Upstream Ag Insights - January 15th 2023

Essential news and analysis for agribusiness leaders

Jan 15

Welcome to the 151st Edition of Upstream Ag Insights!

Index for the week:

Mineral Becomes Alphabet (Google) Company
Looking Ahead to the Future for Ag Retail
Activist Investor Jeff Ubben Acquires Stake in Germany’s Bayer
Raising the Flag on Soil Carbon Credits
Terramera Announces Seed Funding for New Subsidiary enrichAg
Corn Yield Record Shattered By Farmer’s 459.51 Dryland Bushels
Formulation Agrobiology
The Worst Tech Predictions Ever

Thanks for reading, sharing and subscribing!

1. Mineral Becomes Alphabet (Google) Company - Mineral

After five years incubating our technology at X, Alphabet’s moonshot factory, Mineral is now an Alphabet company. Our mission is to help scale sustainable agriculture. We’re doing this by developing a platform and tools that help gather, organize, and understand never-before known or understood information about the plant world - and make it useful and actionable. Together with our partners across the food production system, we're hopeful that these tools will - over time - drive a deeper understanding of the complex interactions of plant genes, the environment, and farm management practices.

This is notable news out of Alphabet (Google parent company) this week.

I think the first aspect is to acknowledge what it means for a company to graduate out of X, Alphabet’s moonshot factory and into the Alphabet business itself.

X is quite literally where Alphabet attempts to incubate technologies and companies that can “change the world”. It allows Alphabet to build out small (relative to Google for example) bets that could turn into huge impact companies.

Starting in 2017, Mineral was their agriculture focused incubation.

Alphabet only “graduates” these companies once they have proven commercial viability, meaning stand on their own as a successful business unit within the Alphabet business. They are likely to fall under the “Other Bets” business with the likes of Waymo (self driving cars). This business unit generated $753 million in revenue in 2021 for context.

This indicates that it isn’t just some “side project” for Alphabet. Projects simply do not graduate out of X unless there is commercial viability. That’s an indicator that for us in the ag industry, Alphabet is here for a long haul.

What exactly does that look like?

Mineral has actively been working with agribusinesses like Driscoll’s and Syngenta, among others. I think that’s a key point of emphasis. Mineral is working to enable agribusinesses, not compete with them or go direct to the farmer.

I have repeatedly stated this is the likely, and hopeful approach, in how entities like Alphabet and Microsoft come to agriculture and go to market. Take this quote from a 2021 Upstream Ag Insights:

When I think about decision making capabilities in digital agriculture my mind inherently goes towards non-input selling companies and companies that are software/analytics first driven. I would guess this is where organizations like Microsoft or Google would come into the fray to participate in the tech and data aspects of agriculture. Their core competency is data and software and their entire model is around empowering people and businesses, which could easily be farmers and agribusinesses.

Alphabet’s core competency is organizing and making sense of the worlds information, making the lives of people better. There is unlikely to be many better positioned companies from a core competency perspective to begin delivering on the promises of artificial intelligence in agriculture. When you combine the technical prowess of Alphabet with the fundamental agriculture industry knowledge, you can get 1 + 1 = 3 sort of scenario.

Specifically, Mineral will be focusing in three areas:

developing sensing technology that can generate rich data sets about plants
organizing agriculture data from disparate sources for machine learning (ML) and building powerful software algorithms
conducting research that can meaningfully advance our fundamental understanding of plant kind.

Let’s unpack the first two specifically:

