I used to specialize in Agtech as a consultant. It's a weird space.
[Disclaimer, this is my experience of the Australia and NZ scene, not much Silicon Valley involved]
The gap between "people who know ag" and "people who know tech" is just vast. And I sensed no great urgency to bridge the gap. I tried myself - talked to every farmer I could, moved to a working farm, even had a go at weighing cattle and applying treatment [0] (yes I am wearing slacks here!) but I didn't see a lot of people around me really interested in understanding the "other side".
There's also reluctance to address the connectivity issues. For most software developers internet connectivity is an ever present resource and it's a fault if it's gone. That doesn't really apply if you're deploying rurally, but multi-master distributed systems are hard and few people want to accept that's what they're dealing with.
In the end I came to the conclusion that no one appreciated my industry knowledge in AgTech (I must be one of the most experienced livestock identification software professionals on planet earth...) and it wasn't really worth investing my time in. That's not to say I would never work in the field again, but it would be in the capacity of working in offline systems, not my fairly extensive knowledge of cattle identification tags.
As someone involved in ocean tech, I sometimes see a similar attitude, although the number of former naval personnel looking for new opportunities helps to bridge that gap a little bit.
I've had tech people occasionally propose ideas, or ask if something sounds feasible, and one of the first things I often remind them is that the ocean is brutal to most forms of human engineering; something like Murphys Law incarnate. Weird physical stresses, bad wireless comms, corrosion of all sorts, biofouling of all the things, large temperature swings, etc. Sometimes a good day in ocean tech isn't "it worked", it's "it didn't work, but we got our equipment back out of the water." The domain knowledge involved in building or operating even moderately-reliable systems like that is hard to get without first-hand experience. There are lots of important, interesting problems offered by the ocean, but they always come hand-in-hand with these not-so-fun ones, which in my experience tends to dampen tech people's enthusiasm a bit.
I've often wanted to look into AgTech (grew up chasing sheep in Canterbury) but people I've talked to in that space say that it's incredibly hard to grow sales, farmers are inherently conservative on new expenses, what was your experience of that?
I can answer this one. I have an AgTech startup that deals with sheep.
The industry is incredibly conservative (which I knew, having worked in it for over a decade) but on top of that, it's also incredibly insular. You get one shot to get something right. You get one short per year due to seasonality.
You better be prepared to travel a lot. In-person training is not negotiable for farmers using a piece of new technology. Often they will be trying to use it well out of any kind of network access, your tech better work off-line and you better be there in person.
Different countries with the same livestock often have wildly different ways of dealing with livestock. You have to be adaptable to the local practices, often whatever you are doing has to fit in with some other thing the farmer wants to do with the sheep, because they don't want to have to get them in the yards unnecessarily. That goes for the availability of labour as well, some places you will be overrun with helpers, sometimes it's just you and the farmer and you need to be able to adapt to that.
The tech has to be rugged. I used off the shelf Android RFID scanners for part of my project and we have broken nearly everything we have tried.
Compared to Australia/NZ, the US market is just about impenetrable for anybody outside of it. I did travel to Silicon Valley looking for investors but got nothing but blank stares and head scratches when it came to things like genetics. Silicon Valley is about the last place I would ever look for funding again as there are too many fools with their hands on the investment funds looking for dogshit fintechs or web3 stupidity rather than actual businesses.
Out of curiosity, did you try approaching any of the Universities with big agricultural programs (Texas A&M as an example)? They'd have alumni willing to invest in Ag.
Texas A&M have some of my gear they are using with goats. Haven't really tried leveraging that as travel to Texas has been off the cards for a few years.
Heh. The cost of the individual animals means that all that dairy tech isn't really useful for sheep which is why I think renting tech in that space is the way to go. Distribution is still a problem for us though.
My commercial knowledge of this is all second hand, but I hear it's driven by a few big customers - essentially people with huge commercial farms who are willing to experiment. The idea of selling an expensive AgTech system to your local farmer at the pub doesn't really happen.
I recall my father hating technology because the irrigation systems for crop circles would call him when they stop functioning at night. Likely could just wait until morning for a motor or solenoid to be replaced.
I took a sabbatical a while back after a couple of back to back bad contracting gigs that had me thinking maybe it was not them but me.
My plan was to work on software for permaculture, but there were a number of confounding factors that limited that and I ended up abandoning the work. Portable devices were still not great in full light settings. Although that has improved in the interim, the bigger problem was that at least half of the permaculture people I knew were anti technology. More than a few were borderline pastoralists, and I found a group of them having an antivax conversation (pre-Covid, pre-Trump). I can’t recall the last time I was that disappointed.
Connectivity is a big issue as you say, but also the ergonomics are just bad for many settings. Glare is bad, especially when combined with dust. Farming and landscaping is dirty work. Touch screens do not like dirt, especially abrasive dirt. Work gloves prevent blisters and some dirt, but some gets through the gloves. But also getting gloves on and off repeatedly is a huge pain in the butt, which is something I only came to discover more recently when my partner or work would try to message me when I was working in the yard. It takes you out of your flow.
