Vertical Farm Basics and Overview

You Cannot Protect a Vertical Farm's Safety with "Pesticide-Free"

A vertical farm’s business plan almost always says “closed environment, therefore safe.” But that one line is not what the buyer at your shipment destination looks at in a QA audit. When a microbe shows up in the circulating nutrient solution, how far back through the lots do you trace the recall? When the HVAC system goes down and a whole zone is ruined, how do the contractual liability and the insurance move? What the audit checklist pushes in your face is not “is it pesticide-free” but “is this designed so you can stop the bleeding when something happens.” If your own verification items still stop at “are residual pesticides within limits,” this article is the conversation to have before that audit.

The Three Layers Bundled Inside the Word “Safety”

A vertical farm is talked about as “safe because it is closed.” But being closed also means that when something happens inside, everything is connected through the water and the HVAC. If a microbe appears on one rack, it can ride the water channel and carry over to the next line. Hasn’t this snagged you, somewhere entirely separate from the question of whether it is pesticide-free? Seen that way, safety is not a matter of “is the growing clean” but of “when it spreads, how much does it drag in.” That said, this is not a deterministic claim of “closed = inevitably wiped out.” My own view is rather the opposite: a hydroponic system that circulates soundly at sufficient flow and is properly managed carries a low risk of pathogens spreading in the first place. The point is that the unit you watch for safety becomes not “one plant” but “the range connected by the water channel,” and the questions become how you partition that range and how you keep the circulation sound. I will get into it in detail later, but I stand on the side of getting it right through sound management rather than holding things down with sterilization equipment. In fact, when I talk with shipment destinations, what they care about is not the presence or absence of pesticides but the arrangements for recalls, insurance, and contracts when something happens. These ought to be separate matters, yet in practice they all get lumped together under the single word “safety.” One more thing: what I have in mind here is the hydroponic type that circulates nutrient solution, a vertical farm centered on leafy greens. With fruiting vegetables or a greenhouse, how microbes spread and how the countermeasures work both change.

The label question of whether something is pesticide-free, and this nagging point, are separate layers. The label only guarantees “what was not put in at the entrance”; it says nothing about what can happen inside the facility. If anything, an environment that has been closed and made uniform can also become an easy place to move around for a microbe once it gets in. There is no soil, no natural predators, the temperature and humidity are constant, and the water circulates. Being clean and being a place where microbes spread easily can hold true at the same time. Clean does not necessarily mean it won’t multiply. So the unit of safety becomes not “one plant” or “one rack” but “the range connected by the water channel.” But this is a matter of “the routes being connected” — and precisely because they are connected, whether you can keep that water channel in a sound state becomes the dividing line. The cleanliness of the growing is a matter of points, but the risk moves along a line — and whether you can keep the flow and circulation steady so that line doesn’t run wild is where the management battle is decided. The reason shipment destinations care about recalls, insurance, and contracts rather than pesticides is exactly that they are looking at “when it spreads, how much does it drag in” in terms of money and liability. For them, safety is not cultivation quality but the arrangement of who bears how much after an incident occurs. To organize it: “pesticide-free” is a product label, that is, a guarantee about a point. A vertical farm’s safety, on the other hand, is the bundle of business risks made of the routes by which a microbe can spread through the water channel, the operation that holds it down through sound circulation and management, and contract, insurance, and recall all moving at once when it does spread. The same word “safety” is really bundling three things that belong to different layers.

And this paradox is already embedded from the very starting point of the design. Soilless cultivation — the hydroponic type grown in the nutrient solution mentioned earlier — was developed precisely to avoid soil-borne diseases, yet circulating the nutrient solution itself creates a water environment convenient for pathogens that release zoospores, such as Pythium and Phytophthora; that is how it has been laid out. The same review, however, also notes alongside this that such pathogens can usually be held down with management (treatment including disinfection) (ref: 1). The method chosen to avoid soil-borne disease newly produces a disease route specific to hydroponics. Yet that route is described as something that can be held down depending on the treatment. My own view is that, rather than tilting that treatment toward strong sterilization, the main line is to secure flow, circulate soundly, and manage the water systems by keeping them separated. Lean too far toward sterilization and, as I will touch on in a later section, the side effects show up on the nutrient side of the nutrient solution. What rides the water channel and spreads easily here is, first of all, the plant pathogens that ruin the crop itself — in other words, the business risk of no longer being able to ship. So what about food safety for the consumer? There are even ways in which it is more favorable than open-field growing. There is measured data showing that the bacterial count on lettuce leaves grown in a laboratory was 10 to 100 times lower than that of open-field leaves (ref: 2). But that cleanliness does not mean “microbes won’t spread.” Even with a small absolute amount of microbes, the problem remains that once it gets in somewhere, the circulating water can become the route, and the business risks of recalls and contracts do not become zero. Reading the clean indoor numbers with the same sense you’d use for open-field growing is itself already a stretch.

