A $7.6m contract argues the Army's power problem is routing, not generation

The contract is small next to the money already inside the company. Chariot has raised $41 million since being founded in August 2024 ^[14], so the $7.6 million from DIU ^[1] is around 19 percent of its funding to date ^[19]. What that buys is not revenue at scale. It is access to hardware no lab can stand in for: whatever tactical generators, solar arrays and vehicles a particular brigade happens to own ^[4], on a 12-month delivery clock ^[3].

Read the fuel claim against the mechanism the company describes and it resolves into a statement about run time rather than burn rate. Amphora is presented as a power management computer rather than a conventional battery pack, moving energy in both directions and routing it to where it is needed ^[5]; the route to lower consumption is reducing how much generators are needed, with stored energy and other sources carrying the load under software control ^[22]. So the 60 percent depends almost entirely on the shape of demand. A unit whose generators are oversized for its average load, and whose peaks are short, has a lot of idle capacity to reclaim. A unit running close to capacity has very little. "Up to 60 percent in combat scenarios" ^[7] stacks two qualifiers on one figure and states no baseline, and the 36-hour silent operation number ^[8] names no load it is silent under.

The less advertised part of the package may outlast the fuel number. Amphora collects energy data and feeds it to command and control networks ^[9]. Metering changes the character of the resource: power stops being something a unit finds out about when a generator strains and becomes a quantity that gets reported and planned against. It also makes the vendor's own claim checkable by the customer, which is a different kind of scrutiny than a datasheet gets.

Adam Warmoth's framing is that power availability decides what equipment troops bring to the fight ^[10], and that a shortage forces a unit to choose which systems stay online ^[11]. Follow that logic and the control layer is the thing making the choice, or at least the thing the choice is made through. Chariot builds neither drones nor electronic warfare systems ^[17], which keeps it out of competition with the loads whose demand it arbitrates ^[12], and Warmoth came from Anduril, where he worked on scaling counter-drone programs ^[16].

What PRISM tests, then, is narrower than the marketing suggests and harder. The question is whether a single management architecture holds across a brigade's existing inventory ^[18][6], not whether the software works on a bench. Generators, solar arrays and vehicles bought over years will not present themselves identically, and the failure mode here is a compatibility list rather than a broken box. Fuel deliveries, heat and noise are the standing costs of the generator-first arrangement ^[13]; a routing layer only touches them to the extent it can talk to everything already on the ground.