I think we will get there, but it will take OLED or better displays, and many more iterations on scratch resistant coatings (I gouged the shit out of my Apple Watch 5 by wearing it gardening), and I think then the path will still be voice recording, note taking and location tracking as the primary tools to split data collection in the field from copy editing accomplished while sitting at the kitchen table or office desk.
As someone with a foot in both worlds, I have been in tech for 30+ years and we also own a small cattle farm (can't even call it a ranch), there is a staggering amount of distrust in technology of any kind in the ag world. Every new piece of tech comes saddled with some license, or dealer only issue, or the old tractor/equipment could break down and I could fix it in the field on my own in a few hours, vs my new equipment broke down and now I have to wait for some service tech (there are never enough when you need them) to come out and reset some computer or replace a piece of software/tech at a ridiculous cost. And I can't work on it on my own.
Add to that, many farmers have been farming the same piece of ground for generations and they know everything they need to know about their land to at least break even most years and no piece of tech or software will improve those numbers enough for them to justify the cost/risk or the developers to justify the minimal profit growth.
I’ve seen the tech distrust with landscapers (who were also permaculture enthusiasts, but I’m not sure what that does to the Venn Diagram). And protest all you want, but this is a group that is exquisitely sensitive to UX concerns and we are barely adequate at managing UX for moderately tech savvy people, let alone technophobes. The Hands in Soil community needs an Apple serving their needs, but as you say they can’t afford premium products, so they will continue to be underserved until there’s a glut of empathy, and… well, don’t hold your breath.
It seems like the likely route is a combination of adaptation of Open Source and some amount of more humane freemium model than currently exists.
What’s the Open Source Equipment community been up to lately? That seemed like a promising direction but I haven’t heard a peep in a long time.
I hear the same about new tech-enabled equipment. I've always been curious, is there a market for continuing production of old-school, simple tractors and the like? Or is it one of those things that people complain about a lot but won't actually spend their own money to buy dumb equipment that may be less convenient or productive?
I'm more curious about why there isn't a market to offer what farmers want. Why not make modern, serviceable equipment with open code and published APIs and schematics? This just screams "market failure" to me.
Caveat: this is totally pub talk level speculation, read at your own risk.
There's a bunch of stories about how pissed the users of farm equipment are with everything being controlled by computers that are totally locked down to the manufacturer. They're the kind of "easy agreement" story that flourishes. I don't doubt that they're true.
What I suspect is that this segment of the market is too small and too conservative to support a new manufacturer. The end result being not enough sales.
If you're buying a product that lasts 30+ years, and there's relatively little follow-on revenue after the initial purchase, then the market needs to be pretty big to support a manufacturer. And if you're new to the game, who's going to trust you for a 30+ year product life, when you've got no track record?
I'd love to see someone try though: you could try to promote the openness of the system as a counter to the "untrusted manufacturer", I suppose.
Meanwhile, "big Ag" is perfectly fine with an equipment-as-a-service model, since it gets capital costs off their balance sheets, and manufacturers are happy to play along, because they can get a much simpler and more consistent revenue model?
My primary tractor is 4 years old and it has been in the shop more often than my 35 year old tractor. Most of us, especially the smaller guys, run equipment until it is quite literally gone.
The biggest issue that bringing new equipment to market is incredibly expensive and you are competing with all of the green, red and blue (and now orange) guys that have been building equipment for generations. The new issue is looking for a revenue stream for the manufacturers. My Case while not on a service contract, has very few owner servicable parts to it. My '94 Kubota still does the job but Kubota makes about $200/ year on me at most since I can fix everything on it myself.
It is by no means a market failure, the market is working exactly as it's supposed to. It's just that people think that the market is there for the farmers when it's really there for the equipment and inputs manufacturers.
I am not in manufacturing or anything. But I really think there could be a market out there for No-Frills appliances and equipment. Tractors that are just a tractor that work well and easily serviceable, long tail money made on replacement parts. Expand this to Refrigerators, Washers / Dryers, etc. So many products are made worse but cheaper, and have unused extra features that most people do not actually want.
A big part of this I think is using electronic components and no chips / pcb, so you don't need to have specialized skills to repair and service.
Question - how well known is US Cattle Trace? (Industry lead cattle identification scheme, for traceability).
I followed it with interest - they had leap-frogged the state mandated scheme in AU and NZ by pushing UHF RFIDs instead of the LF ones we're stuck with down under - but it's hard to tell if most US cattle farmers have even heard of it.
Cattle Trace is great for people with 10s of thousands of acres to manage. But that's not all that many cattle operations in the grand scheme of things. Most of us are small operations that don't interact with major processing plants and so on. We all know when someone has a sick cow and we all talk to each other.
I was part of a pilot program using drones and RFIDs in Brazil some 10 or 15 years ago. It was cool, but the ranchers still ended up having to go out and put eyes on cows when some went missing or died so it didn't really solve a problem, just narrowed down the area a bit.
I have a small drone here that I use so I can sit on the deck and still check water tanks and cows, or look for babies so I don't have to go out in the mud. But I don't know anyone else in 20 miles of me that does that.
I assume FAA Part 107, the basic license you need as a drone operator, wouldn't fully allow you to do what you're doing in the US at least. You're using the drone to go check on things you can't see from home. If you're lucky you'll be able to maintain visual contact with the drone, but you'd still need to go through the trouble to get a license.
