Economics and Profitability
Vertical farm profitability — judging it by a single year leads you astray
Many of you, I expect, are weighing up a vertical farm project right now, staring at the plan and the P&L you have in hand. And in most cases, you try to decide “at this profitability we go ahead / we hold off” from the profit and loss for the first year or the first few years. But having been involved in a number of startups on the ground, what I have come to think is that a vertical farm business has a shape in time — it carries losses in the startup phase, then the operation matures from there, and only then does it turn the corner toward payback. A single figure called profitability does not tell you when that turn comes. With the very same plan, depending on which one changes — the crop, the startup speed, or the contract length — the answer comes out completely different.
The year you turn a profit and the year you fully recover your investment are different things
About vertical farms, you often see it put as “how many years to profitability.” But when you look at a few cases, something starts to nag at you. Take a vertical farm written up as “we turned a single-year profit in year three” — over the one or two years before that, it must have been piling up considerable losses, and recovering that comes much later. And yet only “profitable in three years” takes on a life of its own. Doesn’t that nag at you too?
What is interesting, too, is this: even at a vertical farm where the year it turns a single-year profit comes fairly early, if the startup losses are large, the point where it turns the corner toward payback on a cumulative basis comes much later. Conversely, even if turning a profit is on the late side, if the early losses are shallow, the valley that the cumulative losses trace is shallow too. Thought of that way, what you should be looking at is not “in what year is it profitable,” but “when do the cumulative losses hit bottom and start coming back” — that, it begins to seem.
This is a sound way of looking at it. They are easy to confuse, but these two are separate matters. “Turning a single-year profit” is whether that year’s balance is positive — that is, a matter of flow. “When the cumulative turns the corner,” on the other hand, is how much of the hole you have dug so far you have filled back in — a matter of stock. When you judge whether an investment is viable, the one that really counts is the latter, and it is also the latter that directly answers the question of whether you can fully recover the money that went out at the start.
These two move independently. If the startup losses are deep, then even with a profit from the next year on, it takes years for that profit to fill back in the deep valley. So if you pull out the single point “profitable in three years,” the information about the depth of the valley drops out entirely. Even with the same “profitable in three years,” depending on the weight of the initial investment and what kind of losses piled up during the startup phase, the view of payback turns out to be a completely different thing.
A vertical farm is a type of business where this bites especially hard. Even within what I have seen on the ground, the initial capital investment is heavy, and the losses from low packout yield and from utility bills until operation stabilizes pile up easily. So if you look only at the single-year break-even, you misread the economics far too optimistically. What you should be looking at is when the curve traced by the cumulative cash flow hits bottom and turns back upward. That turning point, and the depth of the valley. Only when you look at these two as a set can you say whether the money you put into that vertical farm comes back.
This view that “a vertical farm is a type of business where it bites especially hard” is not just a matter of gut feeling on the ground. Multiple studies — differing in region and in research method, from case studies to economic models to reviews — all point the same way on one point: that these two things, the weight of construction and initial capital and the continuing operating cost (mainly electricity), are the biggest walls to adoption and to establishing profitability (see: 1, 2). What is backed up is this structural part, “these two are the walls”; the way of splitting things into valley, width, and climb that comes up later is just a restatement of that, put so it is easier to handle on the ground. Either way, the reason the valley tends to get deep at startup is not that the vertical farm does things badly, but that the type of business is structurally that way — that is what it amounts to.
The valley up to payback is set by its depth and the time spent at the bottom
The depth of the cumulative valley varies quite a lot from one vertical farm to another. Looking over vertical farm cases makes that plain. So what, in the end, decides “how deep the valley can be dug”? Is it still the weight of the initial capital investment that matters most, or is it how long the startup phase drags out the losses — that is, the width of the valley? At the same depth, a valley stretched out sideways looks like it would make recovery tougher — haven’t you had that feeling? Let me lay out the difference in how these bite, here.
The depth of the cumulative valley is set by two things, the initial investment and the startup losses, but they are completely different in character. The initial investment is a fixed depth dug once before the vertical farm runs; the period over which losses drag on is the length of time spent at that bottom. A vertical farm’s initial costs pile up heavily across many fronts — not just the building, but LED lighting, HVAC, and hydroponic equipment. And that weight keeps biting even after operation starts, as depreciation. In a survey of PFAL operators too, depreciation is listed alongside labor and electricity as one of the major cost items that weigh heavily on overall cost (see: 9). It is the kind of thing where you keep paying off the depth you dug, a little at a time, over many years.
