Identifying The Lack Of AgTech Hockey Sticks

I have been thinking a lot lately about why AgTech struggles for exits. It's an important topic because without either initial public offerings (IPOs) or mergers and acquisitions (M&A), capital is not returned to the limited partners (LPs) that write checks to fund venture capital firms, who in turn write checks to startups. When IPOs and M&A are humming along, the virtuous cash cycle happens and funds from both those events are returned to LPs and VCs and future funds are raised in larger numbers and the cycle continues. From 2013 to 2021, AgTech had it pretty good as everyone came by for a look when Monsanto bought Climate Corp for $1B. Billion dollar valuations are nice - billion dollar exits are nicer. Well done Climate Corp team - Freidberg and his merry band of climateers got things moving for a roughly 20-25x lift to $53B in 2021 because founders saw that there were real problems to be solved (and being solved) in agriculture and funders saw a real life unicorn outcome that they could model against - excellent work all around AgTech Ecosystem team!

On top of the Climate Corp deal, there were acquisitions in multiple segments. If we just look at automation, John Deere bought Blue River and Bear Flag for $300M and $250M - not unicorns, but certainly very pretty ponies. CNHi bought Raven (startup or not - $2.1B is nice but to the how to classify it part it is a one off and tough to wash rinse repeat with Raven - Google if interested) and has done 3 other $100M acquisitions in the past 5 years (2 of which were for satellite navigation and machine vision technologies). But that's a pretty small count for over a decade of activity and, always worth remembering, it's been all M&A and no IPOs. Part of this is the nature of AgTech - it's proving hard for AgTech to get sufficient revenue and revenue growth to achieve exit velocity to complete an IPO. That takes away an exit option for startups and shrinks the available exit options from two to one. It's also worth remembering that the unofficial or whisper metrics for likely IPO success has gone up the last 10 years.

So in an era when venture capital has reduced by 50% ($340B to $170B from 2021 to 2023) and AgriFoodTech has reduced by 70% ($53B in 2021 to $16B in 2023 and again in 2024), it's apparent that exit activity needs to ramp back up to bring LPs back to the table with checks for VCs that will then write checks to startups and for AgTech that means relying on M&A, not IPOs. So where do we look for M&A? Well, two of the likely segments are equipment manufacturers (OEMs like John Deere, Case New Holland, Kubota, and Yamaha) and chemical input manufacturers (Bayer, Syngenta). Their revenues depend on growth and margins for growers at a level that allows them to buy equipment and inputs. How does the revenue look the last 2 years? Not very good - agriculture revenue nationwide was down 40% overall for 2023 and 2024. So growers are buying fewer (or trying to buy fewer) tractors and inputs. This means revenues for Deere and Bayer are going down. This means that capital available for M&A by the major segments of likely M&A buyers is down. This means there will likely be no increase in M&A, at least from the usual suspects for at least 2025 - we'll see about 2026.

So as you look at M&A and what could drive it back into play, it's useful to think about some Silicon Valley technology examples and the growth they provided and how they got there and compare those examples to AgTech. First, let's look at eBay. How did eBay make money the first 10 years when it had some hyper-growth periods (still makes money this way but less hyper-growth)? It got paid fees for listing items and for successfully selling items, as well as for promotion of items or sellers. What problem was eBay solving? It took a lot of local offline markets and turned them into online global markets overnight. If you had a collectible item (trading card, antique), your choices before eBay were to do a classified listing in the local paper or an actual ad in a local publication and make the local audience aware of that item. This meant a lot of potential buyers were not made aware of the item and the seller frequently sold the item for a "local discount" but was hopefully able to sell the item. Another option was garage sales, but the prospect of getting organized for a garage sale, standing in the driveway for a couple of hours and then putting everything away that didn't sell is a lot for most people and often sales are lower than hoped because (again) you only get to a local audience.