This is a shame really because it really sounds like a pretty good use case that would be easy to adopt if you could just go down to some farm supply company and pickup a drone and be flying it over your farm the same day. Instead you end up having to figure out licensing requirements and worrying about the restrictions when you don't really want to deal with interpreting and understanding FAA regulations and you shouldn't really have to if you're using the tech over your own land in a rural setting.
Would be cool if one day you could buy a Matterport or Leica Disto equipped LIDAR photo scanner (like Fixar or Wingtra) to survey the entire property based on GIS tax parcel data and then in the software you could select things to photograph on a regular schedule and all you'd have to do is look through the photos each day. Eventually machine learning could be used on the photos and you'd be able to have something like pagerduty for agtech as someone mentioned elsewhere to tell you where to focus your time.
Not sure how well this would work with some livestock since I can't imagine they would like drones buzzing about.
That said, I've got no ag experience so not really sure how much labor this really saves. Just cool to think about.
I live on about 100 acres. I can see 80% of the land from my deck. The drone comes in handy for looking into the tree areas where the cows like to spend most of their without me having to go look for them. Besides, I figure if they won't do anything about the meth cook 3/4 mile away, they aren't going to care about me flying a small drone over my own land.
This is done all the time. When evaluating a new seed variety, a different seed spacing, a new nitrogen application, farmers will do a test plot next to the control. If things looks good then they'll expand the new change to more acreage.
The US Dept of Agriculture funds comparative research and agricultural extension agents help put it into practice. They actually know quite a bit about technology adoption in agriculture.
I came in to make the same point. You can reserve a small part of your crop for the experiment.
This is exactly akin to how social media companies do growth hacking. Try something new out on x% of the users/crops, measure the results. You don't have to risk your whole farm on it, unless there is a huge capital expense on equipment.
That is a pretty wild statement. I would assume A/B testing in farming has happened for centuries. Farms are made up of a multitude of fields. And so are a perfect environment for A/B testing. I did A in one field but stuck with method B in the other and saw that the former crops had higher yield, etc. etc.
There's farms near me that have a field of corn, every 5 rows has a different sign up with some sort of code and a logo for a seed company. I'd always assumed thats what they were testing.
I have worked on a corn seed farm detassling odd rows of corn. It is every other row for seed hybridization. The rows were marked, but it was pretty obvious because the two heirloom varieties were distinct. If there were more that two varieties, it is not seed production.
I used to live in a town in farm country and saw a lot of it from the back of a bicycle. There’s not much to look at when the corn is high except the corn. I don’t know if things have changed in the last couple of decades but it was pretty common for detassling to remove the pollen from four or five rows at a go, rather than odd/even pairings.
I’m in the permaculture community and most of these guys have stories about their neighbors looking at them like they’re crazy for years until there’s a catastrophic event and they see the land across their fence is doing better than theirs and only then do they start asking for advice, which they then take very slowly. For the people who think this is where we need to go to fight soil loss and global warming, this sort of learning by osmosis is very, very frustrating.
People blame these guys for making their money off of speaking engagements and consulting, but it’s going to take a lot to get to the point where your neighbors are on board. If you want to teach someone you have to seek them out, not just work locally.
Some of the newer farm equipment can actually apply different amounts of fertilizer (or whatever) to different parts of the same field, based on GPS and a grid of soil samples.
Someone wa pitching a project for this when I was at NCSA (which was less than two miles from corn fields) in the 90’s, as Netscape was taking over browser development and they needed something else to do (NSF is not fond of competing with commercial interests.)
That would have been an excellent match but I was barely interested and others were less enthusiastic. They started a collaborative software project instead. Which was cool but CRDT wouldn’t be invented for another five to ten years so I wonder what happened with that project.
It's hard to have a truly controlled experiment though. The result will be valid for that farm, on that year - but next year the conditions will be different, etc.
That's true of all A/B testing for real world things isn't it?
You could have measured increased traffic because something was trending on twitter/HN instead of the UI changes you just deployed. Might be completely different tomorrow.
The key insight Silicon Valley lacks about agriculture is that there is little potential for a big win. Major improvements mean 10% of something, not 10x. US agriculture is already highly efficient.
On the down side, there's potential for a big loss if you screw up.
I'd go 1 level deeper than that. Agriculture isn't a real market at all. It is closer to govt infrastructure contract work, than selling a commodity. There are no wins, juts minimized losses.
Govts decide farming subsidies, allowable environmental damage & even exploitative immigrant labor practices based on the prices their voters would be willing to pay. This means that an order of magnitude improvement in efficiency will likely lead to reduced subsidies rather than increased profits. Also, the value of food is subjective. Increasing lobster or caviar supply will decrease its perceived desirability just as fast. So infinite scaling laws stop applying pretty quickly.
The one agricultural industry that is ripe for disruption is the fake-DOP industry. A lot of European premium processed DOP goods are expensive because they are manual labor intensive. If you can train some minimum-wage laborers in a 3rd world nation to do the manual labor, then you can sell practically-perfect Parmigiano-Reggiano or Jamon-Iberico for a fraction of their price. As of now, all the fake DOP products are vastly inferior (it isn't rocket science) and still incur huge labor costs because they're still produced in 1st world countries.