At the same depth, a valley stretched out sideways makes recovery tougher. There are two reasons. First, simply, the more years you spend at the bottom, the more losses get added on, and the valley itself gets deeper still. Stretching out is not unrelated to depth; it bites in the direction of digging the depth deeper. The other is the climb after you turn the corner. A vertical farm whose startup is dragging on is still in a state where packout yield and operation have not fully stabilized during that time, so the very speed of filling back in is hard to raise. Dug deep, and weak at filling it back in. It tends to become this double disadvantage.
So rather than lining up “depth” and “width” as separate factors, it fits the feel on the ground better to grasp it as: width turns into depth. The initial investment is a given that is largely fixed at the blueprint stage, but the length of the startup phase still has room to move through operation. That is exactly why, if you want to keep the valley shallow, how short you can cut the time spent sitting at the bottom matters as much as keeping an eye on the weight of the initial investment. What sways the view of payback is, surprisingly, this “when can you start climbing up from the bottom” side.
You want to think “if we just make the scale bigger, the valley will get shallower too,” but here the numbers come out in a peculiar way. There are economies of scale in PFAL construction cost, and there is an estimate that increasing scale a hundredfold lowers construction cost per unit by roughly 55% (see: 3). This fits the feel on the ground too. Even when you make the facility bigger, work space and the like do not grow at the same rate as growing space, so the bigger the vertical farm, the easier it is to raise the share of area usable for cultivation, and the fixed cost riding on each plant spreads thinner.
But what thins out is the fixed-cost side, and operating cost is another matter. The study just mentioned, in order to isolate and measure the economies of scale in construction cost, treats the operating cost (utilities and so on) as not spreading thinner with scale (see: 3). In fact, utility costs are large even among running costs, and there is a survey showing that for PFAL, electricity alone accounts for roughly a quarter (24%) of total cost (see: 9). Even if construction cost and the per-plant fixed cost can be spread thinner by scale, a variable cost like utilities, which keeps going out the whole time you sit at the bottom, barely falls with scale. So scale can make the depth at the mouth of the valley somewhat shallower, but it barely touches the spending for the time spent sitting at the bottom — the “width” part just discussed. This asymmetry exists.
Whether you can turn the corner toward payback splits by the crop
Up to here, on the premise that you “turn the corner eventually,” we have looked at whether that timing is early or late. But looking broadly across cases, even at the same vertical farm, depending on what you are growing, whether the cumulative even turns the corner toward payback at all changes. Between something with a high unit price and fast crop cycles, like leafy greens or herbs, and a crop with a much lower unit price, there is a qualitative difference in how the valley comes back.
This is a matter of the slope of the climb. The depth and width of the valley we have looked at so far represented “how far you dug” and “how long you stayed at the bottom,” but how steep a gradient you can climb after turning the corner is set by how much gross profit that crop can carry per crop cycle, and how many times a year you can repeat that. Leafy greens and herbs, with a high unit price and fast crop cycles, have a large gross profit times the number of crop cycles. So the slope of the climb gets steep.
A vertical farm carries, to begin with, a structure that is higher-cost than open-field growing. With a low-unit-price crop, the gross profit carried per crop cycle is thin, and most of it gets eaten by operating cost before it can be turned to filling back in the valley. Then the slope of the climb lies flat. But here the stages split by the type of crop. Low-unit-price fruit vegetables — tomatoes, for instance — are the domain where it does turn the corner, but the climb lies terribly flat and payback is greatly delayed. Grains are further still beyond that: far from the slope lying flat, they cannot even sustain a climb steep enough to turn the corner toward payback in the first place. Up to here it was a matter of “when do you turn the corner,” early or late; here, depending on the crop, the very question of “can you turn the corner at all” — the possibility of payback itself — is at stake. That is where the qualitative difference shows.
So crop choice is, on a separate axis from the initial-investment story that sets the depth of the valley, the variable that sets the slope of the climb. That vertical farm cases lean toward leafy greens and herbs is not a matter of taste. It is more accurate to see it as a narrowing-down from the economic side: under this high-cost structure, the crops that can sustain a climb steep enough to turn the cumulative firmly toward payback are, for now, in practice limited to that band.
