Vertical Farm Basics and Overview

The challenges of a vertical farm are mostly decided before you build it

Why do so many vertical farms get stuck on the same problems, as if stamped from the same mold? If the troubles differed from farm to farm, that would be individual bad luck. But in reality, the list of challenges overlaps to an uncanny degree. When coincidences line up this neatly, the first thing to suspect is not the floor itself, but the design choices made one step before it.

Power bills and labor costs are mostly on the symptom side

The rows of frilled lettuce look more impressive the longer you look. I have seen plenty of PFAL vertical farms, and the gradations of green line up beautifully; the technology itself I admire without reservation. But the moment I think, “where is this actually being sold?”, the story usually turns vague. On the supermarket shelf, next to ordinary lettuce, priced just a little higher. And when I listen to struggling farms — in the PFAL operations I have seen — it nearly always ends up at power bills and labor costs. But think it through and something nags. Back at the point where they decided to “grow lettuce” and “wholesale to supermarkets,” wasn’t something already decided? Weren’t the farms that failed carrying a combination of what they grow and where they sell that was tight from the very start? That is what comes to mind.

Power bills and labor costs are, no question, painful. But most of them are on the symptom side. It is like seeing a fever and reaching for the fever reducer: the underlying weight of the cost is decided much further upstream. Where is it decided? At the moment you decide “what to grow, and where to sell it.” Here, quietly, the ceiling on the cost that farm can carry gets set. Compared with open-field farming on soil, a vertical farm is structured to take on a “lump of fixed cost” right at the start, in the form of the building, the equipment, and the electricity. So unless you have chosen a way of selling that still leaves a profit after taking on that heavy cost, no matter how hard the floor works, there remains a cost floor it cannot cut away.

Let me lay out the map of this article up front. The upstream choice of “what to grow, and where to sell it” sets the ceiling on the profit that farm can reach in a year: in other words, the floor on its cost. Daily operation, on the other hand (how you use electricity, how you run people, your marketable yield), decides how close to that fixed ceiling you can bring your actual results: the rate at which you reach it. The upstream sets the ceiling, and the people on the floor set the rate at which you reach it. Both move the revenue. So as a matter of order, before you build, you raise the ceiling itself upstream; after you build, you take everything you can through operation. Keep this view, and the rest of the discussion follows more easily.

What you often see is an upstream choice pointing the wrong way. You adopt the most cost-heavy way of growing, and then carry it to the shelf where the price is most fixed and where you are most easily compared on price. Lettuce is exactly that: a shelf with ordinary lettuce sitting next to yours is a place where the buyer sees it as “the more expensive version of the same thing.” The moment you place it there, you are close to declaring, of your own accord, “I will compete on price.” But if you compete on price, the side carrying heavy fixed cost cannot win. In other words, a good number of the farms that failed did not so much fail on the floor as, at that first combination, choose for themselves a hard-to-win arena: a shelf where you get compared on price. In the range of PFAL I have seen, that is how it feels in many cases. Ending up at the power-bill story is partly the result of running desperately in that arena; it is not the whole of the cause. And conversely, this is also the place with the most hope. Swapping out equipment is hard, but “who you sell to, and how you get them to feel the value” still has room to be reconsidered, more often than not.

The literature points the same way. Vertical-farm production that runs on a commercial basis leans almost entirely toward leafy greens, herbs, and berries, and these sit in a band of only a few percent of the world’s calorie supply. Once you get to staple grains like rice or wheat, with today’s technology and cost structure the numbers do not work. A review 2 and an opinion piece 1 point the same way. So “what to grow,” before any talk of taste or craft, has already decided in advance whether you are even in an arena where the numbers can work.

“If you compete on price, the side with heavy fixed cost cannot win” is also pointed out plainly in one essay. A vertical farm has to fight inside the thin-margin distribution of food while carrying high upfront investment and high energy cost. Whether it can survive economically is the single biggest barrier to its spread, it argues (3). There may be room to raise efficiency technically, but that does not translate directly into the price the consumer pays or into profit.

Why so many farms gather in the same arena

Then just choose, from the start, a sales channel where you do not get compared on price: you might think. In fact, the people at the struggling farms probably knew that too. And yet, somehow, so many people drift toward growing lettuce and wholesaling it to supermarkets. Even when they faintly sense it is a hard-to-win arena, there is a reason they choose it anyway.

