Los Alamos National Lab spinoff Descartes Labs just raised $20M in Series B, bringing its total $58M. Also, DARPA is one of the organizations backing the startup, having provided $7.2M in 2018. The startup refers to itself as a data refinery, collecting geospatial data from a variety of sources including satellites, and using the data to build AI models that can help identify issues around the world like crop disturbances, food shortages, oil disruptions, constructions starts, and more.
Working with imagery from satellites, probes, and other sources generate petabytes of data. In order to deal with the huge volume of data, the Descartes team took their experience in building a supercomputer at Los Alamos, and built one in the public cloud. The result, in 2019 the team developed a supercomputer from AWS spot resources that was able to perform at 1.926 petaflops (Linpack Benchmark), placing them at #136 on the Top500 List for supercomputers. The cost of the building and running that system was only $5,000, charged to the company credit card at that time. It doesn’t say for how long the system ran for but it’s a testament to the power of cloud computing
Uses Cases
- Construction Starts: Created a proprietary model to track constructions starts using synthetic aperture radar, which can be used to monitor infrastructure growth
- Crop Classification: Developed a model for surveying field boundaries for land with crops, in order to classify the type of crop being grown within each field
- Wind Turbine Detection: Created a model to identify all the physical wind turbines across the world
- Methane Monitoring: Developed a system to monitor methane emission in New Mexico
Profile
- Company: Descartes Labs
- Founded: 2014
- HQ: Santa Fe, New Mexico
- # of Employees: 113
- Funding: $58.3M
- Founders: Mark Johnson (CEO) and Mike Warren (CTO)
- Industry: Computer vision
- Customers: DARPA, governments, academia, and more
