The big money moves just keep on coming. This morning, edge computing company Armada announced a $131M strategic funding round, which it will use to build a new “megawatt-scale” modular data center called Leviathan.
New investors Pinegrove, Veriten, and Glade Brook joined the round, alongside returning backers Founders Fund, Lux Capital, Shield Capital, 8090 Industries, M12 (Microsoft’s venture arm), Overmatch, Silent Ventures, Felicis, and Marlinspike. This brings Armada’s total raised to over $200M, according to CEO Dan Wright.
On the edge: In case you didn’t know, AI requires, like, a heck-ton of compute. Models produced by companies like OpenAI and Anthropic are propped up by enormous, power-hungry data centers. (Remember Stargate?)
Most models today run inference via the cloud: compute happens centrally, and the results are sent back to users over the internet. But in austere environments—without stable connectivity—this model falls apart.
Out of the box: Armada was founded in 2022 to solve this problem. Co-founders Dan Wright and Jon Runyan set out to find a way to bring AI, compute, and connectivity to the world’s harshest, most out-there environments, including disaster areas and war zones.
The company produces modular, ruggedized “data center[s] in a box” called Galleon (think a data center in a shipping container), along with a platform for connected assets (like Starlink and sensors) and a marketplace for edge computing applications, software, and hardware.
Armada has partnered with Starlink for comms, and in May, the company inked another partnership with Microsoft and Second Front to build AI-driven tools at the edge. The company is truly dual-use: their products, they say, are as appropriate for the frontline as they are for an oil rig.
Galleon—the company’s flagship product—now comes in four different sizes:
- Beacon: A briefcase-sized unit.
- Cruiser: A 20’ container with “3 racks of compute,” designed to be “compact [and] agile,” designed for environments where space is an issue.
- Triton: A 40’ container containing “5 racks of compute” designed for the “heaviest workloads,” according to the company.
- Leviathan: The newest product, a larger-scale data center made up of two 45’ containers (one for compute and one for power distribution) and a 20’ container for cooling.
Wright told Tectonic and other reporters on a press call that Leviathan will be ten times more powerful than Triton. He did not reveal a price point for the unit, but said it was a “small fraction of a traditional data center.”
Big boys: When Leviathan gets deployed, said Wright, the major difference is that users won’t only be able to run AI models at the edge—they’ll be able to train them, too. Wright said that customers can use this to “create AI factories” in a much faster and cheaper way than building a mega data center. Leviathan can be easily upgraded with advanced chips and cooling capacity, Wright said.
They’re also teaming up with energy companies throughout the country—from North Dakota to West Virginia—who are trying to stand up AI infrastructure. Leviathan is energy-agnostic and can run on everything from fuel to nuclear to solar.
“You can park these Leviathans at those energy sources, and you have a plug-and-play AI cluster,” Wright said.
Far out: Armada’s units are already being tested by the DoD. In February, the company announced that it’d delivered a Galleon unit to the US Navy (specifically, the Naval Information Warfare Center (NIWC) Atlantic) as part of a cooperative research and development agreement looking at “advanced network management, command and control, and edge computing solutions.”
According to Wright, the units are being used to “[process] data from autonomous technologies, drones, et cetera, at the edge in remote, hostile, disconnected environments.”
Wright told reporters that Galleon units (including Triton) had already been tested in a “variety of hostile environments.” However, when asked if Leviathan had been field-tested, Wright said that the product was still in development.
“We’ve done a lot of testing for Leviathan, and we also have been working very collaboratively with…partners to make sure it fits the event,” he said.
Big brain: Armada is taking a pretty ambitious view of where its edge compute platforms are headed. As AI becomes all the more powerful—and all the more central to warfare—they argue, the more data centers will be needed. And a lot of the places where computing will be needed (on the frontline, for example) don’t have access to traditional data centers.
Plus, the more distributed a computing network is, Wright said, the harder it is to take out.
“Because it is distributed, you can’t just attack one brain in order to destroy it,” he told reporters, “You’d have to destroy all of the different brains, and they’re distributed all over the world.”