Non-traditional contractor Torch.AI seeks increased defense work

By Marjorie Censer / March 18, 2021 at 3:42 PM

Artificial intelligence firm Torch.AI, buoyed by a new $30 million investment, is pursuing additional defense work, particularly in what the company's chief executive describes as areas where it can "showcase" its capabilities.

The Leawood, KS-based company specializes in using machine learning to enable high-performance data processing. Brian Weaver, the company's founder and CEO, said the company has already been working for several years on background investigations for the Defense Counterintelligence and Security Agency.

Torch.AI has partnered with Perspecta on the services piece of the work, he said, but the government has also directly acquired licenses for Torch.AI's platform.

"We've opted to partner with systems integrators and value-added resellers . . . as often as we can because we feel like that's a better, healthier strategy," Weaver said.

This week, Torch.AI announced it has raised $30 million in Series A funding to scale its Nexus AI platform. The funding was led by WestCap Group, a San Francisco-based investment firm.

Last year, the company announced it was growing its DC-area presence. The company added Marion Kennedy, an executive who has worked for intelligence contractors, to lead Torch.AI's expansion among defense, intelligence and civilian agencies and tapped Joe Imorde, a former Army intelligence officer and industry executive, to focus on growth with combatant commands, the Army and defense intelligence missions.

Now, Weaver said, Torch.AI is looking for work where the company can "showcase" its capabilities. He said the company is being "strategic" about the kind of work it pursues.

"We want to deal with really, really big problems, but we don't necessarily need them to be some huge financial win," Weaver said.

He told Inside Defense Torch.AI is particularly interested in the sensor space.

"If you think about sensors, and [internet of things] devices in general, the information is largely unstructured -- many, many classes of information that all need to be synthesized into a common operational picture," Weaver said. "We're very keenly involved in several sensor programs and very interested in that particular space."