Current States - AI & Farmland

I am back with the new format. Today I want to chat further about investing in AI as well as farmland.

AI

Whether computer science grads like it or not, the use of artificial intelligence by itself is a commodity. Anyone can download Tensorflow and Keras and then use it to create an AI for their data set. After all AI is advanced pattern recognition (wiki). The reason for this is that most of the AI algorithm are open source and researchers do not always like doing proprietary work (exhibit B). Meaning that it is unlikely you will want to, or can invest in the algorithms itself.

How easy it is to use start using machine learning software That leaves two other interesting parts of AI around. Hardware and proprietary datasets. On the hardware sides of things Google and Nvidia have interesting specialised processors called TPU's, they speed up neural network learning, and Google is putting that into their cloud based computing software (so is Amazon's AWS & Microsoft's Azure). The appeal is obvious, the faster the machine learning the better. Although as I discussed last week, quantum computing will be quicker than the bits on these versions of TPU's.

Whilst I believe that AI is a commodity, what is not is data. Good data with commercial opportunity is where those who want to bet on the future of AI will go to. A good example of how important data is becoming is in the agricultural area. What is called precision ag(riculture). If you look at John Deere, the tractor and truck company, all their machines now comes with sensors. These sensors have over seen great improvements in production efficiency and field resolution times as well as a significant decrease in warranty expenses. By collecting data and letting the farmers understand it, they have a commercial opportunity.

This is why Facebook and Google have done so well. Large datasets with actionable data on the consumer allows the best targeted ads, which advertisers will pay for - obviously. These ads can nudge us (also see my last post on this blog) to do things we would not otherwise do, and the insights from the data, allows these companies to know us better than we know ourselves. Why else is Mastercard and Microsoft now working together to even better understand who is who, and what do they do online?

The takeaway is: novel datasets and the solutions and insights that they can give is where the money is at.

Farmland

I have been meaning to talk more about this for some time. So let us begin. In economics there is an idea that gets thrown around a bit, Knightian Uncertainty. That is that we do not know the true unknowns. Agriculture as a business has a number of variables in which we cannot control things like the weather, CO2 levels, soil health and the like. All of this and more affects the yield of any harvest, as well as the value of agricultural products and the farmland itself.

Investors dislike things that they cannot control. But what if I told you that we could lessen our exposure to the vagaries of the wind and broader weather issues make money on future of our species wanting to eat. After all everyone knows that that the population around the world is expanding and by 2025 the population will have gone from 7.2bn to 8.1bn mouths people.

That alone gives me great confidence in owning - for the longterm - farmland. This is rather than actually investing in, and being apart of the actual farming itself. One cannot hit back to us that increasing harvest yields and technology means that over time, even with inflation, the cost of grains have fallen (particularly good read here). They may say that farmers who do not own the land will not treat it right; but that is a risk in every endeavour where we limit skin in the game. And even as grain prices have fallen, farmland has a high correlation with inflation, meaning for those that are worried about a return to inflation; buy the land.

According to Nijs (2014) farmland has 6 main drivers (1) global population growth, (2) changing dietary habits, (3) agricultural productivity (which is both an issue and an opportunity), (4) changing climate patterns, (5) the rise of biofuels production and consumption, and (6) the increasingly limited availability of farmland. All great drivers for the longterm and patient investor.

There is more to it, but buying farmland is a smart move. The two REITs on the ASX (disclosure: I own both) have a yield of ~5% and ~8%. Both have land that has different types of crops & animals and are geographically spread out. One of them Rural Funds Group (RFF.asx) publishes this infographic below which I find particularly interesting for anyone interested in the area.

From: Rural Funds Group website (here)

There is a lot of nuance in this area, and truthfully I probably do not know enough about it. But I like the idea that I own something that will always be needed. It seems to be low risk and medium reward (also see Clarke et al 2012). And when someone asks about water, and about changes in soil composition I get to say that does not worry me. I am not the farmer merely the one that owns the land, I can collect rent and get a good capital gain from ownership. Smart management of farmland REITs will make sure to be geographically spread out, have different resource entitlements and not take big concentrated risks.

WJD