Khosla and a16z put $7M into counter-drone startup Mara, at 1/50th their Hermeus cheque

The argument here is arithmetic, not engineering. Co-founder and CEO Daniel Kofman says small drones "distributed the cost and capability of precise attack" while most defences "still concentrate both replacement cost and line of fire," and that Spike answers with a mesh of small, cheap nodes that can each aim and shoot in several directions at once ^[5]. That is a claim about exchange ratios: whoever spends less per engagement can afford to be wrong more often.

What the source does not contain is a price. Mara says the total cost of deploying a complete Spike system comes in under the cost of the vehicle, aircraft or facility it protects ^[19], which is a system-level statement, not a per-shot one. Per-shot is the number that decides a swarm. For scale, Anduril's reusable Roadrunner-M interceptor runs about $200,000 each ^[6], so Mara's entire round equals roughly 35 of them ^[14]. The 250-gram, 200 km/h Seeker is tube-launched with onboard targeting ^[4], which sets an upper bound on what it can plausibly cost, but the company has not published the figure.

The demand side is already booking revenue rather than forecasting it. DroneShield, the only exchange-listed pure-play in the category, reported $125.8 million in the first half of 2026 against full-year guidance of $250 million to $270 million ^[7][8]. Take the midpoint and the implied second half is $134.2 million, only about 6.7 percent above the first ^[15]: guidance that assumes the current run rate holds, not that it accelerates. Fortem Technologies, meanwhile, has raised around $122 million and took $25 million from Lockheed Martin in April 2026 for its DroneHunter net-capture system ^[9]. Strategic money went to capture, not to expendable interceptors.

Khosla's own sizing is the more revealing signal. Its previous defence cheque since April 2026 led a $350 million raise for Hermeus and unmanned high-speed aircraft ^[2], fifty times the Mara round ^[16]. Same lane, two very different risk prices, and the smaller one is priced as an option rather than a build-out.

Which matters because of the timetable. Mara employs about seven people in San Francisco ^[3] and intends to ship ground-mounted units to major customers by the end of 2026, with vehicle-mounted and portable versions in 2027 ^[18]. Seven million dollars across seven staff is roughly $1 million a head ^[17]: hiring and tooling money, not a production line. Field trials to fielded hardware inside a year, at that headcount, is the load-bearing assumption in the whole thesis.

One design detail will draw procurement questions before price does. Spotter can automatically trigger a Seeker launch, while the human operator sets the rules of engagement and keeps the ability to abort ^[10]. That is a defensible split, and it is also the exact seam that customer legal teams will probe.

As for the premise underneath all of it, Mara states that low-cost FPV drones have become the principal cause of combat casualties worldwide ^[20]. That is the company's assertion as reported, with no independent figure attached, and it is doing a lot of work in the pitch. It is also the part investors cannot verify by inspecting the product.