Mineral has been open about developing a rover with sensors to be able to assess plant phenotype, ID weeds, plant health etc. It’s important to state, the rover IS NOT the product. The rover is a mechanism for more efficiently acquiring data via the sensors across fields.
According to CEO Elliott Grant (emphasis mine):
We found that most companies are not collecting the quantity, diversity or quality of data needed to take full advantage of machine learning. That’s why we built tools to better capture, curate, clean and augment multimodal data
Essentially, these sensors are what can acquire data that is, in Grant’s words, “curated and clean” because Mineral can assure it. When this occurs, Alphabet can more effectively train their machine learning (ML) models and deliver tools to agribusinesses. The cool thing is that once they train the model via data from the rover, they can take the sensor acquisition to other places; cells phones, drones or even satellites. This allows them to do something else that Grant has emphasized: deliver at scale.
At a practical level this might allow plant breeders to acquire a host of data on phenotypes in an automated fashion and then be able to layer on ML models on top to identify outliers or opportunities unable to be calculated by the human mind and rapidly speed up breeding outcomes, or in the case of Driscolls, be able to assess berry quality for example.
To bring this to an example that I reference frequently in Upstream, this could allow equipment companies to work with Mineral to access a model that enables sensors to assess weeds at high speeds with low latency, which the equipment company could leverage to speed up their time to market and the quality of product they have in the market as it pertains to see and spray capabilities (though there is something to be said about an integrated capability here, but that’s for a different post).
I think before moving ahead, it’s important to highlight exactly what ML is - It’s talked about a lot in agriculture, and I’ve talked about it a lot without explicitly highlighting it. ML is:
the use and development of computer systems that are able to learn and adapt without following explicit instructions, by using algorithms and statistical models to analyze and draw inferences from patterns in data.
Essentially, an algorithm that identifies patterns in data and predicts.
(Note: I have learned a lot from the authors of Prediction Machines. You can check out their books and numerous presentations here)
This isn’t very different than what humans do to learn and how we make decisions. Some of the most effective people tend to do three things:
1. Seek out and consume high quality information. Eg: Read books from Harvard Business Review, Scientific Literature etc vs. Us Weekly or tabloids.
2. Consume more and diverse information. eg: Listen to more podcasts, read more books across various topics vs. watch more streaming TV.
3. Consume in a structured and organized way. eg: They learn the foundations and build up and do so in a focused way and they don’t read a physics article at the same time they read a history article.
  This all informs how we see the world, how we make decisions, treat people and more. Then, as we make decisions we get “feedback” from people, business that help us understand best where and how to apply the information to get the best outcomes. Now if we think about it from an ML specific scenario in agriculture we can see that generally:
  1. There hasn’t always been high quality data to leverage. eg: not cleaned, connected and contextualized
  2. Data has been expensive to acquire leading it to be less robust (eg: not enough geographies for dialled in agronomic recommendations)
  3. There hasn’t been a consistent structure and approach to acquiring, storing and analyzing the data in a coherent way.
    To top it off, we haven’t always had the hardware (computer) capabilities to run incredibly complex ML on.
    Mineral aims to begin solving all of these issues to lay a strong foundation for the future of ML and augmented human decision making in agriculture.
Even though I got started with it above, the best place to move to with “organizing agriculture data from disparate sources” is with this statement from Mineral to AgFunder:

Mineral has already surveyed and analyzed 10% of the world’s farmland

What this means is that Mineral has combined disparate data sources like SURGO, field boundaries, satellite imagery and more. They then have 80 models that they can run on this data set to support companies, with infinitely more to come I am sure. This may not sound that compelling at first, but it allows all agribusinesses to stand on the shoulders of a giant; agribusinesses can then plug in their own proprietary data and work with Mineral to access more data they need, or create their own ML models that help them achieve their own goals - whether that be yield prediction, quality forecasts or something else entirely in the world of business intelligence. And remember #1, they can also leverage Mineral capabilities to acquire more data.

This means data is still a valuable asset to companies and they can leverage it more readily via working with an organization like Mineral.

Last week I talked about “the ultimate roadmap”, which essentially is using data to identify potential outcomes and augment decision making where I shared this Gartner chart:

Source: Gartner

What does this mean?

While I don’t believe this gets us to anything like knowing exactly what a crop will yield the moment you put it into the ground, I think having a group like Alphabet structure data properly, access data more readily along with layering powerful ML on top gets us closer to getting beyond gut feel in decision making regarding agronomic decisions, breeding decisions, insurance decisions etc. For example, Mineral might bring us closer to the “percent likelihood” of an outcome occurring. By that meaning, Mineral could help creat software that gave agronomists the ability to see that given the current conditions, environment (and every other important thing) using a fungicide product gives the farmer a 80% chance of an economic yield increase.