Premium sausages, Premium cheeses, Premium sliced meats are ripe for disruption. It won't replace these products in top-tier $$$-$$$$ restaurants. But the more numerous $-$$ restaurants would be salivating at the prospect of such a replacement.
I am already seeing some of this. ~~Columbian~~ (Colombian) Burrata is both cheaper and a better product for US customers due to faster supply chains. I'd like to see more of this from developing and underdeveloped countries with the right climate, soil and water table.
For those unaware: DOP == "Protected designation of origin", or in Italian, "Denominazione d'Origine Protetta". Effectively a form of terroir or appellatian
From a bakery equipment maker: "To meet the burgeoning demand for products that look and taste hand-crafted, bakers had to go back to the drawing board with their ingredients and OEMs had to retool their equipment to handle the different types of dough used to make these products. The success of this effort can be seen in the bakery department of your average grocery store."
How artisanal bread is really made.[1]
I didn't know Tetra Pak made machinery for artisanal cheese automation, but they do.[2]
I strongly disagree. Food manufacturing is a process with a massive amount of variables, which we can neither model nor control for in a factory.
Every batch of supplies is slightly different. The flavor trends vary by season, and even in the most controlled environment, things go wrong. Additionally, cooking/food manufacturing often involves extra 'learned' steps that aren't properly documented or understood.
It is not that cooking/food-manufacturing can't be automated. But perfect automation of it requires so many sensors, so much multivariate optimization, incredible amount of manual verification and so much modelling.... that it quickly becomes cheaper to just have manual labor balance those all by feel. As someone who worked in highly automated manufacturing lines, the machines we work with all are all faulty in their own little ways. Being able to have an intuition for what just-right looks like and being able to reproduce it manually, has continued to be a difficult problem for all manufacturers. This is doubly true with food, where the scientific processes for certain things are still unknown.
I have tried aged American Parmesan cheese blocks and they are objectively worse. Barbeque seems like the most automatable process, yet nothing comes close to a a real pitmaster's brisket. Sometimes the main task of food preservation is simply keen inspection & disposal of bad batches, which remains economically un-automatable.
Artisanal products aren't always better (sometimes you need steel extruded dried spaghetti & dusty parmesan), but they are often different & hard to reproduce with a machine.
This is actually the big insight big time agriculture lacks about agriculture. They are obsessed with volumes and consider their farms to be large spread out factories making commodities and the tons of commodities produced to be the only metric that matters.
There are big wins to be had in making better food -- both healthier and better tasting. Most food in America tastes kind of bad and is very bad for you, so there is huge room for improvement. US agriculture is very efficient at making bad quality food. There is also a sub-industry of US agriculture-- organic farming -- who are very inefficient at making good quality food. Furthermore, there is a relatively large wealthy class in the world which have realized that health and a long life is the only thing that they cannot outright buy and every opportunity where you can trade money for some extra bit of health is worth taking. So there definitely is a market for good quality food.
There are other issues that are not considered critical yet but very soon will be. These are quality of soil and water. You can get good improvements if you make land that was previously unfarmable farmable.
But arable land is mostly a function of rainfall and flatness.
You can turn almost any flat land into arable land given enough effort.
Terraced farming cannot be mechanised economically because you need very small combines that need to be transported up or down and building the terrace with excavators etc disrupts the soil which means you have to build it up which means hardly any productivity in the first decade. You got a huge equipment, logistics and construction problem which requires massive amounts of upfront capital for basically no cash flow during the most important years. We would need huge advances in agricultural robotics for this to be cost effective.
What makes you say the organic market is so inefficient? And what room for improvement is there outside of the kinds of tactics that would make the food non-organic?
For the first question, one answer is the high price of organics. For your second question, there are many options. One promising option is targeted pest and weed control using cameras and image recognition and maybe even lasers. A system like that was actually mentioned on HN a couple of weeks ago.
Another is using biochemistry to make pesticides that degrade and disappear before the food gets to market.
Another is to bury seeds individually in the soil without plowing the soil, thus protecting the seeds from pests and giving them advantage over weeds whose seeds have to start on top of the soil.
I think there is an argument that agriculture is highly inefficient because we can't track externalities well. US topsoil in our bread basket was 14 inches a hundred years ago. It is down to 4 inches now. When it goes, we go.
I'm under the impression that an acre inch of soil weighs ~170k pounds.
So if we're losing .14 inches of topsoil per year - you only need to replace ~23k pounds of topsoil per year. Probably less than half of that needs to be compost.
I'm also assuming loss was worse in the past than it is currently. So you'd need even less compost.
This sounds bad, but not like something that can't be worked around. And honestly, it sounds like it could be a good thing for there to be huge demand for compost.
This is not a good take. The company that enables an improvement of 10% can still make tens of billions of dollars. I assure you that adtech knows all about the value of optimizing at the margin.
Vertical farming can do 10X, GMOs can do 10x, urban farming can cut out 90% of transportation costs, (not exactly what you mean but getting people to eat less meat would 100x our efficiencies) there is still room for improvement.
