This post was updated on November 24th, 2020.
Last year I've heard about a startup that will launch satellites into orbit using a massive centrifuge in Silicon Valley — for a fraction of the cost of any other solution. The cost of satellite imagery is reducing dramatically and the quality of the photos is improving. At ONEVC, when we are looking at a startup we want to invest in, we always think "Why now?." Why didn’t this startup work 10 years ago, and what are the enablers that make it viable now? In the context of this post, cheaper and accessible high-quality satellite images are enablers of crop yield prediction — an important metric for better estimating the mortgage collateral used in rural loans.
A couple of months ago, I interviewed Hans Tung, managing partner at GGV Capital, for Brazil at Silicon Valley podcast. He emphasized that FoodTech and AgTech were interesting investments for Brazil. In his words:
"There will be food shortage in the future […] and there is value in marrying science with tech to solve real-world problem. Brazil has a lot of natural resources. It will be good to put together resources with tech and build something productive to help society".
Hans Tung, managing partner at GGV Capital
At ONEVC, we decided to look more closely at this space and found compelling reasons to believe that it is definitely a good time to invest in AgTech, especially the ones doing data analytics and lending.
Why should you start an AgTech company in Brazil?
1 —Why now? Satellite imagery and sensors are affordable
Satellite imagery is an enabler for many AI-based companies. With cheaper and better-quality images, companies can train their algorithms more precisely and at a lower cost. In agriculture, for example, startups are using satellite data and weather sensors to predict the productivity of each farm. This is important for lenders, who want more accurate estimates of each farmer’s collateral and also when a crop will be harvested. These two data points are enablers of credit underwriting for agro.
2 — Long iterations cycles for Agro suit the VC model
Each iteration of our machine learning model takes anything from six to twelve months to train depending on the crop type being monitored, so it takes years to develop truly accurate prediction models.
Brazil has a natural advantage for this: we have two harvests per year and no threat of frost. This can speed up our models and give an unfair advantage versus any US-based startup.
From an investor’s perspective, this long cycle can be positive: a dataset that is hard to build becomes a barrier to entry by new players. As a consequence, this type of business suits the risk-profile of a venture capitalist: a defensible business that takes a long time to be profitable. But once built, the potential for these kinds of businesses is immense.
To dig a little deeper technically, defensibility also depends on how the algorithms are trained and how hard it is to get data. To “train” the machine learning models that predict harvest size, you need real data from each farm — and that data is hard to get. The more farms you have in your platform, the more data you have, and the better are the prediction for each farm. As a consequence, the more farms you have, the cheaper the loan tends to be, creating a barrier of entry for a new fintech with a smaller client-base and a higher cost of the loan. James Currier describes this as data network effects.
3— Agribusiness is a massive market in Brazil
According to PwC, "Agribusiness represents 22% of Brazil’s GDP, 1/3 of all employment and almost 40% of exports."
4— Recent regulatory changes will bring more financing for agro
- Digital contracts: Very recently, a new regulation called “MP do Agro” was approved, allowing online registration of a very popular loan certificate in agro. “CPR” is an online contract promising the delivery of a certain amount of a commodity in a given location in exchange for financing. This online process brings transparency and reduces fraud in a very important source of financing for the industry. Also, this new regulation allows the loan to be registered in foreign currency by residents or foreigners. ONEVC’s portfolio company Docket already benefits from this change. According to a post by Valor Economico, the new CPR will inject billions of reais in the industry.
"The objective is to make the paper more attractive and dynamic, so that it can be used in a larger number of credit operations and can help to leverage tens — or hundreds — of billions of reais in offering resources for rural operations in the future".
Valor Econômico — December 12th, 2019
5 —Credit is not easily available for smaller farmers in Brazil
“Farmers in emerging countries have limited access to financing and insurance options, and interest rates are not attractive enough to sustain growth.”
PwC Report —"Agribusines in Brazil: an overview"
Small farmers still lack credit from financial institutions but depend on it to survive. Banks are not prepared to serve small farmers "locally" and other players such as local retailers, trading companies, and chemical companies "have to" do it to fill this gap. Among the players described, the ones that give more credit to the small farmers are the retailers. More likely, retailers would prefer to be paid now, with a financial institution holding the credit risk. Startups have the opportunity to sell data to big banks or to underwrite credit themselves.
Last Year, BTG Pactual invested in a startup doing data analytics for the agribusiness. According to Roberto Sallouti, CEO at BTG Pactual, tech allowed BTG to serve the middle market segment and the same logic applies for agro.
