Using data science to grow leafy greens

They call themselves “plant whisperers,” but they don’t tenderly bend over kale and chervil. They focus on data science and automate operations.

Close your eyes and think organic leafy greens and herbs. Do you see artificially lit warehouses or shipping containers where computer-driven systems mist the bare roots of plants, stacked high on walls, with just the right mix of nutrients and micronutrients at scientifically calibrated intervals?

Probably not. But you may have eaten greens produced that way, at home or at a high-end restaurant.

Indoor vertical farms have sprung up around the world to produce locally grown produce. While producing only a small part of the global food supply today, vertical farms may be key to doubling the world’s supply of food by 2050, which the United Nations says will be needed as population grows. The farms require far less land and water than industrial farming does, and don’t use pesticides.

David Rosenberg, Chief Executive Officer of AeroFarms, the world’s largest indoor vertical aeroponic farming company, gave his view of the future of farming at J.P. Morgan’s Advisor Exchange in Paris in November. 

“About 70% of the world’s fresh water goes to agriculture globally, and about 70% of fresh water contamination comes from agriculture,” Rosenberg said. AeroFarms produces greens using 95% less water than a conventional farm, he said. They are available, under the Dream Greens label, at Whole Foods, ShopRite and from FreshDirect.

The scientific challenge

The firm’s water savings come from its aeroponic technology, which sprays the bare roots of plants with only as much water and nutrients as the plant needs; between sprayings, the roots are exposed to air, which improves their oxygen intake. (Hydroponic farms, by contrast, immerse the roots in a water solution; geoponic farms grow plants the old-fashioned way, in soil.) Thousands of sensors track whether the plants are getting what they need.

“We are trying to understand what an arugula wants, what a kale wants. So, we call ourselves plant whisperers,” Rosenberg said. “We try to pay attention to what the plants are telling us and how to adjust to give the plant what it wants, working with engineering and operations.”

Such systematic study is easier in the controlled environment of an indoor farm than for plants in the field—where there are many uncontrolled variables and scientists can’t isolate variables—than in the controlled environment of an indoor farm.

Environmental health scientist Dickson Despommier first proposed vertical farming as way to make agriculture sustainable in 1999, but for several years, many experts said the need for artificial lighting would make it uneconomic. 

In researching how to solve that problem, Rosenberg read about “Haitz’s Law.” Roland Haitz, a scientist at Aquilent, had observed in 2000 that light-emitting diodes (LEDs) were becoming exponentially more efficient (producing increasingly more units of light per unit of electricity); Haitz said they would continue to become more efficient, much as Moore’s Law said of computers. AeroFarms invested in LEDs to drive down its largest capital cost (for LEDs) and its largest operating cost (for electricity to power them). The result: “We have proven the unit economics in leafy greens,” Rosenberg said.

The firm is expanding into berries and other produce, while seeking to drive down costs further through automation and improved data management. AeroFarms has a partnership with Dell, which owns EMC Pivotal VMWare, to manage the data gathered by the sensors in its indoor farms. Software programmers make sure the data goes immediately to food safety, operations or R&D, and later to finance, sales and marketing.

“In many ways we view ourselves as a data science company as much as an operating and a technology company,” he explained. Or as plant whisperers.