Microbes Move Along a Line; Equipment Fails Across a Plane

Power outage. HVAC shutdown. In a facility that has been closed and made uniform, these mechanical troubles too feel like they hit everything all at once. This “moving along a line” story is perhaps not limited to microbes. Microbes are a story of spreading routes, but is that kind of equipment trouble just a different version of the same “closed, therefore all in the same boat”?

They are kin, yes. But it is better to keep them as separate layers. The reason is that the origin point of how they spread is different. A microbe arises inside and travels along the water channel — it draws a line, so to speak, from within. A power outage or HVAC shutdown drops the whole surface at once, from outside. It does not drag things in one by one along a route; the common foundation gives way and everything falls at the same time. So if microbes are a “line,” equipment is a “plane.” The only thing they share is that, being closed and uniform, the damage comes out all aligned — but how they move is a different beast. This distinction matters because the countermeasures that work point in opposite directions. For microbes you defend by cutting the route, that is, by splitting lines, splitting water systems, and keeping flow and circulation sound. For equipment, conversely, you defend by adding redundancy, having backup generators, and doubling up the systems. Even with the same word “split,” microbes you split to stop the spread, and equipment you split so that the rest stays alive even if one fails. Mix them up, and the countermeasure for one becomes powerless against the other.

Looking a bit more closely at the microbe side’s “moving along a line,” how much it multiplies changes quite a lot with temperature. In an experiment that watched a certain opportunistic bacterium in growth medium and basil juice, at 4°C it barely multiplied over six days, but at 20°C it multiplied by more than five orders of magnitude (100,000-fold) within three days (ref: 3). For viruses, the direction is rather the opposite. The number of days for a certain virus in nutrient solution to drop to one-tenth (to be inactivated) falls from as long as 48 days at 15°C to about 7 days at 37°C (ref: 5). For both bacteria and viruses (a surrogate strain), the “safety line” moves with the same variable of temperature. But bacteria multiply at high temperatures while viruses disappear faster at high temperatures, so the direction of the movement is opposite. That is exactly why you cannot say “what temperature is safe” without reference to which microbe you are dealing with. And as for where the contamination comes in from, one study reports that in hydroponic spinach, water rather than the growth medium is the main contamination route (ref: 4). If so, it becomes clear that the line to cut is, first, the water.

Reading the Growing Side by Cause, Binding the Shipment Side by Result

Microbes arise inside and travel along the water channel, a “line”; equipment trouble drops the whole surface at once from outside, a “plane” — so far I have cut the world by cause. So, are the contracts with shipment destinations written with an eye to even that difference in cause? In most cases they don’t spell out the cause in detail. What they write is the “result.” What happens if the delivery doesn’t come in. What happens if off-spec product appears. Who bears the cost if a recall becomes necessary. Whether you are wiped out by a power outage or part of it is ruined by a microbe, from the shipment destination’s view it is the same result: “the goods we were promised didn’t arrive.” So they bind you by result, without separating the layers of cause.

Here you notice that, depending on your position, the way you see the world is reversed. The growing side sees the world in the layer of cause and needs to separate microbe from equipment. The shipment side, on the other hand, sees in the layer of result and deliberately does not separate. The single word “safety” bundles three things that belong to different layers, but where you cut that bundle is reversed depending on your position. The growing side cuts by cause; the shipment side cuts by result. This is the point where contract negotiations fail to mesh.

And after an incident occurs, that “result” splits further into three kinds of damage and moves. The first is the direct loss that goes out of your pocket on the spot through recalls, disposal, and redoing inspections. The second is the loss from transactions being halted or from being dropped, and this drags on longer than a single incident. The third is the reputational harm of “that place had an incident.” The first two connect directly to money and contracts in the layer of result and are easy to estimate. The troublesome one is the third, which works on the layer of attitude I’ll talk about next, and is written nowhere in the contract as a figure. Contract, insurance, and recall all moving at once after an incident occurs — this is what it means to see safety in the “layer of result.”

Consumer Anxiety and Operator Risk Are on Different Axes

The third of these damages, reputational harm, is connected to another gap that lies between you and the consumer. Haven’t you ever felt that vegetables from a vertical farm seem “somehow not quite safe”? At that moment, the content of your own anxiety — whether it is pesticides, microbes, or a discomfort with “artificial” technology — is often not clear even to the person feeling it. Meanwhile, the risk an operator actually shoulders — the spread of microbes via the water channel, or the conditions under which microbes survive depending on temperature — does not correspond to the content of what the consumer is anxious about.