Let’s see you sell your easy ten thousand percent increase in farming efficiency.
The reason it’s hard to sell hype to farmers is they live in the real world and they’re not impressed by idiots that are so wrong that stupid isn’t a big enough word to describe them.
It takes 9 acres of solar panels to light 1 acre of vertical farm crops.
Theoretically, you could cover all of Nevada (~70M acres) & Arizona (~73M acres) with solar panels.
That would only give you enough vertical farm space for ~16M acres, and apparently the US only has ~10M acres for farmland for fruits & vegetables.
Point being - this seems implausible - and I'm not sure how you work around that.
IIUC, it takes about 1 acre to produce ~.12MW solar (obviously differs based on where the panels are). The world installed 168GW of solar in 2021 - so that's 1.4M acres of solar.
It'd take 100 years of solar production to cover that much area with solar panels.
And for what?
Cattle uses 4.5x more land by itself. We use 3x as much space to grow most of our calories (grains and oils) - replacing that with vertical farms seems like something from an alien civilization that is way to far away to imagine.
Note that vertical farms have about 3-4x yield compared to an equal area of land. They also require significantly less pesticides, water, or fertilizers. Though currently they struggle with many types of vegetables.
3-4x yield for which crops exactly? They are only used for growing leafy greens and low volume high cost fruits such as strawberries. Do they produce 4x corn and soy? 4x potatoes?
The capital investment required to grow corn indoors would be enormous relative to the per bushel cost. I reckon the operational cost would even outweigh the revenue.
Average corn yields in Iowa were 205 bushels per acre in 2021. Let's say corn is $10 a bushel (trading around $7 on the spot market right now). So $2050 per acre for conventional ag.
Let's assume 4x yield for indoor production, so roughly 1000 bushels/acre. Or $10k per acre at $10.
I had trouble finding any good information on energy consumption in an indoor cropping system, but I'd bet lighting energy consumption would eat up the additional revenue from higher yields.
Edit: Please let me know if anyone has access to scientific literature/experience on indoor farming energy use, I'm genuinely interested to know
Perhaps you can collect light vertically as well? And deliver it to crops either via the conversion to electricity and back, or even by some sort of optical/waveguide/handwave thing?
There's only so much sunlight that reaches any point on the Earth throughout the day. You could build something really tall to collect more light from the sun at more angles, but that will cast a shadow and reduce what's available for other collectors. So you end up with a lot of collectors receiving little sunlight or fewer collectors spread far apart.
It is also unreasonable since there are better competitors like artificial starch synthesis where you can use bioreactors to produce starch directly from CO2 and electricity at 8 times the efficiency of corn. If you add artificial lighting then vertical farming is at least 16 times worse than artificial starch synthesis.
The freed up land would allow us to then grow normal non nutrient dense food in its place instead which obsoletes the need for vertical farming.
Vertical farming can grow micro-greens pretty well but it will never produce 90% of our calories, even with optimistic assumptions. It would also be an environmental disaster.
Transportation is a teeny tiny fraction of the environmental cost of food as well.
I'm also not aware of any GMOs that are 10x better than their non-GMO counterpart although there are certainly improvements to be had here.
GMO yields are actually lower than non-GMO in ideal conditions. The savings come in other ways, such as no manual weeding with glyphosate resistance (or higher yields compared to letting the weeds rob NPK), or less insecticide use and spraying for BT corn. As a rough estimate, yields are about 6-10% lower for GMO seeds, but labor savings can be huge. GMO cotton cuts labor costs by 90%.
> GMO yields are actually lower than non-GMO in ideal conditions.
If, as is true of the most commom GM crops, the specific GMO trait at issue is a non-yield trait like, say, glyphosate resistance, sure, that’s what you would expect. (Though the yield is substantially higher in the intended conditions; real crops rarely grow in ideal conditions.)
Of course, there are crops modified for yield specifically, too.
It makes perfect sense from a biological perspective. I am not aware of any transgenic methods to increase yield as that is done through breeding and hybridization. Do you know of a specific seed that has enhanced yields through a crisper like process? I am aware of accelerated mutations by irradiating seeds, though I think this is rare and I only know of red grapefruits that successfully used this technique.
> I am not aware of any transgenic methods to increase yield as that is done through breeding and hybridization. Do you know of a specific seed that has enhanced yields through a crisper like process?
Not sure that anything particularly CRISPR-like is involved (there are other transgenic techniques), but:
It depends upon what you're defining efficiency as. In terms of yield per acre, there are estimates that yield per acre can increase by a lot more [1]. I haven't seen anything more efficient energy or cost wise (outside of a couple types of plants, really just microgreens).
Yeah sure at 1.5 million watts per acre of grow surface in LED lighting. The average American eats 2.5 acres worth of food and produce per year, which is a 3.75 million watts of lighting for 12-16 hours a day per person.
Now sure that number can be improved, but it also isn't accounting for additional HVAC costs, and you are still looking at atleast 2 million watts per person.
I ask myself this. Are the vertical farm fans really this ignorant? Or is it an act? But there are also people who will pop up and advocate for "personal rapid transit" (they are known as "pod people"), hyperloops, NFTs, and all types of other things that are impossible, nonsense, or pointless. Vertical farms are just the latest of these: a stupid idea with no future whatsoever, but with a vocal army of online boosters.