How does Mineral Monetize?

Mineral hasn’t stated specifically what their business model will be publicly. The easy answer is likely through some initial project based approaches, but Alphabet wouldn’t want to be focusing on something that doesn’t have scaling capabilities, as well as a scalable business model.

When I had a conversation with Elliott Grant he made an interesting comment: ML as a service.

If we think what platforms have been built in the past two decades, we have seen software as a service explode. Algorithms can be a layer on top of that enabling better outcomes from software and from using data effectively, adding incrementally more value into the software.

This can help teams with:

out-of-the box predictive analysis for various use cases
data pre-processing
model training and tuning (eg: combat model drift)
run orchestration
model deployment.

This makes teams more cost efficient, faster, and gives them access to better tools - just like in-house software development generally is not as good as enterprise quality software from a pure play SaaS company, we could expect the same in an ML setting coming from Mineral.

Do they have competitors?

Over the last couple years we have seen announcements talking about similar things to what Mineral is wanting to do from ag specific companies:

INTENT announced in November. You can read my write-up on them here.
AgMatix announced in March of 2022 their efforts. You can read my write-up on them here.
Cropin announced in late 2022 their capabilities, which was wonderfully broken down by Rhishi Pethe (he also happens to be the Product Lead at Mineral, so he knows a thing or two about this area!)
EIWA as well, who I haven’t written extensively on.

Final Thoughts

Agribusinesses generally do not have a core competency in data and analytics. Alphabet doesn’t have the same depth of agriculture specific understanding that an agribusiness does. Together, they can leverage each other’s strengths and core competencies to get to better outcomes, whether a larger agribusiness, or start-up.

I know we have heard about the potential for a decade plus surrounding artificial intelligence. With more players focusing on artificial intelligence I am more optimistic than ever that we can begin to see this science bear fruit within the agriculture industry.

Share Upstream Ag Insights

2. Looking Ahead to the Future for Ag Retail - Crop Life

One of the most frequent questions I get is “what does the future of ag retail look like?” (second is probably “what will the grower of the future look like?”) I’ve given probably a dozen presentations on my views, but I think it’s most useful to provide frameworks for how to think of what the evolution could look like.

The key thing to consider about the future of ag retail is that it won’t just happen - ag retailers like those reading this make it happen. Not industry analysts or newsletter writers.

The way to make it happen is via coherent , differentiated strategy, which I think is best identified by strategist Roger Martin’s framework:

What is our winning aspiration?
Where will we play?
How will we win where we have chosen to play?
What capabilities must be in place to win?
What management systems are required to ensure the capabilities are in place?

Answering these questions comes down to understanding four important areas:

technology - what technology will be out there to either leverage or that could hinder your efforts? This could be important to incorporating to how you view the novel services and partnerships you want to focus on.
the business economics - eg: assets light vs. capital intensive have very different needs for example or maybe it means vertical integration.
customers and the future - what will customers look like in the future? What will they be needing? How will you segment and target? What trends are influencing this? This view will shape the efforts of retailers.
supplier and external threat dynamics - this is essentially the ability to analyze through a Porters Five Forces Framework.

Every organization is going to have different views on this, which is a key point.

Today, ag retail is generally pretty homogenous: the emphasis is on more manual services, more assets and a knowledge based value added approach. Everyone executes on this to varying degrees though (which is why in my opinion you can tell a lot of what you need to know about the quality of an ag retailer by their investment in training and support infrastructure).

But like Ken Zuckerberg highlights in the article, this homogeneity will change. I have believed this for several years and even though Ken and I approach the topic with different lenses, we come out to a very similar conclusion:

You will see consolidation of low margin operators, strategic combinations and partnerships, and possibly input retailers combining operations with equipment dealers and paradigm market shifts will take place. The ag retail market might split into various groups…. large full-service retailers, regional innovators/specialists, independent advice providers that are asset light, specific product-only sellers and logistics providers, and next generation service providers specializing in ag tech and partnerships.