This stage difference shows clearly in the numbers. First, the low-unit-price fruit vegetables. In a trial calculation for a small-scale vertical farm with small wind turbines attached, basil and lettuce can exceed a 100% internal rate of return, while tomatoes stay at 2.5–11.3% in all cases (see: 4). It is positive, so it does turn the corner, but just swapping the crop in the same facility changes the speed of payback by an order of magnitude. Note, though, that this [4] is a small-scale trial calculation for a special configuration with small wind turbines attached, so the absolute value of the internal rate of return itself cannot be applied as-is to general cases. What you should read off is the way it bites — “the crop changes it by an order of magnitude.” And further beyond that are the grains. For indoor wheat, the revenue obtained stays at about one forty-sixth of the cost spent, and there is a trial calculation that even with advancing technology the theoretical limit is around one-sixth. For soybeans, it is 1,000 to 2,000 times the energy of open-field growing (see: 5, 6). This is no longer “the climb lying flat” — it is a domain that cannot sustain a climb steep enough to turn the corner toward payback in the first place.
Don’t take the finished numbers — draw the cumulative curve yourself
You have a single project or business plan in hand, and you want to size up its economics. At a time like that, where, concretely, should you start? It tends to happen that you start to judge by looking at the profitability figure that came out, or the one line of “how many years to profitability.” On the basis of what we have covered so far, let me lay out what to redraw first and what to set yourself — that first move.
Stop looking at the numbers, and draw a single curve yourself — start there. Concretely, it goes in this order.
First, what to redraw. The finished numbers that came out — “profitability X%,” “X years to profitability” — set them aside for now. Instead, what you should pick up from that plan is the raw materials. The total initial investment, how much loss you assume in each year of the startup phase, and the annual cash flow after stable operation. Pull out just these three. Profitability is a conclusion, not an input, so don’t work backward from the conclusion.
That said, this does not mean throw away the finished profitability. The “profitability X%” you see out there is, in almost all cases, a single point’s number from after it has settled into steady-state operation. Apply that as-is to your own project, which includes the startup phase, and the valley drops out entirely. Use the benchmark only as a reference value for the slope after you have finished climbing, and set the depth and width of the valley separately, yourself. Not mixing startup-phase and steady-state numbers is what this distinction means.
Next, what to set. Using those three, draw a single broken line of cumulative cash flow yourself. The horizontal axis is years, the vertical axis is the cumulative in-and-out. Dig down once with the initial investment, the bottom extends with the startup losses, and you climb back up with the profit after stabilization. Ten rows in Excel is enough to draw it. With this, the “depth of the valley,” the “width of sitting at the bottom,” and the “slope of the climb” come out at a glance. In what year the turning point comes, and how many years later that is than the plan’s stated “profitability in X years.” If a gap between the two shows up here, you can no longer go back to the one line of numbers.
On top of that, shake the premises you set, yourself. Three in particular. If the startup phase runs a year or two longer than assumed, how deep does the valley get? If the crop’s unit price or turnover drops by a tenth, how flat does the slope of the climb lie? If the initial investment swells by a tenth, how far back does the turning point shift? A plan usually sets these three at a single convenient point. If you move there and the curve falls apart badly, that means the plan’s economics are merely riding on that premise.
If, from here, I had to name one single point to go set yourself, it is the length of the startup phase. The initial investment and the crop are givens largely fixed at the blueprint stage, but the time spent at the bottom is the one where a plan’s optimism most easily slips in, and it is also the pressure point that bites on both the depth of the valley and the delay in climbing. But there is something I have to say honestly here. Much of what causes the startup phase to run long is held by external factors — demand, sales channels, working capital — and only part of that is something you can freely shorten by operational effort alone. Even so, there is room where the operating side bites, and from what I have seen on the ground, it is the retention of people and the accumulation of technique on the ground. The startup phase is also a learning period in which the ground gets a grip on the quirks of new equipment and stabilizes packout yield, and if people keep turning over, this accumulation gets reset each time and the time spent at the bottom drags on and on. Conversely, a vertical farm that can keep this stable starts climbing up from the bottom early. So take “can this vertical farm really start climbing up from the bottom in this number of years,” and set it again not only as a story of equipment, but as a story of the people who run it and the ground. That alone changes the view enough.