I think they are really this ignorant. The tech suburbanite ideas about farming are insane.
Vertical farming is just one of the cringy ideas. Farm underneath solar panels instead of putting the panels on one side of the farm, monocropping is pure evil and farmers are simply stupid for doing it. Grow cover crops so we have to dedicate 50% more land for agriculture.
The best greens I've had come from a vertical farm. Its too bad - I otherwise think its a wasteful, inefficient technology. I suspect the high quality didnt really have much to do with the vertical-ness of the farm though, and could likely be reproduced in a more traditional greenhouse (with some supplemental lighting during the shorter days).
I think most people have no idea how thin agricultural margins are to start with. They see that some farmer bought a $500,000 combine and think they got money just to throw around, but that combine is expected to last the next 40+ years and work near flawlessly for every harvest while performing multiple different functions simultaneously in a dirty and uneven environment. There is no alternative to a combine harvester since it is replacing the work of reaping, threshing, and winnowing all at once which without a combine would take thousands of man hours to do by hand every harvest. They might not expect for it to pay itself off for 30+ years.
Nearly all farms operate on single digit percentage returns, a large number of them have been floating along on 1% or less for decades hoping to hit a bumper crop once or twice more in their life time. If they throw a million dollars at fertilizer, fuel, seeds, maintenance for their tractors, barns, silos, dryers, and any other specialized equipment for their particular crop, then earning $50,000 on that million dollar investment would be considered a great success. Earning $10,000 would still be considered some success because it is not uncommon to run multiple negative years in a row.
Yield can vary 30% in a year from normal weather patterns, and the market can vary just as much or more, and they don't always align in a farmer's favor.
So if your tech can't handle 99.99% uptime, and be fixed in the field when it does break on short notice with basic hand tools, farmers don't want it. A single failure at the wrong time can dump half of your million dollar investment down the drain which you were only hoping to make a couple dollars from to start with. It could set them back a literal decade or more financially because some shitty widget or sensor failed and they either couldn't repair it or the fault wasn't detected until it was too late.
And don’t forget that many farmers take low season jobs for supplemental income to keep things afloat. Some only make profit from this extra work. Their farm is in fact underwater.
We talk about consolidation like it’s some evil conspiracy. But some farm kids realize dad is just holding into a dream that will never be true, and if none of your kids want the farm, on principle or because half a farm can’t support a used combine, then you’re going to sell or the kids will.
Gabe Brown, if I’m remembering his back story properly, ended up with about a third of his parents’ farm, split with his siblings. He got hit with hail every time there was a storm, and tried permaculture (lite) as a Hail Mary. He scratched and clawed his way out of the red and eventually reassembled the family farm buying from neighbors and siblings.
For his kids he established subsidiary enterprises and gave them to his kids instead of the land. Some stayed in At and some didn’t. One of his kids is slowly taking over management of his farm and he will get it eventually. Gabe probably has 10-15 years left in him.
I was looking into this recently and while I didn't find a great analysis, the impression I got was that financialization of the economy is a big part of this.
There are tons of tax benefits for owning land and for many of them you don't even need to produce anything or even have a working farm. You just need to own eligible land. Combine this with the fact that farmers are often capital poor so a lot of farmland is leased from those that own it. We basically have the same issue that renters are facing in modern urban America but with farmers.
Getting the financial industry out of assets we can't make more of seems like the obvious choice.
Farmers also hate to be dependant on closed technology they can't fix themselves. When you are out in the field and something breaks, you can't wait 3 days for the dealer to come around and plug a computer in to fix things... There is a reason farmers hack their John Deer or use older equipment like my dad does.
Silicon Valley is great at tech because it has a high concentration of tech companies and top universities. It doesn't have that for agriculture.
The "Silicon Valley" of agriculture is in Wageningen, a small university town in The Netherlands. In the surrounding area you have a huge number of food/agri-tech firms.
There is a huge amount of agriculture within driving distance of Silicon Valley. The main obstacle is culture, not distance. Most people I meet in Silicon Valley have zero clue how food is actually produced, and lack the proper respect for farmers and ranchers.
It's not exactly Silicon Valley, but UC Davis, which has a lot of Ag programs, is within fairly short driving distance of the bay area (to the point where some consider it potentially part of the 'greater bay area')
I had a visit to Wageningen and really enjoyed it. As a gift for delivering a lecture I received a cookbook for insects. I never used it, but the cookbook has pride of place on my shelf.
i’m not sure if silicon valley is great at technical prowess anymore. the hype machine has run its course and the tech is more proliferated and distributed than ever before.
Pretty sure you can subdivide a farm however you want and use different methods across them. Iteration cycles are in the year time frame but otherwise seems pretty easy to test whatever you want.
A buddy of mine works at Simplot and does exactly that. He works on a team that designs experiments and run them on crops.
Now smaller farms probably don’t, but then again smaller startups often don’t have A/B infrastructure setup either.
Edit: Come to think of it, despite this being an article on what SV doesn’t get about AG they talk of A/B testing as some special SV only thing. In every other field they’d just call it “scientific experiments”.