Here is my conclusion from a few years ago in the Bundling and Unbundling of Ag Retail:

We will see the large players like Nutrien, Helena, Simplot etc all continue to acquire and focus on offering a scalable, best in class service and differentiated offering. But this will increasingly leave pockets of opportunities for retails to further reinforce or create niche bundles and focuses themselves.
The future of ag retail is not homogenous; it’s specialized.

We have a specific example of niche strategy today in the US market: Meristem Crop Performance.

They have identified the type of products they will and will not sell (eg: no crop protection products), they have decided how they are going to go about patenting, formulating and procuring products (build themselves and identify novel ways to formulate) and most importantly they have focused in on a very specific type of sophisticated and asset heavy farmers to be able to handle their go-to-market approach (direct to the farmer for large chunks of their revenue). They are entirely differentiated in their business compared to the traditional market.

To be clear, I am not saying everyone will look like Meristem. I am saying they are an example of a group that targeted a niche and exploited it because they were different and invested in the infrastructure necessary to deliver. My assumption is if you were to ask CEO Mitch Eviston or President and COO Rob McClelland, they would not only be able to answer all of Roger Martin’s questions concisely, they would be able to tell you what they have done and are doing today surrounding them and they would have a very explicit vision around the technology and the customer they have today and how they will evolve plus understand the economics associated with it all to a T.

Once the questions above have been answered and prioritized, there is the ability to go deeper because it fundamentally comes down to “what do you want the customer to experience when they do business with you?”

When I think about experience, I often reference the Uber experience (could use Amazon as well). It’s very easy to breakdown what Uber did to challenge the taxi industry: they reduced friction.

Y Combinator founder Paul Graham talks about a concept he calls Schlep Blindness.

Schlep was originally a Yiddish word but has passed into use in North America. It means a tedious, unpleasant task.

The “blindness” comes from ignoring and taking these “schleps” as a rule of the system rather than opportunity for innovating around them (aka “that’s just the way things are”).

To really differentiate an ag retail strategy, customer experience and the elimination of “schleps” will be even more important in the future as our exposure to increasingly pristine consumer experiences marrying the physical and digital (eg: Uber, Amazon) influences the expectations of farmers.

Basic principles of friction reduction are:

enhance convenience
decrease uncertainty
enhance transparency

At what points in the customer journey to transact can all three of those areas be managed positively? Again, the answers will vary by the focus areas of the retailer.

Today, we don’t always see the entirety of the customer experience being considered. But the retailer of the future will have a very precise view of how to reduce their target customer friction points.

It’s easy to continue to do what’s always been done, sell products that have always been sold, use business models that have always been used and partner with the same companies. But for the ag retails that are Finding Asymmetric Upside in Agribusiness, they will be leaning into the above and doing the hard work of building their own unique future state, not standing idle on the sidelines.

Share Upstream Ag Insights

3. Activist Investor Jeff Ubben Acquires Stake in Germany’s Bayer - BNN

This week a couple activist investors purchased significant stakes in Bayer, with the view that there is a need to split out the agriculture businesses from the consumer health and pharmaceutical business.

For financial background, Bayer purchased Monsanto for $63 billion USD.

At the beginning of 2018 before the close of the deal, Bayer had a market capitalization of over $100 billion USD.

When the Monsanto deal closed, Bayer shed 30% of its value within weeks of the close due to the glyphosate lawsuits.

It has only went down from there.

Before the activist investor purchases increased the stock price, the market capitalization was about $55 billion USD in early January 2023 showing an almost 50% decline in value since the acquisition, with the decline almost equalling the entire cost of the acquisition and Bayer now being worth LESS in totality than what they paid for Monsanto.

This is what has made activist investors circle: the pharma and consumer healthcare businesses are being dragged down in value by an unrelated business, which is what the activists see as an opportunity to spin out and separate from the agriculture business.

