Machine Learning and the Future of Agriculture

The world faces a fundamental food problem: quantity versus quality. For the last several decades these two notions have been at odds, with the gap widening. What can put a stop to this?



Can Plenty Solve the World’s Food Problem?

The continued use of big commercial farming to generate the world’s food supply has caused a major tradeoff in which increased food production comes with adverse quality [1]. These large-scale farming practices combined with long supply chains have not only diminished nutrition, but increased waste and risk to food safety.

Venture capitalists are betting on agtech to solve this problem. More specifically they are betting on Plenty, a Silicon Valley based agtech startup whose focus on machine learning has drawn over $200 million in funding from the likes of Jeff Bezos, Eric Schmidt, and Softbank Vision Fund [2].

Plenty: Using Machine Learning to Ignite a New Age in Agriculture

Plenty is a leader in the agricultural technology space. Operating as a vertical indoor farm located close to city centers, Plenty uses gravity-fed hydroponics (water based soil-less horticulture) to grow organic food. Through this method, Plenty grows crops that contain higher levels of nutrition and enhanced flavor, all while minimizing water usage and increasing yield [3]. While this approach may seem innovative, Plenty is not the first player to enter this market. In fact, others such as PodPonics, LocalGarden, and FarmedHere all tried and failed in the hydroponic indoor-farming sector [4].

Source: Plenty

So what’s different about Plenty and why are investors so bullish? The answer lies in Plenty’s use of machine learning. By machine learning, Plenty optimizes the growing conditions of plants. A multitude of cameras and sensors gather and analyze data of plant images with different issues such as nitrogen or iron deficiency. Powered with this information, Plenty’s software can prevent such issues [5] and create the ideal growing environment customized by species. Machine learning reduces the variability in agricultural practices and distills the elements of plant production down into its fundamental components. Temperature, oxygen, water, light, nutrients, and other conditional factors are all evaluated and adjusted in an iterative fashion. This approach, “can achieve yields of up to 350 times greater than traditional agriculture while using 1 percent of the water and barely any land compared to conventional methods.” [6].

What Will Plenty Do Next?

Plenty plans to build aggressively on this machine learning enabled business model. Having recently added a Seattle farming operation to its San Francisco headquarters [7], Plenty is looking to move abroad. Specifically, China stands out as the next target market because it lacks the necessary land area to feed its high population density [8]. Plenty is primed to serve this type of market because it appears to meet the need for an extremely efficient and high yield method of agriculture production in a relatively small footprint.

Plenty believes they hold the key to unlocking healthier eating for everyone not just in the US, but around the world. “Plenty has the capital and connections to accelerate its endgame: building massive indoor farms on the outskirts of every major city on Earth, some 500 in all.” [9]. With this type of growth ambition, Plenty is doubling down on its machine learning enabled farms with the expectation that they will continue to drive the yield gains necessary to offset high capital costs. Should this assumption prove correct, the rapid growth trajectory of Plenty is astounding.

As Machine Learning Drives Growth, there is Plenty to Consider 

  1. Understand Target Consumer Market

As Plenty grows, it will need to appropriately communicate its value proposition. Consumers are increasingly skeptical of scientific changes in agricultural practices. Therefore, educating the consumer on the benefits of Plenty’s machine learning farm practices will be critical in generating demand for Plenty’s products. Just as Plenty created experience centers to promote transparency and trust amongst locals [8], more of these types of programs should be launched to accelerate awareness. Garnering this ground level momentum will drive change in areas like government where regulatory issues will likely arise.

  1. Narrow Varietal Selection

With the seemingly limitless varietal options that Plenty can test, there’s a temptation to explore. However, given one of Plenty’s challenges is managing capital-intensive operations while meeting consumer demands for low cost produce [10], Plenty needs to narrow its varietal selection to seek the highest performing crops. By focusing on select plants, Plenty can collect even more data that will hone and further increase the efficiency yield of these crops through machine learning.

World Impact

Many investors see Plenty’s potential as revolutionary. Will Plenty’s approach to indoor farming lead to a fundamental shift in food production? If this is the future of agriculture, how does this affect the labor force in farming? Should this be a concern?





[1] Scheer, Roddy & Moss, Doug. “Dirt Poor: Have Fruits and Vegetables Become Less Nutritious?” Scientific American,, accessed November 2018.

[2] De Clercq, Matthieu, Anshu, Vats, & Biel, Alvaro. “Agriculture 4.0: The Future of Farming Technology.” World Government Summit, February 2018,, accessed November 2018.

[3] “How Indoor Farming Will Change the Way We Eat.” Plenty Blog, 16 November 2018,, accessed November 2018.

[4] Baker, David. “Vertical farms have nailed leafy greens. Next up: tasty fruit.” Wired, 19 March 2018,, accessed November 2018.

[5] Fortune Staff. “7 Ways AI Is Changing How You Shop, Eat, and Live.” Fortune, 22 October 2018,, accessed November 2018.

[6] Soper, Taylor. “Jeff Bezos-backed indoor farming startup Plenty opens 100K square-foot facility in Seattle region.” GeekWire, 3 November 2017,, accessed November 2018.

[7] Ohnsman, Alan. “Veteran Tesla Engineer Leaving For Greener Pastures: AgTech Startup Plenty.” Forbes, 7 June 2018,, accessed November 2018.

[8] Reuters. “California Indoor Farming Startup Plenty Plans to Launch Global Operations in China and Japan.” Fortune, 17 January 2018,, accessed November 2018.

[9] Wang, Selina. “This High-Tech Vertical Farm Promises Whole Foods Quality at Walmart Prices.” Bloomberg Businessweek, 6 September 2017,, accessed November 2018.

[10] Bunge, Jacob & Brown, Eliot. “Indoor Farming Takes Root at California Startup.” The Wall Street Journal, 13 February 2017,, accessed November 2018.




Chasing Glory: How Ferrari S.p.A. is embracing Additive Manufacturing to Win Again


That Friend Who Always Knows What to Say: How Google is Helping Brands Talk to Consumers

Student comments on Machine Learning and the Future of Agriculture

  1. Nice to learn about this new technology in the agricultural sector. Your point about the capital allocation is very interesting but I don’t think that plenty should focus on the highest performing crops but on the one with the highest market potential.
    For the moment it seems that they are building indoor farms and collecting and analysing data from those. But at what point they will have to convince farmers of the benefits of the product and as seen in the Indigo Agriculture case, farmers are really reluctant to change. How will they change the farmers mentality and is it not better to make them part of the process and to develop the system with them on real outdoor farms and not controlled indoor farms.

  2. The idea of using machine learning in agriculture is really interesting and solves a very critical social challenge in the world today. While I agree that the use of machine learning will fundamentally shift the way we farm; however, to answer your question as to its potential impact, I wonder if there’s a ceiling to the gains to be made. What I mean is that once ideal growing conditions for a particular crop have been found, will there come a point wherein conditions can no longer be optimized? If the world continues to suffer from a hunger crisis, Plenty may have to look at other ways to increase yield beyond growing conditions. Another potential use of this data is to use it as additional input into agricultural genetic modification. Understanding ideal growing conditions can help determine what genes may need to be modified for a crop to grow exponentially. I do not believe that Plenty will solve this problem, but it could expand its revenue stream by selling its data to companies that are tackling these issues on a daily basis. While the ethics of this continue to be debated, in a scenario of world food shortage, this may be an interesting option to explore.

Leave a comment