One reason is simply that it is easy to start. Lettuce to the supermarket: both the way to grow it and the place to sell it are already “there.” Visible. So the anxiety of starting is small. A “sales channel where you do not get compared on price,” on the other hand, has no visible outline at first, so stepping out toward it all at once carries the uneasiness of putting your foot out into fog. As a result, what you are choosing by leans toward “can I start” rather than “can I win.” Choose by ease of starting, and everyone gets pulled toward the same entrance.

And the choice of lettuce to a mass retailer has its own reasonable logic. Among leafy greens, lettuce has the most established cultivation technique, and its failure rate in the startup phase is low. A deal with a mass retailer, since the volume is consolidated, supports your cash flow through the startup phase. The way subsidy programs are designed also tends to push these easy-to-grow, easy-to-sell models. So this is less “because they did not study” than, in large part, people gathering at the same entrance as the result of a rational choice. And the same entrance connects straight to the same arena. So in the end, people naturally pile up at the most crowded place, the place where they cut each other down most on price.

What makes it troublesome is that this choice does not wear the face of “failure.” Grow it and it takes shape; it lines up on the shelf too. At a glance, it looks like proper forward progress. The hardship comes to the surface much later, once the fixed cost starts to bite. So even the person who started it can hardly tell where they got the arena wrong. The easy-to-start road sometimes leads to a hard-to-win place. If so, what is being asked is not “did you know,” but rather “did you have the resolve to spend time, from the start, searching for the unseen sales channel.” This question of how to search for and design that unseen sales channel is a large theme that deserves to stand on its own, so I will not go deep into it here.

“People pile up at the place where they cut each other down on price” is backed up by surveys of the buying side too. Ask consumers about vegetables produced in a vertical farm, and the result is that price weighs decisively on the decision to buy or not. What is more, a sizable share voice concerns like “isn’t it artificial” or “is it safe,” and in one survey over 60 percent raised some kind of unease (a Taiwan survey, 390 respondents, 4). In other words, put it straight onto an ordinary shelf and the buyer tends to set it in the price arena as “pricey lettuce, with a bit of unease about it.” But the same survey also shows that when “who grew it” is visible, as with a brand tied to a university or a company, people respond that they are willing to pay a little more. That meshes neatly with the idea that the arena changes depending on how you choose your sales channel, doesn’t it?

Separate the challenges into the layer that sets the ceiling and the layer that closes the gap

I have been saying that “what to grow” binds where you sell. This works not only on where you sell but on the scale you need. Choosing a crop is not just choosing “what to plant”; it is choosing, along with it, the “smallest size” at which that crop can stand. A leafy green turns over many times on a short cycle and lets you stack the racks many tiers high. On the other hand, a crop that takes time to bear fruit, stands tall so you cannot stack many tiers, and on top of that yields little per unit area. For these two, the floor area you need to reach the same sales is, from the start, a different world. So the moment you decide the crop, you quietly bring along the size of building you need and the weight of fixed cost you will carry.

What I want to flag here is that scale cannot be added smoothly after the fact. The building and the equipment go in together in a certain unit, so it is hard to adjust by “I’m a little short, so let me widen it just a bit.” This is also what I felt firsthand watching PFAL startups. So the choice of crop, just as much as it binds your sales channel, sets the floor on cost through scale. It is, in fact, the same root as the sales-channel story.

This property of “the crop deciding even the scale you need” shows up in a fairly extreme form in model estimates too. In an estimate for one group of vertical-farm facilities, the smallest cultivation area at which lettuce reaches commercial break-even is put roughly at the order of tens of square meters, whereas trying to do strawberries on the same facility assumptions (in the PFAL, closed-type estimate), the area needed to break even is estimated to leap up to anywhere from ten-odd thousand to over a hundred thousand square meters. Just because the crop differs, the order of magnitude of the area you need changes wholesale. And the scale you need is not decided by the crop alone. Even for the same crop, the area needed to break even moves a great deal depending on the price you assume you can sell at (5). “What, and at what assumed selling price” is quietly deciding the scale you need.

On top of that, it is easier to organize the challenges by looking at which layer decides them, split roughly into three layers. This is not a “the main causes are these three” attempt at completeness; I am putting it down as an easy-to-organize frame for telling apart the layers you can move later from the layers you cannot.

At the very bottom is the combination of building and crop. This is the layer that decides the ceiling on the profit that farm can reach in a year: the total itself. It is the hardest, and once you have built, no amount of effort on the floor can undo it. The things you cannot do anything about without rewinding all the way to the design are, usually, in this layer.

