the accumulation of one research group, and have not reached the point where you can flatly say it will turn out this way at commercial scale (ref: 13, 14). So it is best read at the level of “there is room to move, depending on type.”

What to look at is not the sign of the loss but its slope

The view of two remaining cards aptly captures the discussion up to here. Finally, one thing I want to add. So far I have been telling a story of “there is room to move,” but there are also, in reality, projects where you move both sales channels and utilization and still cannot thin the fixed cost—that is, projects already finished at the design stage. The question is where to draw that line. And to simply rule “red is bad, black is good” is, probably, also not right. The loss that is unavoidable in the ramp-up period and the loss that is structurally inescapable need to be handled separately. So how should this line be drawn?

For drawing the line, the very same three questions can in fact be used as they are. Fill the sales channel at a fixed unit price, take utilization to a realistically reachable level, and if the cost structure still does not reach profitability, then it is a project where, at the design stage, the fixed cost exceeds the ceiling of revenue. The sum of electricity and depreciation exceeds, from the start, the revenue that comes in at a realistic utilization x fixed unit price. This is not a shortfall of effort; it is settled as arithmetic. Once you are here, regardless of whether it is the ramp-up period, there is no escape as long as you make that crop with that equipment. The line that recognizes withdrawal or a change of crop as a number, not an emotion, is here.

Here, one real example. Brick Street Farms in Florida, USA, fell into financial difficulty again the year after its founder transferred the business, now under a new owner, and the building went into receivership. Even when the owner changes, if the revenue structure itself is not touched, the red carries over to the next owner. This is exactly a story about the slope (whether the structure moved), not the sign (who holds it, whether this term is in the red).

On that basis, to connect to the idea of not judging losses as good or bad: what you should look at is not the sign of the loss but which way the loss is moving over time. The ramp-up-period loss shrinks every month as the utilization rate rises and the yield stabilizes. Even within the same loss, if the slope is heading toward the black, that is a normal process of the investment heading toward recovery, and not something to rush to judge. Conversely, if the loss does not budge even after both utilization and yield have risen fully, that is no longer a ramp-up problem but proof that the structure has hit its limit there and will not move. The criterion that separates the two is “whether time is on your side.” If the loss shrinks the more time passes, it is a ramp-up-period loss; if the loss does not move however much time passes, it is a structural loss. What the former needs is cash-flow management and patience; what the latter needs is a redesign or a withdrawal—the prescriptions are completely different. To lump everything under the single word “loss” and slap a good-or-bad verdict on it is the most dangerous of all. Because under the same sign, losses that vanish if you wait and losses that do not vanish however long you wait are mixed together.

This view that it is “settled as arithmetic” is consistent with the research side, too. High initial capital and electricity cost structurally limit profitability. For low-value-added crops like staple grains in particular, profitability does not hold up with current technology—on this point, multiple studies and cases agree (ref: 7, 9, 10, 15, 16). The convergence that profitability and cost are the greatest challenge appears repeatedly in prior research, from Japan’s field surveys to overseas reviews (ref: 17, 18, 19). That projects where “you moved both sales channels and utilization and still could not thin the fixed cost” exist in reality is, less a difference of effort, because the ceiling is decided by the combination of the first crop and equipment.

What to look at is not the sign but the slope, and what decides that slope is, in the end, the sales channels and utilization described up to here, and the ceiling of fixed cost beyond them. Stop surveying the whole through the single phrase “a vertical farm is a loss-making industry”; re-examine your one project through the meshing of three factors, and read it by whether that loss shrinks over time or stays put. Read it that far, and you become able to decide not by entrusting your judgment to the single word “loss,” but by the slope specific to your own project.