This Atlanta Startup Just Raised $350M to Build Autonomous Hypersonic Warplanes. Also, a Nicer Way to Get to Paris.

There is a startup in Atlanta — not San Francisco, Atlanta — that just raised $350 million to build autonomous hypersonic fighter aircraft. Khosla Ventures led the round. The company is called Hermeus. Their test program is called Quarterhorse. Their military drone is called Darkhorse. Their civilian passenger jet — yes, they also have a civilian passenger jet — is called Halcyon.

I have been covering tech startups for long enough to have developed a high tolerance for ambitious branding. But "Darkhorse to Halcyon" is a particular kind of poetry. One product is designed to outmaneuver adversaries at Mach 5. The other is designed to get you to London in time for dinner. Both are coming from the same team, in the same office, presumably with the same coffee maker.

The Quarterhorse Saga: A Brief History of Going Very Fast

To understand what $350 million is buying, you need to understand Hermeus's test program. They've been building a sequence of increasingly fast unmanned aircraft called Quarterhorses — each one faster than the last, like an engineering team that read a speedrunning tutorial and decided to apply it to aerospace.

• Mk 1: Subsonic. Unveiled March 2024. Went from requirements-setting to rollout in roughly 200 days, which for aerospace is approximately the speed of light.
• Mk 2.2: Supersonic. Recently completed two successful flight demonstrations. This is the one that just secured the $350M round.
• Mk 3: Targeting Mach 3.3 — faster than the SR-71 Blackbird, which spent the Cold War being the coolest thing in the sky and has not been surpassed in civilian imagination since.
• Eventual goal: Mach 5+. The actual hypersonic threshold. The place where conventional materials fail and the engineering problems become genuinely exotic.

CEO AJ Piplica, describing the pace of development, said: "It's a pace of development we haven't seen in this country for a very long time." This is a statement that is simultaneously true, impressive, and exactly the kind of thing a startup CEO says right before a $350 million raise. I am not criticizing it. I am simply noting it.

The VC Thesis, Which Is Actually Coherent

Here is the thing about defense tech startups that I find genuinely interesting, as opposed to just satirically interesting: the underlying argument is not stupid. Traditional defense contractors — your Lockheeds, your Northrops, your Raytheons — are optimized for cost-plus contracts and 30-year procurement timelines. They are not optimized for moving fast. Startups, in theory, are.

The Pentagon has noticed. The Defense Innovation Unit, which exists specifically to route commercial technology into defense applications, has already put money into Hermeus. The Air Force has contracts with them. Total government investment so far: over $60 million. Which means the $350M Series C is not just VCs betting on the vibe — it's VCs betting alongside an institution that has been writing checks and watching flight tests.

We've spent a lot of time recently asking whether AI agents actually make money, or whether the whole thing is an elaborate confidence exercise. Defense tech is a different animal. The customer has a defense budget. The defense budget is not going away. The only question is whether a startup can build what the customer needs before the customer loses patience and goes back to Lockheed. Hermeus is betting yes.

“A Dozen or So” — The Manufacturing Cadence

I want to dwell on one detail, because it is doing a lot of work.

Hermeus currently has the manufacturing capacity to produce, at their Atlanta facility, approximately “a dozen or so” Darkhorse drones per year.

A dozen or so.

This is not a criticism! Every production program starts somewhere, and the F-35 — which the United States has been building since roughly the Clinton administration and has cost approximately $400 billion — is not exactly a triumph of manufacturing efficiency either. But there is something specifically delightful about a company that has just closed the largest single funding round of its existence and describes its output capacity with the same phrase your aunt uses when talking about how many Christmas cookies she plans to bake.

“A dozen or so. Maybe more if we get the Jacksonville facility running on time.”

The Jacksonville facility, to be clear, is real. Hermeus broke ground on a hypersonic engine and flight test facility at Cecil Airport. It is called the HEAT facility, which is either a very deliberate pun about hypersonic aerodynamic heating or a remarkable coincidence. Either way, I respect it.

The Dual-Use Question, Which Nobody Wants to Answer

There is a version of the “defense-adjacent tech” conversation that Silicon Valley has been having with itself for about five years now, and it goes roughly like this: Is it okay to build things that could be used to hurt people, as long as they could theoretically also be used to help people? Hermeus has operationalized this question into two separate product lines. Darkhorse is for the Department of Defense. Halcyon is for the person who wants to get to London in 90 minutes and doesn’t want to think too hard about where the underlying technology came from.

I am not here to adjudicate this. The defense tech conversation is long and complicated and involves a lot of people with more clearances than I have. What I will say is that the branding — Darkhorse for the warplane, Halcyon for the passenger jet — is doing something very specific. “Halcyon” means peaceful, calm, golden. It’s the word you’d use to describe a simpler time. It is not the word you’d use to describe a Mach 5 military drone program.

Someone in the Hermeus marketing department understood the assignment.

What Hypersonic Actually Means, For the People in the Back

Because I know some of you just registered “hypersonic” as “very fast” and moved on: Mach 5 is approximately 3,800 miles per hour. At that speed, the air friction generates temperatures that will destroy aluminum, titanium, and most conventional materials. You need specialized thermal protection systems just to keep the aircraft from disintegrating. The SR-71 Blackbird, which could hit Mach 3.2 and was operated by the CIA, leaked fuel on the ground because its fuel tanks only sealed properly at operating temperature — that is, after the aircraft had already been flying at supersonic speeds for a while.

Hermeus wants to go faster than that. With a hybrid turbine-ramjet engine. Built by a startup. Funded by venture capital. In Atlanta.

I find this genuinely exciting in a way that I find, say, smart glasses startups do not. Smart glasses are trying to solve a problem you didn’t know you had. Hypersonic fighters are trying to solve a problem that the United States Air Force has been very specific about having. The market pull is real. The technical challenge is also real. The gap between $350 million and “we build hypersonic aircraft at scale” is significant — but it is a gap that has a plausible path through it, which is more than you can say for a lot of the things getting funded right now.

The Closing Question Nobody Wants to Ask Out Loud

Here is where I land on Hermeus: it is one of the more legitimately interesting defense startups to get a major funding round in a while. The test program is real. The flights happened. The government is already paying them. The technology — the hybrid turbine-ramjet engine — is genuinely novel. The 275-person team in Atlanta has more aerospace credentials per capita than most companies that describe themselves as aerospace companies.

The question I keep coming back to is not “can they build it?” They have, demonstrably, been building it. The question is “can they build it at scale, fast enough to matter, for a price the Pentagon will actually pay, while also somehow developing a civilian supersonic passenger jet?” That is four questions that each contain several other questions.

AJ Piplica says they’re moving at a pace we haven’t seen in this country for a very long time. He’s right. Whether that pace is fast enough — for the technology, the budget, the geopolitical moment, and the entirely separate civilian aviation market they’ve also committed to addressing — is a question the next few Quarterhorse flights will start to answer.

I’ll be watching from a safe distance. Specifically, the distance required when something is traveling at Mach 3.3 and you’d prefer your hair to remain attached to your head.

Godspeed, Atlanta. Literally.