Is this gap something to go and fill in? In the range of my talking with shipment destinations and visitors on site, my honest sense is that the moment you try to fill it in, it more often gets tangled instead. The consumer’s anxiety is the feel of “it’s artificial, so it’s dangerous,” and this is a matter of attitude, not a matter of content. If you respond to that with the content of the technology — “we keep the water systems separated,” “we cut the microbe routes” — it gets received as “do you have to manage it that much for it to be dangerous?”, and there have been times when it made the anxiety bigger instead. Because you are answering the layer of anxiety with a technical answer, it not only fails to mesh but looks as if it got worse. At least in my own hands, there have been several such scenes.

So rather than filling it in, it is better to separate the way you speak from the start. But if you take “separate” to mean the double-talk of “soothing lies for consumers, real risk for operators,” that too is a cheap move. Instead, what you should return to consumers is the guarantee of “who is watching over this with responsibility.” Not the content of the routes or the microbe conditions, but whether this is someone you can trust. What the consumer really wants to confirm may be, more than the fine details of microbes, whether “a proper person is doing this.” So rather than putting the transparency of the technology up front, show where the responsibility lies. Even so, to the content you are asked about — microbe countermeasures, inspection results — disclose honestly. It means not hiding the content, but not pushing the content you weren’t asked about. That said, important safety information the consumer has no way of even raising as a question — for instance, the transmission risk inherent to a closed environment and how you hold it down through sound circulation management — you disclose actively even when you aren’t asked. I draw the line on this from the start as an exception to “don’t push.”

So, does it only click with someone who works in the same layer? This is half true and half not. With someone in the same layer, it is certainly easier. With shipment destinations you can talk in the layer of result, and with on-site engineers in the layer of cause. But the truly important work lies more in the translation that crosses layers. How do you map the consumer’s “somehow anxious” onto the operator’s “conditions to inspect”? How do you bring down the shipment destination’s “binding by result” into the on-site “countermeasures for cause”? Who takes on the role of translator between layer and layer? What secures safety itself is the design of sound circulation management, records, and redundancy, but whether you can reduce the discrepancies that form between layer and layer — the failure of contract negotiations to mesh, or the worsening of consumer anxiety — changes greatly depending on whether this translator is present.

This “gap between what the consumer fears and what the operator should inspect” can be backed up in the form of surveys, too. About vegetables grown in cities, a survey in Bologna found that over 60 percent of people worry about contamination from the air or soil, while a report that measured heavy metals at urban farms (soil-based) in San Francisco found those values to be below the FAO and WHO standards (ref: 6, 7). If even soil-based urban farms come in below the standards, then at least in facilities that use managed growth media or water sources, the content people worry about and the reality once measured do not necessarily match. And what the consumer vaguely pictures is contamination originating from soil or air, while what the vertical farm operator actually shoulders is microbes via the water channel — a different axis altogether.

Build Inspection Items in the Three Tiers of Cause, Result, and Attitude

Whether you are drawing up a new business plan or preparing an existing facility for a quality-assurance audit, the first place you get stuck is the same. As inspection items, what do you list, and how far? You can think of areas like hygiene, microbes, traceability, and contractual guarantees, but you have no idea in what order or at what granularity to put them down on the list. That is the kind of moment.

If you line up the inspection items by area from the start — “hygiene,” “microbes,” “contracts” — you usually end up with a thick list for audits, but it doesn’t get used on site. It’s because three layers of different natures — cause, result, attitude — are lined up all mixed together. So before dividing by area, you first sort once by these three layers.

Concretely, you lay them down in three tiers. The first tier is the layer of cause, that is, the items the site itself knocks out. For microbes: the separation of water systems, records of cleaning and disinfection, and logs of temperature deviations; for equipment: the holding time during a power outage and inspection of the redundant systems. This is exactly the story of the countermeasures pointing in opposite directions for “route” and “plane,” and you write the items separately too. The second tier is the layer of result, the items you owe to the shipment destination: missed deliveries, off-spec product, the trigger conditions and cost-bearing for recalls, and traceability goes here. Given that contamination tends to come in first from the water, if you keep it traceable which water system connects to which lot, you can narrow down the recall scope when it counts. You decide the granularity with the thinking of: at what unit do you cut the lots so that, when an incident occurs, you can make the recall scope smallest. The third tier is the layer of attitude, where the responsibility shown to consumers lies. This is not a technical item but whether you can state “who is guaranteeing this” in a single line. There is just one knack for granularity: “when an incident occurs, can you trace that item and arrive at the cause or the responsible party?” An item you can’t trace back to a cause or a responsible party is just decoration, so you drop it.

