The U.S. Department of Defense is a massive organization. Not only does it employ millions of military personnel and hundreds of thousands of civilians, it also has the world's largest military budget, with more equipment purchased and maintained than can be contained in a single paragraph.
There's a lot to coordinate. Operators across the Defense Department's various agencies must decide how to plan operations, coordinate resources and stay within budget for potentially contested events, like hurricanes or hostile forces.
Two years after it was founded, Virginia-based startup Defcon AI has raised a $44 million seed round to solve this intractable problem.
Consider Air Transport Command, the headquarters of the U.S. Air Force. When operators plan an airlift, they have to consider a variety of variables, including available aircraft, the number of crew members needed, where the crew rest, where they refuel, the associated airfields, cargo handling locations, and more. Defcon AI has developed a set of software that allows operators to set these parameters on the front end and “let the software do the rest,” Defcon co-founder and chief strategy officer Paul Selva, a former U.S. Air Force general, told TechCrunch. The software essentially operates based on these parameters and inputs to create an optimal plan, including cost tables, resource requirements, and schedules.
Planning like this is difficult enough in the best of circumstances, but in times of crisis, defense operators don't have a day to spare to allocate resources. That's where Defcon AI comes in.
“All the work that we've done that's really had an impact on has been done by me,” Selva said. During his long military career, Selva held many positions, including commander of Air Mobility Command, which oversees nearly all of the Air Force's airlift fleet. He later became commander of U.S. Transportation Command, which coordinates transportation missions around the world, including transportation by ship, truck, train and other modes of transportation. Before retiring in 2019, he was nominated by President Barack Obama to be vice chairman of the Joint Chiefs of Staff.
He co-founded Defcon in 2022 with Redcell Partners co-founders Yishroel Bloomer and Grant Verstandig. Redcell has an interesting business model: the company invests internally, but also incubates companies (including Defcon), often finding promising entrepreneurs to lead them. Entrepreneurs sometimes approach Redcell before founding their company, and the company handles board formation, legal, HR, finance, and other aspects while the company grows.
In the case of DEFCON, Silva says the company was founded “because Air Mobility Command identified a mission need that wasn't being met by industry.” The trio “talked about whether this was a tractable problem, and… our intuition was that it was a tractable problem mathematically and software-wise, but that it needed to be addressed in a different way.”
Both Bloomer and Verstandig have impressive resumes. Prior to joining Red Cell, Bloomer served as acting director of OSD/Cape (Office of Cost and Program Evaluation) at the Department of Defense, a big role he said essentially functioned as the Pentagon's “chief analyst” and oversaw the budget submission process. Verstandig is an entrepreneur who has helped incubate and grow businesses such as Rally Health and defense startup Epirus through Red Cell.
Defcon AI targets problems of “maximum complexity,” Bloomer said. The startup's system combines a variety of algorithms, including machine learning and mathematical optimization algorithms, to simulate specific scenarios and generate the best logistics outcome to meet them. In the early stages of product development, Defcon used a data-free reinforcement learning algorithm, but the company says it now increasingly incorporates data provided by the Department of Defense to enhance the software. Operators can choose whether to have the system simulate ways an enemy might disrupt operations, or even tell it to optimize different variables, such as speed and cost-effectiveness.
Just two years after founding, the company has won nearly $15 million in government contracts and delivered a production version deployed in operational operations by Air Mobility Command. The company is currently in the process of certifying its software to handle classified and secret information, expanding its use with the Department of Defense to incorporate more data, and expanding its planning and simulation software to include trucks, trains, and ships.
Defcon has no plans to slow down. The company sees more uses across the Defense Department where its software could make an operational difference, and Bloomer said it's seeing “very strong signs of demand” for its products from the private sector. Overall, the company says working closely with end users results in better products, creating a true competitive edge in adversarial situations.
“Operational planners are really trying to assess risk for their commanders,” Selva said, “and they're probably the people who are the most skeptical of decision support tools. So the more we can work with them, the better the results we can achieve.”