Above it is the layer of “where you sell and how you sell.” This is the layer that decides how aggressively you can push toward the fixed ceiling. It looks hard, but in fact there is still room to work. Even for the same crop, who you hand it to and as what kind of value often still has room to be reconsidered. The “search for a sales channel where you do not get compared on price” I touched on earlier is a matter of moving this layer.

At the very top is the layer of “daily operation.” How you use electricity, how you run people, your marketable yield, and so on belong here. The layer that surfaces as your troubles is this one, but its role is to close the gap to the ceiling set by the two layers below: to bring your actual results as close to that ceiling as possible. So if I were to draw the lines: “building and crop” is the unmovable, given layer that sets the total, “where you sell and how you sell” is the layer that pushes on the ceiling, and “daily operation” is the layer that closes the gap to the fixed ceiling. As a matter of order: before you build, you go raise the ceiling itself in the lower two layers; after you build, you take everything you can through operation. Get this order wrong, and try to claw it all back through operation alone while the ceiling stays low, and the range you can cut tends to hit a wall.

The rate of reaching it on the floor is decided by people

Let me go a little deeper into that “daily operation” layer. Even putting out the same crop to the same shelf, farms split into the black and the red. The latest field survey also shows that, when you split by their financials, the redder the operator, the higher their cost ratios for labor, electricity, and depreciation (the FY2025 field survey, 7). Given the same ceiling, the rate of reaching it differs from farm to farm. And what actually moves that rate is the people on the floor.

Why does marketable yield swing with people? Managing a vertical farm demands, within a single person, the power to cut across plant physiology, equipment, sanitation, production management, and cost control. Reading light, temperature, humidity, CO2, and nutrient solution for each growth stage; troubleshooting HVAC and the nutrient-delivery system; improving from sensor data; hygiene management; and a view of cost. A person who bundles this whole skill set is, to begin with, hard to hire and takes time to develop. Almost no one has studied both plant physiology and engineering systematically, so in practice it is more common to hire someone without experience and grow them on the floor. So from one floor to the next, the thickness of that “bundling power” differs, and that shows up as the difference in marketable yield and uptime.

What is more, this is not a vertical-farm-only circumstance. Manufacturing as a whole is short of hands. In a 2017 survey by the Ministry of Economy, Trade and Industry, over 94 percent of manufacturing firms answered that they were short of hands (8). In a 2023 survey by the Ministry of Health, Labour and Welfare, the job-openings-to-applicants ratio in manufacturing was about 1.74, far above the all-industry average of 1.27 (9). A vertical farm is no exception to this, and on top of it rides the particular difficulty of “needing cross-cutting skills.”

Large-scale farms have one more structural circumstance. The larger the scale, the more capital it takes, so they often begin as a division or affiliate of a big company, and it is common for a transferee from headquarters or the parent company to come in as the farm operations manager. A transfer, as a company mechanism, rotates on a set period, so if the knowledge painstakingly built up on the floor does not pass well at handover time, it tends to return to square one every time the person changes. This is not a story of someone’s effort being insufficient; it is a matter of fit between a mechanism that periodically shifts people around and a floor where mastery takes time. I have been involved in operations support for PFAL farms at Farmship, and whether or not it was a transfer, how to leave the built-up knowledge on the floor was something that mattered at every farm.

So this is in no way a story of “polishing operation being pointless.” Quite the opposite: the rate of reaching the fixed ceiling is actually moved by the people on the floor and by ingenuity in operation. The whack-a-mole feeling, where troubles keep welling up: go to hold down power bills and marketable yield drops, cut people and somewhere else you run short of hands, that feeling is certainly real. But it does not mean “the total is fixed, so nothing you do matters.” It is true that dealing with it only at the symptom layer tends to look like reshuffling, but once you take the upstream-set ceiling as given and then put operation in order, the rate of reaching it actually rises. When it feels like knocking down one makes another pop up, usually you are trying to claw it back through operation alone without taking the upstream constraints into account. So before you swing, go take a look at “which upstream choice called up this symptom.” And then close the gap through operation. It is a problem of order.

If marketable yield is hard no matter what, separate the part that is the consequence of having chosen that crop and that scale from the part you can recover through the rate of reaching it on the floor. If power bills stay heavy no matter how you cut, separate the part that is the consequence of having chosen a shelf where you get compared on price from the part you can close through operation. Trying to claw back through operation alone something whose cause lies in a lower layer wears you out; but stopping your hand by blaming “the upstream” even for something that genuinely has room in the operation layer is a waste too. Look at which hole the nest is under before you put your hand in. That, in the end, is faster. What to close and how in the operation layer holds a whole book’s worth of ingenuity. What we have gathered as the 172 hints for raising a vertical farm’s profitability is exactly a collection of concrete measures for raising this rate of reaching it.

