The problem Perseus Defense is trying to solve fits in one uncomfortable sentence: a hobby drone that costs $500 can now threaten a tank, a base, or a convoy worth millions. For most of the last century, air defense assumed the thing overhead was expensive too - a jet, a helicopter, a cruise missile. Cheap, swarming quadcopters break that assumption, and they break it in favor of whoever is willing to lose a lot of small machines.
Perseus Defense, a Y Combinator Summer 2025 company, is a bet that the counter can be just as cheap. The company builds guided micro-missiles - small enough to carry, made to be manufactured in volume - that detect and intercept small drones for a target price under $10,000 a shot. Compared with many existing interceptors that run past $100,000, that is roughly a 20x cut. In a war of attrition, the side with the cheaper shot usually wins, and Perseus has organized the entire company around being that side.
The companyShrinking a missile until it is affordable
The flagship product is the Harpe interceptor, a guided micro-missile roughly 24 inches long with a range of about two kilometers. Co-founder Steve Messinger has described the design idea plainly: take something like a Sidewinder missile and shrink it down to a miniature scale that can be built in high volume at very low cost. The missile rides on the Perseus Launcher, a networked pod that weighs under 80 pounds, sets up in under a minute, and can run for a 24-hour watch. It is meant to bolt onto a truck, sit on a boat, or hold down a fixed installation.
The system is built to handle Group 1 and Group 2 drones - the small and medium classes that have become the signature weapon of recent conflicts. Detection, tracking, and engagement are bundled into one modular package, so a single operator is not stitching together a radar from one vendor and an effector from another.
Cost per intercept · the asymmetry Perseus targets
Relative cost per shot. The point is not that Perseus is cheap in absolute terms - it is that the gap between attacker and defender shrinks by an order of magnitude. Figures are approximate, drawn from public statements.
"We're coming in at less than five grand, and we might be able to get that down even more."Steve Messinger, Co-Founder & CTO
The foundersFrom a Saturn helicopter to a quadcopter killer
Chief executive Jason Cornelius spent his earlier career at NASA, where he worked on the Titan Dragonfly - a rotorcraft designed to fly in the atmosphere of Saturn's largest moon - along with next-generation Mars helicopter work and autonomous flight systems. He holds a PhD in aerospace engineering and has taught aerodynamics at Stanford. His co-founder and chief technology officer, Steve Messinger, came from Boeing, where he led autonomy work and cut his teeth on swarm robotics and drone competitions.
That pairing matters. One founder knows how to make small aircraft fly precisely in punishing conditions; the other knows how swarms of drones actually behave when they are trying not to be caught. According to the company, they had a functional prototype together within roughly four days of starting. In June 2025 they built their first two spin-stabilized guided rockets, the Mk-I and Mk-II, and the same month they won the YC S25 batch preview day.
Founder toolkit
| Founder | Prior work | Edge it brings |
|---|---|---|
| Jason Cornelius | NASA - Titan Dragonfly, Mars rotorcraft | Precision flight, guidance, aero design |
| Steve Messinger | Boeing - autonomy lead, swarm robotics | Autonomy, target behavior, tracking |
Two aerospace backgrounds pointed at opposite ends of the same problem: flying the interceptor and predicting the target.
The customerBuilt from the truck bed up
Perseus is selling to U.S. government buyers - the Department of Defense and the Department of Homeland Security - along with the defense integrators who bundle systems for them. Before writing much code, the team ran more than 60 discovery calls with people in that world. That shows up in the design constraints: under 80 pounds because someone has to carry it, 60-second setup because the situation will not wait, 24-hour operation because threats do not keep office hours.
The system has already been in front of soldiers. During the Army's Golden Shield exercise, elements of the 1st Cavalry Division tested the Harpe receiving external radar cues to engage drone targets, including multi-launch capability and fully active guidance. One Pentagon official, quoted in press coverage, put the ambition bluntly.
"If you build this, it will be mandatory equipment for every truck, boat, and convoy in the U.S. military."Pentagon official, via press reports
The marketWhere Perseus sits among the alternatives
Counter-drone defense is a crowded field, but the incumbents each carry a catch. Jammers and electronic warfare fail against drones that fly autonomously without a radio link. Guns are cheap per round but hard to aim at a small, fast, maneuvering target. Directed-energy weapons are promising but still bulky and power-hungry. Traditional missiles work well and cost a fortune. Perseus is aiming at the open lane: a guided hit-to-kill effector that is far more accurate than a gun and far cheaper than a conventional missile.
| Approach | Strength | Weakness |
|---|---|---|
| Jamming / EW | Low cost per use | Beaten by autonomous drones |
| Guns | Cheap ammo | Hard to hit small movers |
| Directed energy | Deep magazine | Bulky, power-hungry |
| Legacy missiles | Reliable kill | $100K+ per shot |
| Perseus Harpe | Guided, low cost | Early-stage, unproven at scale |
The framing the press reached for is "Iron Dome for drones," and it is a useful shorthand. The sharper insight underneath it is about economics rather than hardware. Israel's Iron Dome is celebrated but expensive per intercept, which is fine against rockets and untenable against thousands of cheap drones. Perseus's whole thesis is that if you drive the cost of a single intercept down far enough, defense stops being a losing trade.
The businessVolume is the strategy, not the byproduct
The business model is straightforward defense manufacturing: sell interceptors and launchers to government customers, and make money on units. What is less common is that Perseus started with mass production as the design constraint. A missile that performs beautifully but costs six figures does not solve the drone problem; a missile that is merely good but can be stamped out by the tens of thousands does. Every choice - the spin-stabilized airframe, the shrunken form factor, the modular launcher - reads as an answer to the question "how do we make a lot of these cheaply."
The company left San Francisco and set up headquarters in Buda, Texas, closer to the manufacturing and open-range testing that a hardware defense startup needs. In September 2025 it closed a $6M seed round led by Bow Capital, with Rebel Fund, Pioneer Fund, Alumni Ventures, Multimodal Ventures, Lobster Capital and YC co-founder Paul Graham among more than a dozen backers. By January 2026, the team reported achieving fully guided flight for a direct hit-to-kill intercept.
None of this makes Perseus a finished company. It is early, the hardest scaling questions are still ahead, and a live-fire demo is not the same as a contract to arm every convoy. But the wager is clear and unusually legible: drones changed the arithmetic of who can threaten what, and Perseus is trying to change it back one cheap, guided shot at a time.