An interview with John Paul of OrderJunky on the build decisions that cost operators millions.
Most cannabis facility failures are not failures of cultivation. They are failures of procurement, made long before the plant ever touches the room.
John Paul has spent twenty years on the build side of that equation. He walks into cannabis facilities for a living. The first thing he looks at is not the canopy. It’s the entrance.
The lab coat lie: where biosecurity fails first
“All it takes is one bug, one spore to really mess up your whole facility. I’ve been to a lot of really big facilitieswhere all they require is for the person to put a lab coat on, and that’s just not acceptable.”
A lab coat is just a coat. It is not a pressure boundary, or a change-out protocol, or an SOP.
The biosecurity envelope, the part of the building that is supposed to keep one spore from ending a quarter, is the part that gets value-engineered down to a lab coat.
The dehumidification trap: when the spec sheet lies
Walk twenty feet past that lab coat and you are in a room where the next CapEx mistake is already humming on the wall.
Most cultivators have heard the acronym AHAM. It is the certification that says a dehumidifier was tested under specific conditions. It is a reference point, not a promise. John has a name for what shows up when the reference point is not your room.
“We have a saying. It’s not a real acronym, but if it’s not made in the US, it’s most likely made in China. We call that CHAM, China AHAM. Chinese manufacturers are notorious for bolstering their numbers. They’re typically rating those things at a set point in the 90s. For every couple percentages of humidity that we go down, you’re losing a considerable amount of efficiency.”
The manufacturers that publish honest performance curves at cannabis set points are a short list. A handful on the standalone side. A few more if the room is being designed with integrated HVAC from the start. All of them will tell you what the unit actually does at 75 degrees and 60 percent RH, not what it does in a lab at 90.
That is the number that matters once the canopy is in. If the wrong people buy the wrong number, it does not show up as a procurement miss. It shows up later, when the dew points crash and the compressors freeze.
The 1100 watt problem: when operators outrun the engineering
The dehumidifier was rated honestly. The HVAC was sized correctly. The room is in spec for two months. Then somebody in the grow room reads an article about getting certain colors and decides to push the fixture past the spec.
“The fixture might go up to 1100 watts. At a certain point in the engineering process, they decided they were only going to maximum take it up to 900. I call it America because if there is a way to go to 1100, somebody in the grow room is going to figure that out and do it without talking to anybody.”
The fixture is capable of 1100 watts. The engineer ran the room load math at 900. The 200 watt delta is sitting there as a permission slip. The grower will find it.
What happens next is not theoretical. Compressors sized for a 900 watt heat load start freezing up. Dew points crash. The canopy reads it as stress.
The fix is not buying a different fixture. The fix is documenting the engineered operating envelope and putting it in front of the operator before the fixture ships. The 1100 watt problem is a procurement problem dressed up as an operations problem. The procurement layer is where it gets caught.
The smoking deal: why cheap fixtures cost the most
There is a second version of the lighting problem, and this one shows up before the fixture even gets installed.
“Manufacturers are in the business of making money. When they’re giving you a smoking deal on a light, it’s because they engineered around getting you that price. On the LED side, they did an inferior driver, they did a less than adequate QC process on their diode array. Some of the higher-end LED guys are doing voiding, where they’re x-raying every diode array to make sure there’s a connection.”
The savings always come from somewhere. A cheaper driver. A thinner QC process. A diode array that was assembled fast and never x-rayed for bad connections. Each decision is invisible at unboxing. Each will show up later, when failure rates start clustering in a way that is not random.
The high-end manufacturers are doing what John calls voiding. They x-ray every diode array before the fixture ships because the failure mode they care about is the one nobody sees until warranty claims start coming in. That x-ray costs money. It is part of why the smoking deal is not a smoking deal.
The cheaper fixture sits next to the more expensive fixture on the same screen, with the same kind of marketing photo and a similar PPF number. The difference is a manufacturing decision the spec sheet was never going to disclose.
The retrofit that should have been a new build: the $3M to $30M story
Some CapEx mistakes are made one fixture at a time. Others get baked in on the day an operator looks at an existing building and decides it is a head start.
John has a client, one he is careful not to name. The operator bought a property in a quiet part of the country because the price made sense on paper. The operator pitched the project at $3 million. By the time the engineering work was done, the number was closer to $30 million.
“Somehow the project went from him thinking he was going to finish the whole thing for 3 million to closer to 30 million. It would have been cheaper to build a brand new building than to retrofit. We were trying to fit a square peg in a round hole. He was committed to keeping this building because he thought it was going to get us to completion quicker, when it was really holding up the project.”
The operator had thirty acres of usable land. New construction next door would have penciled out cheaper than the retrofit, once the mechanical and structural work was accounted for. The hard part was emotional, not financial. The existing building felt like progress already paid for, and that feeling is one of the most expensive instincts in cannabis CapEx.