Pull back to the whole industry and this view holds up too. In the latest field survey of Japan’s large-scale protected cultivation and vertical farms (FY2025), over 60 percent of operators overall were in the black or at break-even. But this splits sharply by type. Greenhouse and hybrid types are both over 70 percent in the black or breaking even (Greenhouse alone is over half in the black), whereas for PFAL about half are still in the red (7). Even with the technology and the capital in, PFAL is still at this level. The picture comes out just as it is: the height of the ceiling differs by type, and within the same type, black and red split according to the rate at which they reach it. Further, a meta-analysis that gathered thirty years of research to look at vertical-farm energy consumption points out that, year over year, there is no sign of a noticeable industry-average improvement in energy efficiency, no so-called learning curve (6). But the paper itself lists effective moves that genuinely exist, such as the case of cutting consumption sharply through root-zone cooling. It is just that they are not widespread, not that they are structurally impossible. In other words, it is a story of “effective moves exist but have not reached everyone, the adoption barrier,” and it looks consistent with the outer macro figures too. By the way, how to read the industry’s share of loss-making operators like this is itself a theme prone to much misunderstanding.

Scale amplifies the black but does not rescue the red

So far I have looked at the view of “going back upstream,” but this was not a story of improvement in the operation layer being pointless; it was a story of order: the upstream decides the ceiling, operation decides the rate of reaching it. On top of that, what you can do differs between someone who has already built and someone about to build. Especially for someone about to enter, the most upstream “building and crop” is a layer that, once decided, cannot be undone on the floor. If so, that is not a place to decide on a whim; there are things you should confirm before you build.

What often comes up here is the expectation that “make it bigger and the unit cost falls and you go into the black.” Even though it is a large judgment, hard to undo later, it is also a place people tend to think about too casually. “Make it bigger and the unit cost falls” is half true. But it is better to look a little more carefully at which unit cost falls.

What works in your favor as you scale up is usually the “unit cost of building the building and equipment.” Once you have built it all in one go, the more volume rides on top of it, the more the weight of equipment each plant carries gets thinned out. Here, certainly, bigger works in your favor. The daily cost, on the other hand, is less affected by scale than the construction cost is. Both power bills and labor costs rise in step with the volume you add. Of course, it is not that operation has zero scale effect: there is room to develop multi-skilled workers who can each watch a wide area, or to negotiate your contracted power favorably. But it does not fall the way the unit cost of construction thins out. Lump these together and think “make it bigger and everything falls,” and your expectation gets too rosy.

So what happens? If the upstream combination of sales channel and crop stays weak, that weakness too grows in proportion to scale. In an arena where you run a thin loss per plant, just increase the number of plants and the total loss swells right along with it. Scale is something that works precisely because the arena is already reaching the black; it is no magic that flips the red into the black. It is a device that amplifies a good arena, but not a device that rescues a bad one. That is the point.

This matches the actual survey data too. In the latest field survey (FY2025), looking at PFAL by cultivation area, the share in the black or at break-even for 1,000 square meters and above is 50 percent, actually lower than the 65 percent for under 1,000 square meters (7). Bigger is not necessarily closer to the black. The measured data backs up the view that just making it bigger does not turn the red into the black.

This split of “what falls is the unit cost of building it” matches estimates using facility data too. Economies of scale work clearly on a vertical farm’s construction cost, and one estimate puts it that scaling up a hundredfold lowers the construction cost per unit by more than half (roughly the mid-50-percent range) (5). But what the same study holds down is this: this scale effect is strictly a “cost of building” story, and there is no confirmation that the same economies of scale reach daily operating costs like utilities. So it is not that “make it bigger and everything gets cheaper”; what thins out is the upfront-investment side, and there is no confirmation that the same economies of scale reach the daily cost. This view has a proper source.

So what someone considering this from here should first confirm is the part before you get into the scale story. Before you expand, is that combination of crop and sales channel properly reaching the black at a small scale? If you have the feel of the black while small, scale will grow it. But trying to make something that does not work small work by making it big is, usually, the order reversed. First, is it reaching the black while small? Confirm that, and then talk about scale. This is the order for not getting the most irreversible judgment wrong. The challenges of a vertical farm are not a list of symptoms to knock down: the upstream decides the ceiling, and the people on the floor decide the rate at which you reach it. Before you build, you sharpen the upstream; after you build, you take everything you can through operation. They can be read back as a single thread of order.