Descartes Labs, an imagery data refinery platform, announced it has raised an oversubscribed $30 million Series B led by California-based venture capital firm March Capital, and including existing customer, Cargill, and existing investors, Crosslink Capital, and Cultivian Sandbox.
This funding reflects the growing size of rounds being raised in the agricultural AI and data analytics space as the industry matures. In 2014 Descartes Labs raised a $3 million Seed Round led by Crosslink, followed by a $5 million Series A in 2015 led by Cultivian.
Headquartered in Santa Fe, New Mexico, Descartes Labs was spun out from the Los Alamos National Laboratory and co-founded by Mark Johnson, current CEO; Mike Warren, current CTO; Adam Smith, current head of business sales; Mark M. Mathis, thaumaturge; Rick Chartrand, mathematician; Tim Kelton, cloud architect; and Steven Brumby, chief science advisor.
Thanks to the greater availability to cloud-based supercomputing infrastructure in the cloud, Descartes Labs created a cloud-based supercomputer that can process mega-volumes of data – enabling the startup to use 30,000 cores to process a full petabyte of satellite imagery in just 16 hours. Likewise, the use of satellites, which were once relegated to only government objectives, has become much more mainstream and cost effective for use by the private sector. Other simultaneous advances in the areas of AI, sensors, video, and radar have driven the growth and greatly expanded the abilities of imagery analysis in association with agricultural production.
Where Descartes Labs differs from rivals in its field, is in the company’s ability to take massive amounts of imagery data and refine it into a useful tool. Every day the startup gathers 5 terrabytes of satellite imagery, but then filters out inconsistencies, selects the highest quality images, accounts for variations due to the atmosphere, and pieces together a single data set. In addition, the startup gathers data from multiple satellite sources that may have better or updated images and integrates that information to create the best possible final product for customers.
“In working on customer problems, one of the things we’ve uncovered is that it’s rarely a single satellite or dataset alone that will solve the problem,” stated Descartes in a company statement on Medium announcing the close of its Series B. “Only by turning different datasets into a fused sensor, a super-sensor, are we able to solve the problem.”
Johnson told The Verge in August 2016 that the decision by Descartes Labs to focus their technology on the advancement of agriculture was based on numerous factors: climate change and food scarcity; the existing bank of data already compiled by Landsat and MODIS that could be used to train their machine learning models; the availability of historical satellite imagery data that farmers can use for comparison to current findings; and ultimately it was the challenge.
“It’s not like a satellite looking at all the Walmart parking lots and picking out the cars,” said Johnson. “That’s a problem of automation; it’s something a human could do. What we do is an automation of what humans can’t do.”
This high level of precision and refinement has led Descartes disrupting the corn yield prediction process when it began releasing its corn yield estimates prior to the those released by the U.S. Department of Agriculture (USDA) on August 12. In 2015 the company proved more accurate than the USDA by a whole percentage point, and has consistently bested the USDA’s forecasts throughout the growing season, reports The Verge.
To achieve this, Descartes receives data and conducts analysis on every farm in the U.S. every day, and updates its estimates of U.S. corn yields, which drill down to the county-level, every two days, compared to the USDA which updates its forecasts every month, and which only drill down to state-by-state predictions.
The company has begun to apply its technology to crops outside of corn, and has already begun working on soybean imagery, and plans to expand geographically into the major global crop producing regions such as Brazil, China, the Black Sea region, and the EU.
The capital gained through this round will help support Descartes realize its plans to expand its team from the current 40 to possibly 60 by the end of this year, and 100 next year, according to TechCrunch. Descartes also states that it will continue to advance its data refining abilities, and will continue to “solve more wickedly-hard science problems…”
-Lynda Kiernan