On the ranch I grew up on, we would "A/B" test new bulls all the time. Literally the first year we would try a new bull, we'd only give him access to 15-20 cows. If the calves came out easy, and grew fast, he'd get put into circulation with the larger herd.
We also tried different things with crops... plant one side of the field one way, plant the other side a different way.
Different kinds of fencing in different pastures... I can't remember a year when we didn't test a new kind of fence. Fencing takes so much labor, even something that shaved a few days off the maintenance each year would have been welcome improvement.
I remember we tried moving gates to the corners one year, just to see if it made moving cattle between pastures easier. It did!
We tried a lot of different rotation patterns, too. Mostly to try and reduce bugs and have to spray for flies less.
Perpetually testing new medicines. "This pill costs more, but you don't have to give it as often, vs. that pill costs less, but you have to give it every day..." turns out there's benefits to being out among the herd every day.
Ranchers aren't very technical folks, but they tend to know their profession extremely well. Above all else, they need things that reliably work.
I remember switching to electric branding irons... like they made branding faster, gave a more consistent brand, and had faster heal times -- but it wasn't just a decision to switch the whole herd, we had to test them out on a small group... and learn and then apply what we learned to the larger herd the next year.
A ranch was, honestly, a sort of perpetual experiment. There's very little "We do it this way because that's the way it's always been done..." If you talk to one generation of cattlemen vs. their fathers, the sons will inherit a bunch of the equipment, but they almost always want to change things up. I'd wager no group of fathers and sons on the planet butt heads more.
> To feed a growing population, we need to double global food production by 2050, but we can't double agricultural acres. We need more food (or more high-quality calories and nutrients) per acre. That's where technology comes in.
While it's true that increasing global food production is one way to enable food security for a growing population, I think it's also pertinent to analyze the inefficiencies within our current supply chains. Food waste is a big issue[1] and efficiencies gained in this area could also have a big impact.
Food is ridiculously cheap to create right now to the point much of it is turned into fuel. Once it becomes more constrained, food production will increase as demand causes additional supply.
Those inefficiencies similarly become removed once demand is higher.
I suspect a large aspect of this is not merely caloric sufficiency (already an issue), but protein, for which much of the world remains deficient.
Protein, whether animal or plant-based, is both an essential macronutrient and sees far lower productivity than staple (carbohydrate) crops. The Green Revolution and following developments have seen major enhancements in the productivity of rice, wheat, and maize. Far less so of soyabeans, whether for human or animal consumption.
Beef cattle is amongst the least productive (and most desired) animal proteins. But pork, poultry, eggs, and dairy, as well as farmed fish also require feed, including protein-rich feeds, to develop.
(I'm not making arguments over the morality or benefits of meat vs. vegetarian or vegan diets, though that's also a concern, especially concerning land-use, degradation, and greenhouse-gas emissions, of which animal metabolism alone is a very significant source.)
Protein-deficient diets, especially amongst infants and children, are a major contributing factor to developmental and cognitive deficiencies. So improving global standards of living will be dependent on increasing overall protein availability.
Perhaps improvement in the quality of life (=wealth) of the population as well as population increase? The world is incredibly richer today than in 1992 (particularly Asia, which accounts for about 60% of the world's population).
I am friends with some local farmers who have a multigenerational farm with a few thousand acres, mostly wheat production. Talking with them, the single fact that surprised me the most is how short the lifespan of a combine is. Those things break all the time, are very difficult to repair, and parts can be insanely expensive, and only reliably last a decade or so.
Another thing is that no one makes equipment for hills and that makes their lives harder.
Another reason they are distrustful of new startups is they need parts and the ability to repair incredibly expensive equipment for about a decade, and often startups don’t last that long.
The thing that a lot of people tend to get confused about is the difference between an innovation that can delivery some sort of ROI for a farmer and a venture-backable opportunity.
It is true that there is a lot of misunderstanding on the ag and tech sides, but also a lot of mis-application of 'templated' thinking - we've written quite a bit about it - one key piece is here: https://tenacious.ventures/notes/how-silicon-valley-set-agte...
Also, it is important to understand the difference between genuine innovation in the agrifood value chain, and on-farm efficiency improvements. As has been pointed out here, farmers are only part of the story, and whilst they are often the user of an innovation, it is often the case that others downstream of them are the true economic beneficiaries - this can only really be solved with business model design, not just technological innovation. There is a tendency to confuse the user with the beneficiary - https://adnause.am/notes/confusing-user-beneficiary-messy-mi...
In terms of how we review investment opportunities from founders, we need to see both elements of technological innovation and business model innovation. All the comments about the challenges in providing solutions for agriculture are valid and often present major challenges for adoption. But in my experience, adoption at scale is more often a business model problem, not a market or technical problem.
It is definitely the case that many problems that are present in the agrifood value chain aren't inherently 'venture scale' - and often that can be a big challenges for founders who can't self fund. Creating and running a profitable, self-funded company is a great endeavor; possibly very rewarding and impactful, but that is not the same as looking for investment for a company that hasn't yet solved their own venture-scale business model challenges.
Interesting question: what other industries share some of these characteristics (but not all)?