So eBay comes along and says "write a simple listing with a picture and everyone all over can see it and buy it and they'll pay for it online and then you ship it to them." Pretty compelling on both sides - sellers get access to a whole new set of buyers and buyers get access to a whole new set of product outside their area. What was regional becomes global the minute a listing is created. Think of the value destruction to classified ads and the garage sale activity as eBay became a more and more popular way to sell. Now think of the value creation as eBay grew revenue when they get more item listings, when they got more buyers placing bids and using buy it now, and when they added promotions to give listings extra visibility if the seller chose to promote them. And if you really want to think about hyper-growth think about this dynamic - there were many periods during the first 10 years of eBay when we were adding sellers and buyers fast enough so that view item pages (seller listing pages) and the average selling price per item (ASP) were both going up - there are few flywheels working better than that one. That's how you create hockey stick style growth.

Now think about Google. Similarly, Google made it a lot easier for businesses to raise awareness of their business online via Google search. Think about the business owner choices before Google (I know, it's the 90s, seems like a long time ago!) Two commonly used options were Yellow Page listings for the business or media ads in print, radio, or TV for the business or a promotion. Like the eBay competitive set, moving online created a wicked jujitsu. Two of the most impressive assets of the yellow page and media companies pre-Internet were (1) distribution (Yellow Pages were distributed to every house and newspapers were delivered to every subscriber every day they paid for it; radio and TV were available to anyone with a radio or TV that turned them on); and (2) quality sales teams (they had good sales teams with decades of experience and training at how to sell and up sell the right packages to merchants based on some discovery questions).

eBay and Google made those not only irrelevant - they made them boat anchors. eBay and Google said write up a view item page or a website and we'll bring an audience to you - no sales team required and the internet covers distribution for us. Brilliant leverage and a full and optimum replacement for old school distribution and sales teams. Suddenly phone books became a much less interesting item for consumers, as did the home delivery newspaper. And the job of the well trained sales teams got progressively and infinitely more challenging. Over time the largest assets of offline media became significant expense items with a much lower return potential. The economics on media sales teams changed as the marketplace dynamics changed.

And in case you're wondering why I said eBay had a nice hockey stick model for growth, take a look at Google. It's even one more step efficient. By allowing advertisers to set a keyword cost per click price, they optimized for the highest per click price on each search result page. The 10 blue links on the left provided relevant content (and quite routinely the best search result on the internet thanks to Google Pagerank) and the ads generated revenue every time they were clicked. Google had the same multiplicative effect - as search results were growing, so were costs per click and Google revenue because the math on a CPC basis was so much more compelling than the offline media math. And remember the silver lining for Google was that you knew exactly what ads got clicked, how often, and if you did it right whether that click turned into an actual transaction and for how much. End to end marketing analytics that were never possible with offline media were available for free for Google AdWords users. So search and CPC going up at the same time on Google were just like item pages and ASP going up at the same time on eBay.

What really creates the hockey stick is when new users show up on eBay and add new listings or on Google and create new webpages. This creates an improved search result for product buyers on eBay and for business or product seekers on Google - often at very low cost. With both eBay and Google, once the core product is built - the eBay sell item process with online payments and the Google search engine with Pagerank implementations - the incremental cost of another listing or indexed web page is effectively zero. The cost of goods sold is ridiculously low and often boils down to the server and server efficiency tools needed to keep the site up to make sure the listings can be seen and webpages can be found in search results. That is ultimate product leverage.

So now let's think about AgTech - specifically automation for this exercise. There's no Yellow Pages or classified ad competition. The "competition" for automation is the farm labor crews already doing the work. They can be of different quality and cost different amounts based on status (domestic or H-2A). As a baseline, let's put a range of $15-30/hour on farm labor for specialty crops. Farm labor is hard and most labor crews are doing great work in the fields every day they are out there, whether it's weeding, thinning, harvesting, or something else. And unlike most offline ad products, you can measure the results of the crew on a daily (or even smaller slices of time than that) basis. You can also measure the cost increase for a crew as things like adverse effect wage rate (AEWR), minimum wage, overtime rules, and housing rules and costs change over time. Farmers know these numbers because they're such a large expense item on the P&L they need to be managed well for the grower's operation to end up with any margin at the end of a crop season.