One more thing I want to set alongside the startup phase is the contract length of the sales channel. Where startup speed and yield set the depth of the valley and the slope of the climb, this sets how far the climb keeps going. If a several-year sales contract runs out before reaching the steady-state phase and cannot be renewed on the same terms, the climb is cut off there, and the cumulative ends before it turns the corner. So in how many years the contract renews is also, like the length of the startup phase, a variable to set on the curve as your own premise.
That shifting a premise by a tenth makes the curve fall apart has a clear real example. In one model calculation, the minimum scale at which lettuce breaks even leaps up dozens of times from just a 20% drop in price, while conversely for strawberries, just a 20% rise in yield per unit makes the break-even scale orders of magnitude smaller (see: 3). Yield per unit and price, which move during the startup phase, move the scale at which a business breaks even by orders of magnitude. So just how fragile a premise the plan sets at a “single convenient point” is — you can tell at once by shifting it by a tenth yourself and redrawing the curve.
Build the shape of the curve yourself, take the gradations from outside results
Last, let me set down one line. The work of “drawing the curve yourself” we have looked at up to here is an interpretation for grasping the economics anew as a structure with your own hand, without swallowing the numbers that came out. At the same time, the soundness of that curve is, in the end, decided by how well the startup speed and the stable packout-yield value you put into it match reality. This is the part you cannot fill with your own optimism, and grounding only comes once you go to the long-term track record of facilities actually in operation, or to primary information from the equipment side and the operating side. You need to make clear the boundary between where you can build the structure yourself and where you should go and get the outside real data.
The curve itself is structure, not grounding. How the three shapes — the depth, width, and slope of the climb of the valley — combine to decide payback, that logic you can build in hand. Rather, you must build it. It is the work of breaking down the numbers that came out and getting how the variables bite on each other into your own head. This is not something to go and get from outside; it is the part you should do yourself.
But that curve only tells you the shape. How many years startup takes, at what level yield stabilizes, what the cash flow becomes after stabilization — the gradations to put on the line themselves do not come out no matter how hard you stare at the structure. Fill this in with your own gut, and you end up having merely overwritten that earlier “single convenient point” with the shape of a plausible-looking curve. The finer the shape, the more it carries the danger of lending optimism persuasive force instead.
So the boundary is there. Building the relationships of the variables is yours; the values to put into the variables come from outside. The startup speed and the stable packout-yield value in particular are the pressure points where your own wishes most easily slip in, so this is the one place where grounding only comes once you go to the track record of facilities actually in operation, or to the primary numbers the equipment side and the operating side put out. To put it the other way, the purpose of going to get the outside real data, too, is not vaguely “to know the going rate,” but is clear about which single point of the curve you drew you want to back up. Build the structure first, and what you should go get is narrowed down. Drawing the curve yourself and going to get the outside data are not opposed — they are an order. It is precisely because you build the structure first that the primary information you should go get settles on a single point. Only once you have done that can you, at last, speak in your own words about that vertical farm’s economics.
The caution that “fill in the gradations with your own gut and you end up lending optimism persuasive force instead” is a truly important part. In fact, if you gather up individual case studies of hydroponic farms, cheerful figures — internal rates of return of 60–107%, payback periods of under a year — line up almost consistently (see: 7, 8). And yet, looking at the farm sector as a whole, farms turning a profit are rather a minority. In fact, in Japan over the ten years from 2012, roughly 80% of PFAL have vanished, and many of them are reported to have withdrawn or gone bankrupt because the numbers did not add up (see: 3). Abroad too, many protected cultivation operators are pointed out as not reaching the break-even line (see: 6). In other words, the single cases that surface carry a bias in which the ones that went well tend to get selected and reported. So when you go to the outside numbers too, if you paste the high profit rate of a single success case onto your own curve as-is, rather than looking at where it sits within the whole farm population as a set, you end up merely endorsing optimism after all.
Both the depth of the valley and the slope of the climb, you can build with your own hand. But the moment that curve touches reality is the moment you put in just one point of outside primary information. And what actually moves those gradations is, in the end, still the strength of the people on the ground. Even with a vertical farm of the same structure, both how fast it starts climbing up from the bottom and whether it can hold the slope of the climb change with how stably you can run things on the ground. Only when you look at it including that can you stop being jerked around by a single year’s profit or loss, and speak in your own words about when you get back the money you put into that vertical farm — the view of payback.