The structure could not hold the mechanical load the cultivation plan required. Welding and infrastructure adjustments added up to more than a clean-sheet build would have cost. The schedule slipped.
This is the conversation John wishes more cultivators were having before the slab gets poured. Site selection. Building selection. Realistic numbers on what scale cultivation costs in 2026. A facility that was $300 a square foot in 2021 is closer to $500 a square foot now, and that is if everything goes right.
None of this is a knock on the operator. It is the trap the industry hands every new builder. The existing building always looks like savings, until the engineers start opening up walls.
What a procurement platform actually does: more than a catalog
There is a real gap in procurement here, and OrderJunky is John’s response to it. He is direct about what makes the platform different.
“It’s not just another way to buy products. It’s really the support and the people behind it. A kind of prerequisite to be a pro inside of our ethos is to be slightly damaged and slightly burned by this industry, because those are the people that have been through it. If you haven’t been through a couple market corrections, then you really haven’t seen the industry.”
The “slightly burned” line is doing work. The people John has wired into the platform are engineers, equipment specialists, rebate consultants, attorneys, and CPAs who have watched at least one operation fail in a way that taught them something. The platform is the catalog. The pros are the layer that tells you the dehumidifier you almost bought was rated in the 90s in a lab, and your room runs at 60.
It is also the layer that runs the rebate math, models the financing, and handles the documentation. None of that is sexy. All of it is the difference between a vendor and a partner on a multi-million dollar build.
The gap exists because the engineer, the operator, and the buyer are almost never in the same room when the decision gets made. OrderJunky is built for the moment before the PO becomes permanent.
Start over on the plants: the takeover playbook
Most cultivators in 2026 are not breaking ground. They are walking into a building someone else built badly and trying to make it work. John’s advice for that operator is the most prescriptive part of the interview.
“Start over on the plants. Secure all the genetics that are worth anything. Isolate, get them tested. Make sure they don’t have any hop latent virus or any other viruses. Then, if they’re trying to upgrade anything, talk to us. We can see about maybe even getting them their lights for free through some rebate programs.”
Start with the genetics because everything downstream is built on them. After the plants, the building. Inspect the sensors. Service the dosatrons that have not been touched in years. Verify the lights are still putting out the photon counts they claimed on day one.
Then the rebate conversation. Utility rebate programs in mature markets will cover meaningful portions of a lighting upgrade, sometimes the entire fixture cost, if the operator can document the existing and proposed load. Most operators leave that money on the table because nobody told them to ask for it.
The upgrades only work if the foundation underneath them is clean. Cleaning the foundation is the work most operators skip, because it is invisible until it is the only thing that matters.
The audit trail nobody budgets for: paperwork as insurance
There is one more line item that does not show up on a CapEx pro forma until it is the thing standing between an operator and a license renewal. The documentation trail.
“We built some modules for OEM people to be able to track their product back to the manufacturing QC process. It does cover that. Especially in the lighting world, a 4 percent failure rate is pretty common, and being able to track that and have backups for your clients is a good insurance policy.”
Four percent. On a typical flower facility running 2,000 fixtures, that is 80 lights that will fail inside the warranty window. Manufacturers typically require a serial number, purchase date, lot number, and proof the array is dead before they will replace it. Operators who have that paperwork get the replacement at no cost. Operators who do not pay retail and absorb the downtime.
The same documentation trail is what state inspectors want during an audit. It is what investors want during due diligence on a sale.
A procurement platform that produces this trail as a byproduct of the purchase is doing two jobs. It is selling the operator the fixture. It is also giving them the receipt that, three years from now, makes the warranty claim winnable and the audit survivable.
The CapEx budget rarely has a line for this. The operators who think about it before they need it are the ones who keep their licenses.
Plants don’t talk back
One of the things John said in the interview was about cultivators, not about CapEx.
“Growers are really good plant people, but they might not be good people. Plants don’t talk back.”
His point was about staffing. Many seasoned growers are excellent with a canopy and uneven with everything else. They have spent careers in a room where nothing argues with them, and that one-way relationship shapes how they show up with the people around them. John’s advice was for those growers to be self aware about it, build in an HR buffer if they need one, and have the harder conversation with investors before the personnel problem becomes a yield problem.
The line lands harder when you widen the lens. Plants are not the only thing in a cannabis facility that absorbs whatever you give it. The dehumidifier does not push back at 60 percent humidity. The fixture does not push back when somebody cranks it to 1100. And the people who could have pushed back are not in the room anymore.
Every CapEx decision compounds. A dehumidifier spec’d for the wrong operating point later becomes a microbial issue. The 1100 watt fixture turns into a fixture failure no one logged, then a warranty claim with no paperwork, then a line item the next operator inherits when the facility changes hands.
That is the lab coat lie all over again. The weak point was visible at the entrance, but the damage showed up later, when the people who could have spoken up were either gone or never asked.
Build the room for that operator. Hire and listen to the people who will work in it.