Let's take a look at one of the more successful automation segments in specialty crops - weeding robots. There are multiple formats of weeding robots in market. Laser weeders like Carbon Robotics are good for certain crop types. Mechanical weeders like Stout Industrial are good for others. Sprayer robots like Ecorobotix and GUSS work in still other use cases. They compete in different ways for the same dollars. As we have seen from the WG Case Studies, Carbon Robotics was able to save about $800,000 in year 2 where year 1 was labor costs only for weeding of 3,200 acres of 7 types of certified organic crops and year 2 was labor and robot costs combined. The total savings was $800,000, which at $20/hour equates to 40,000 hours which at 2,000 hours/full-time equivalent worker = 20 farm worker FTEs. That is a solid ROI analysis for a weeding robot and growers can do their own math (which can change based on type of tractor and whether it's owned or leased and type of labor i.e. domestic or international) with the WG Economic Templates so they can justify the purchase of a robot before committing to buy it.

So here's where we take a look at why the hockey stick doesn't show up that often in AgTech - particularly in automation. Think of the virtuous circle eBay and Google found themselves in during hyper-growth - the two key metrics for each business (listings and ASP for eBay; searches and CPC for Google) were both going up at the same time. Short of running way too many Super Bowl ads, it's really hard not to have revenue and margin growth when that is occurring. But in AgTech automation, the robots are used to replace labor. The key metrics are reduction in weeding costs overall and hours of operation for the robot so that the ROI period can be actually determined. There are no flywheel metric scenarios in this model - the ROI payback is linear. As hours of robot work replace hours of labor, labor costs are reduced and the robot ROI is delivered one hour or one shift at a time.

Looking beyond weeding, it's similar math and economic scenarios for many of the specialty crop labor functions. Thinning is the same. Harvest is the same. Planting is the same. Harvest assist (which is an alternative to robots that actually do the harvest work) is the same. Each of these solutions works against an hourly labor cost and completes hours and shifts to deliver the ROI on the robot. This makes sense because in all automation scenarios, the problem being solved is the rising cost and decreasing availability of domestic labor. So of course the metric to work against is farm labor hours and the cost of each hour. While this makes the math easier to calculate (but not easy) for growers and startups, it does remove much of the opportunity for hockey stick growth. And the end result of walking through all of this is it makes M&A at the prices (which are just realized valuations) startups need to return enough capital to the VCs so they and their LPs have enough capital coming back to reach Internal Rate of Return (IRR) targets on capital deployed. The way to get more capital from LPs is to outperform other investor segments on an IRR basis so you're in the preferred top spot for deploying the next round of capital.

So in the end the lack of a hockey stick makes attractive M&A outcomes much less likely in the AgTech automation space (not all AgTech, but definitely automation). This in turn means that new in-flows of LP capital are going to be tough to come by because IRR targets for automation startups will be harder to reach than other segments. This means we should all really appreciate the shrinking pool of AgTech-only venture capitalists (by playing only in this segment, they reduce the competitive stack for the IRRs that stack up competing for the next LP check. Oh by the way - they also understand the ag problem statements better than most investors because they focus on our segment).

The one place there may be an opportunity for a hockey stick is if automation vendors can turn the data they are getting from regular field passes into actionable recommendations that help the farmer with other tasks. For example, if the weeding robot could help the grower understand new planting or thinning options that could reduce the overall labor pool, that would be useful and deliver an additional ROI element. Similarly, if a robot that is in the field during the growing season could use data to predict crop yields, that would be a huge benefit to the grower and another ROI element. That said, both of these seem years away and speculative at best. But that's the vision where automation hardware could deliver a +1 ROI element behind the actual labor saving solution.

None of this feels like very good news for the automation space, but I think it's important to walk through the dynamics and get some of this thinking out there. With any luck I'm missing something that folks can help me out with to go deeper with the analysis. I would welcome that feedback and am happy to adjust my thinking with new data or different ways of thinking.