4. Raising the Flag on Soil Carbon Credits - Mitch Rubin Substack

This is a well written article by Mitch Rubin on some of the challenges with carbon credits. He does a great job summarizing issues and opportunities. I think it’s worth the read if you are bullish or bearish on the space.

5. Terramera Announces Seed Funding for New Subsidiary enrichAg - Terramera

Today, I’m proud to share that our hard work is coming to fruition as we announce the launch of our new soil-focused company. The new company is called enrichAg, and it is launching a transformational product called enrichSoil. The enrichSoil platform is a tool that farmers and agronomists can use to more effectively measure and analyze soil, improve variable rates, lower fertilizer costs, and ultimately boost the health and productivity of their agricultural land.

This release has numerous components, but is very vague.

One component of the product itself, an application to make soil testing more efficient and convenient is actually a good idea. There has long been an opportunity for soil labs to make the experience better for farmers and agronomists.

The issue is, there is no background on the soil testing technology itself they have. If you watch a video on their website, they state they have a sensor, but nowhere can you actually find out what it is and the technology in it. In looking at their patents there is nothing surrounding soil at all.

Yet, their CEO is claiming “99%” accuracy on “soil testing”, but then fails to state on what parameters (eg: OM, nitrogen, pH etc). And if you watch the video on their website they actively state that they run their “test” and then have to send to another external lab, which means they aren’t capable of testing across all important soil parameters today. There is also a claim of “real-time”, yet a sample is required to send into the lab still so it’s unclear if the real time dynamic simply helps their lab thru-put.

I reached out to Terramera across numerous mediums this week, if they find time to update me, I will be sure to share next week.

6. Corn Yield Record Shattered By Farmer’s 459.51 Dryland Bushels - AgWeb

I always find world record holder breakdowns to be really interesting. It helps illustrate what has the potential to become more mainstream in the coming decade from a crop input and practice perspective.

What stood out to me about this world record is the fact it appears to be done on a field that’s very similar to the rest of his farm and grown for the traditional economic reasons: manage input costs while optimizing yield. From Russell Hedrick, the record setting farmer:

I want to tell farmers that even if you can only cut back in the right way a little on fertilizer, the profitability side goes up. High yields are awesome, but if profit and improved ROI don’t come with those yields, then we’re spinning our wheels.

If you look at traditional world record corn and soybean growers, they often have specific fields or areas of a field where they will treat it much differently, throwing the economics out the window and purely maximizing yield - as a basic example, they might apply nutrients and specific biostimulants every day or every other day in order to manage nutrient levels in tissue, balance hormones and strategically minimize stress.

This farmer emphasizes profits, soil health and sustainability.

In fact, this farmer actually has a regenerative farming company.

7. Formulation Agrobiology - Harry Teicher

Harry Teicher is a PhD that constantly shares a wealth of knowledge around crop protection and formulation technology.

This week he posted three articles on insecticide, herbicide and fungicide formulation that are great reads.

They are technical, but for those interested more in the science he has many incredible resources.

Non Ag Article

8. The Worst Tech Predictions Ever - Trung Phan

This is a really interesting article highlighting some of the worst tech predictions ever. Many of them from some of the most successful people ever.

One of the big take aways is the famous quote:

It is difficult to get a man to understand something, when his salary depends on his not understanding it

Here is some background:

People just talking their own book: The #1 takeaway for me is that the business execs are just making calls that align with how their companies make money. The following predictions were just people talking their own book.
Ken Olsen on home computing (DEC was a selling mainframe computers)
Andy Grove on mobile computing (Intel was dominant on PC/Data Centers)
Steve Jobs on music subscription services (Apple had iTunes and the $0.99 per song model)
Steve Ballmer on the iPhone (Microsoft had its own mobile ambitions)
Larry Ellison shitting on cloud computing (Oracle dominant in on-prem data centers)
When you’re business is already crushing it, you keep doing what works. But that myopic focus on what works makes you vulnerable or unable to respond to industry shifts [Insert something smart about “the innovator’s dilemma”].

Upstream Ag Insights