“Technology allowed us to serve a new segment of the market last year. With technology and online marketing, we started targeting smaller customers with efficiency. We believe that the same thesis applies to the agribusiness industry: With data and tech, banks can go down-market to serve a new segment efficiently”.
Roberto Sallouti, CEO at BTG Pactual
6— Good tail-winds for Agro FinTechs
- VC money is available to back startups: FinTechs are booming in the region with the IPOs of PagSeguro and Stone and the early success of new challengers such as Nubank, Creditas and Neon. So far, I haven’t found any compelling reason why VCs should not back Agro FinTechs as well.
- Small- and medium-sized farmers are underserved: Many startups target the underserved population. In FinTechs, the initial customer profile is usually young people, SMBs, immigrants, the underbanked, and so on. The reason big banks usually don’t serve small farmers is because the revenue per user is not attractive compared with larger ones. I don’t believe it is easy to serve geographically dispersed farmers, but it is true that more efficient tech startups can provide them with an opportunity. Additionally, most small farmers don’t have any access to credit, so startups will likely be the first ones to offer it. In Brazil, most of the smaller farmers have some sort of credit from retailers, not banks.
- Raising debt is now possible for startups: The significant drop in Brazilian real interest rates created a demand for alternative investments in the market. We see many FinTechs and PropTechs successfully raising funds to lend to their clients. In this case, most of the equity money (or the money raised with VCs) is used to build tech and the team, whereas the debt money is used for lending;
However, there are some challenges to be considered:
- Geographic dispersion: The cost of serving geographically dispersed customers (farms) can be high because, to train the algorithm with real data, you need a relationship with each farm. To address this problem, it is better to have a go-to-market that somehow aggregates the farmers (using co-ops, for example);
- Serving small farms is challenging: The segment that is underserved is harder to serve. Smaller farms will have higher default rates and challenging unit economics (LTV/CAC). Additionally, there is an adverse selection problem of trying to source "the best" from the customers ignored by banks;
- Underwriting for agro is not only about analyzing "new" sexy data: The new data available gives you a competitive advantage in collections and collateral prediction. However, you still need to underwrite credit as good as the other banks, using traditional sources of data. As in any credit segment, fraud exists and need to be mitigated;
- Building a proprietary source of data takes time: You need to find a business model that can generate revenue in the short term, so you can keep raising money from venture capitalists to gradually build a long-term vision with a unique dataset.
We want to meet AgTech entrepreneurs at ONEVC
There are more good reasons to be bullish on Agro in Brazil than there are challenges. Yet, we don’t see a lot of opportunities in agro at ONEVC. One assumption I have is that there is little geographical overlap between the cities where most VCs operate and the regions where the farms are concentrated. Most VCs are too lazy to invest outside their “city”. We are not!
If you are building the next category-defining company in this sector, we want to meet you. In this post, I focused more on data analytics and fintech for agro. However, there are many opportunities in crop productivity, biotech, IoT, and marketplaces.
Updated on November 24th, 202
ONEVC Led TerraMagna's Seed round
Following our investment thesis above on the AgFintech space, we've mapped the AgTech ecosystem in Brazil and found the perfect team to invest in.
ONEVC led TerraMagna’s Seed Round. We are excited to partner with the team and support the company’s mission to provide safe credit for agriculture. Bernardo Fabiani and Rodrigo Marques are friends from ITA with deep expertise in analyzing satellite images and use them for credit underwriting.
TerraMagna is reinventing the way credit is underwritten, provided, and collected in Brazilian agriculture by leveraging technology and alternative data sources to transform a risky and volatile process, which is currently run by less sophisticated players, into a simple and safe experience. With TerraMagna, retailers, industries, and farmers have access to fair credit and investors have access to high-yield, collateralized, market-uncorrelated, and high-volume assets. The drop in the real interest rates and the novel underwriting technology using satellite images allowed TerraMagna to raise its first debt facility this year.
Read it here on Crunchbase News:
"Agriculture fintech company TerraMagna raised $2 million in seed funding for risk mitigation, providing farmers in Brazil access to working capital"
Bruno Yoshimura — General Partner at ONEVC
Bruno Yoshimura is a Co-founder and General Partner at ONEVC. Bruno has accumulated 15 years of experience as a tech founder with extensive involvement in product development, sales, and marketing. Before pursuing an MBA at Stanford, he launched Kekanto, an online guide based on users’ reviews backed by Accel Partner and Kaszek Ventures, and Delivery Direto, a white-label platform for restaurants doing delivery sold to Locaweb in 2019. Bruno holds a major in computer science from the University of Sao Paulo.