Differences in standard by shipment destination also become clearer when you organize them in these three tiers. First, the minimum floor common to any shipment destination is the records of the cause layer (cleaning and disinfection, temperature, water systems) and the lot tracking of the result layer — in other words, the state of “if something happens, you can trace it.” On top of that, the add-ons change by shipment destination. In the range of my talking with shipment destinations, mass retail tends to weight the result layer, often pressing hard for traceability and recall arrangements. The granularity of lot management gets questioned directly. Food service was more often a counterpart that focused on freshness and stable specs, that is, on the quality variation within the cause layer. Institutional catering went a tier heavier still: because the eaters include children and the elderly, I have been pushed on two fronts — stricter scrutiny of the microbe items, and, in the result layer, immediate recall and a contact tree in the event of an incident. This is only a matter of how my own counterparts pressed me, and it isn’t something to generalize as an industry-wide tendency. Even with the same facility, when the shipment destination changes, the “layer you thicken” changes. Rather than rebuilding the list, you switch which of the three tiers you dig into by shipment destination. Build it by layer from the start, and this sorting-by-destination works.

So, should you just disinfect the nutrient solution no matter what? Let me write my position clearly here. I think, rather, that you should not lean entirely on sterilization equipment, and should ensure sound circulation and management with sufficient flow secured. There are two reasons. One is that disinfection is not as all-purpose as you’d think. In a study that lined up and compared chlorine, chlorine dioxide, and ultraviolet against plant pathogens in irrigation water, even with the same treatment the concentration and exposure time needed to kill 99 percent or more differed greatly by type of pathogen, so the premise that one standard treatment works the same way on all of them does not hold (ref: 8). The other is that disinfection’s side effects land on the nutrient solution itself. Ozone microbubbles, while effective against pathogens, have an inherent quirk of also lowering the manganese and iron in the nutrient solution along with it (ref: 9). Strong sterilization like ultraviolet, heating, or ozone can knock out beneficial microbes along with the pathogens, as one review points out (a qualitative observation, granted) (ref: 1). You can reduce the targeted microbe, but you whittle down the “growing side” — the nutrient in the nutrient solution and the balance of microbes. So I stand on the side of circulating soundly with flow maintained and managing the water systems by keeping them separated, rather than holding things down with equipment. There are still scenes where you use disinfection alongside it, but in those cases you don’t lump it as “we disinfected” — you leave it as items: what, in what amount, applied for how long. As a move in the opposite direction, there is also an experiment where using a beneficial microbe like Trichoderma held down Fusarium root rot in lettuce while keeping yield, and this is material you can reframe not as a pesticide-free selling point but as one of the verification items for “what conditions, designed how, lower the hygiene risk” (ref: 10). The direction of killing and the direction of growing are better thought about as separate builds.

The Range of Self-Help and Where to Draw the Line to Hand to Experts

Up to here, I have talked about the “way of seeing” for operators to build their own inspection items. Let me draw one line here. What you can build yourself goes this far. When it comes to actually fixing the microbial management threshold values, obtaining HACCP certification, or nailing down the design of insurance and the wording of contract clauses for when a food incident occurs, this is not a domain to complete in your own style; it is where you should team up with food-hygiene experts and insurance and legal practitioners. What I can hand you here is the framework of “what you recast as the things to inspect,” up to that point; the threshold values for each item and the legal nailing-down are the experts’ domain.

Don’t build your plan on the premise that a food incident won’t occur. This stance of “an incident can occur” is not a consolation; it fits the reality. For instance, microgreens, while they have not been tied to a foodborne outbreak so far, have been the subject of seven voluntary recalls in recent years alone, so incidents and recalls are not a “what if” — they are actually happening (ref: 11). Moreover, young plants like microgreens, with underdeveloped protective tissue, let pathogens get inside more easily than mature vegetables, and once you exceed the 18–25°C suited to growth, microbial multiplication is, if anything, promoted (ref: 12). That inspection tends to get a tier heavier in scenes where the eaters include children and the elderly, as with institutional catering, is a line you can nod to once you look at this plant-side vulnerability together with the reality that recalls are actually happening.

That nagging “was safety a simple matter, after all” you felt at the start probably won’t lessen. If anything, the content of “it isn’t simple” just becomes clear. What started as a vague “there is something separate from pesticide-free” takes shape, divided into three — the layer of cause, the layer of result, the layer of attitude — and further, to the point where the counterpart you watch and the direction of the countermeasures differ by layer. The next time you open a business plan or a risk-inspection sheet, safety should come into view not as one more selling word, but as a row of items where you confirm, area by area, how tightly the verification has been nailed down. And what remains most of all is the question of who translates between layer and layer. From “what is it that I’m feeling uneasy about” to “what was bothering me was that I couldn’t see who this translator is.” The question moves up a notch. Even that should be enough to pull you out of the vague unease that surfaces whenever the topic comes up.