I think some people in tech assume that a lot of advantages we have are present everywhere, but they're not, entirely. A lot of industries share some of this innate conservatism, but in different degrees.
The first one that occurs to me is medical. Elizabeth Holmes' big problem (besides being dishonest, of course) was that she thought she could apply the Steve Jobs paradigm to medical testing. She would say "first they fight you, then you change the world" as if that would work everywhere. It doesn't. Your risks are enormously higher in medical, and there are government regulations everywhere you look.
But there are probably other industries like that, too. Autos, maybe.
Agricultural experiments play a prominent role in the history of statistics [1]. Yes, everything varies, so they do trials using different squares of land in the same field.
So I think it's a little much to say you can't do A/B tests, but it's a good point that they have to be pilot studies, not testing things "in production" on people's farms.
Point 4 seems more like a cultural thing than a technical limitation, at least for large farms. Why not split the farm into a few pieces and apply different procedures to each piece?
I launched a self-funded startup to focus on technologies I developed that have the potential to deliver significant benefits to indoor farming operations. We launched in 2019, yet, the pandemic put everything on hold for almost two years.
This year we finally completed our first installation (paying customer) at a farm in Singapore. Yes, we are US-based. Due to circumstances and opportunity our first installation happens to be abroad.
I applied to YC for multiple batches (I don't remember if it was three or four times). Rejected every time. I think it was an algo-rejection far more so than someone explicitly rejecting the company. We have shipping hardware and software products, a sizable first sale and at least one happy customer. This isn't an idea, it's a validated opportunity with solid execution. Our customer is looking to expand and wants more of our product. Lots more. We also have a product plan that will expand into other related domains.
I don't know if I have a solid opinion about SV and agriculture. Starting 2023 I will be out there making a solid effort to seek funding to be able to press on the accelerator. I think $5MM would be about right. I will also be focusing on IP, which, like it or not, can be crucially important when dealing with hardware that could potentially unlock a billion dollar business.
My only attempt at fund-raising has been applying to YC, conversations with our customer (who might be interested in investing) and a good friend of mine with deep and wide business experience in globally recognized companies. I guess I will soon learn what SV has to say about indoor farming startups.
Being self-funded is great in that you can just drive and not worry about anything that might compromise your ability to focus on the mission. That said, at some point, if you are going to grow, you either need a series of massive (and profitable) sales to supporting customers who will wait for you to deliver (hardware takes time to build, etc.) or a strong cash infusion to front-load development and manufacturing.
In looking at this industry I see companies who have raised hundreds of millions of dollars, even over a billion, and are flaming out. As an SV outsider, I see this as a syndrome of sorts. I view it as companies who might not have a reasonably solid business or product plan just trying to, in a way, force their way into a market position by burning cash. This is very, very different from self-funding a startup, doing all the R&D, manufacturing and shipping product, having a satisfied customer who wants more and, at the same time, walking away with a ~33% gross profit margin. I can understand burning cash to accelerate, yet, I think, the fundamentals have to be there in the first place. If I were to guess, I might say that this could be the part a lot of SV startups get wrong.
Anyone can pretend they know how to sail with good, gentle, plentiful and steady wind.
You appear to have a successful, working product that does not rely on regulatory arbitrage to suck up profits from existing market participants.
Thus, your business is not appropriate for YC. That's a good thing. Outside of the Valley, YC and YC companies have a horrible reputation and would hinder your growth and mission more than it would accelerate it.
I used the term as a general reference to applications not receiving full reviews. Given the tens of thousands of applications YC receives, I have a sense there's a filtration layer that is quick and, for lack of a better term, somewhat blind. I am not being critical of this at all.
I think they do a pretty good job of having you describe things succinctly.
I can't criticize YC at all. Not everyone can get in. I'll keep trying until I either get in or don't have the need because I secured funding elsewhere. That said, the YC ecosystem is valuable, so, yes, I am very interested in coming into the fold.
Farms and farmers have been doing A/B testing for generations with seeds, fertilizers, weed/pest control, feed, medicines, and on and on. And it's a PITA. Running 2 different, smaller, operations takes a lot more time and money than just one larger operation.
My wife is 100% engineer and it took me 6 years to even try something different on our (her) farm.
[Disclaimer, this is my experience of the Australia and NZ scene, not much Silicon Valley involved]
The gap between "people who know ag" and "people who know tech" is just vast. And I sensed no great urgency to bridge the gap. I tried myself - talked to every farmer I could, moved to a working farm, even had a go at weighing cattle and applying treatment [0] (yes I am wearing slacks here!) but I didn't see a lot of people around me really interested in understanding the "other side".
There's also reluctance to address the connectivity issues. For most software developers internet connectivity is an ever present resource and it's a fault if it's gone. That doesn't really apply if you're deploying rurally, but multi-master distributed systems are hard and few people want to accept that's what they're dealing with.
In the end I came to the conclusion that no one appreciated my industry knowledge in AgTech (I must be one of the most experienced livestock identification software professionals on planet earth...) and it wasn't really worth investing my time in. That's not to say I would never work in the field again, but it would be in the capacity of working in offline systems, not my fairly extensive knowledge of cattle identification tags.
[0] https://lewiscampbell.tech/img/CattleTreatment.jpg