The Hermeus Playbook

COMMANDER · ISSUE 17 · October 12, 2026 · Istanbul

In 2018 a small group of engineers who had spent their careers in the low end of the launch business — building rockets small enough to hang under a jet — decided the harder problem was the airplane, not the rocket. They wanted to build a Mach 5 aircraft: a plane that crosses the Atlantic in ninety minutes, flies five times the speed of sound, and does it again the next day. Every serious attempt at that airplane for the previous sixty years had ended in a museum or a canceled program. Hermeus, founded in Atlanta that year by a team led by AJ Piplica, decided the reason was not physics. It was method. In April 2026 the company raised $350 million at a $1 billion valuation, bringing its total to more than half a billion dollars — for an aircraft that has, so far, only flown supersonic once.

The archetypes in this series won on category, on manufacturing, on capital discipline. Hermeus is the one that took the hardest engineering problem in aviation and refused to solve it in one shot. Its rivals — and the dead programs before them — tried to design the final vehicle and fly it. Hermeus is building a staircase of cheaper, flyable airplanes, each one a little faster than the last, and learning in the air on every step. The company that most wants to build the impossible thing is the one least willing to attempt it directly.

The business behind the hardware. Not what they build, but how they built it.

Hermeus did not start a Mach 5 program. It started a Mach 5 staircase. The first aircraft, Quarterhorse, is itself a ladder: Mk 0, a non-flying prototype that only had to prove the subsystems worked on the ground; Mk 1, the first version that had to leave the runway, which it did at Edwards Air Force Base in May 2025; Mk 2, which flew for the first time in March 2026 and went supersonic that May; and Mk 3, aimed at Mach 3-plus. Each rung is a complete, flyable airplane with one job: retire the next chunk of risk. Only after the staircase does the real product — the Darkhorse military jet and the Halcyon passenger aircraft — get built.

This is the opposite of how hypersonic aircraft have always been attempted. The graveyard is full of programs that tried to design the finished vehicle on paper, spend a decade perfecting it, and fly it once. Hermeus decided that a moonshot is only fundable if you break it into rungs a customer can watch you climb. Each aircraft is a milestone you can fly, film, and invoice against — proof, not a promise.

Takeaway: Don't build the impossible product. Build the cheapest flyable rung toward it, then the next. A moonshot broken into milestones is a fundable company; a moonshot attempted whole is a research paper.

The single most expensive thing in a high-speed aircraft is the engine. Hermeus refused to invent one from scratch. Its Chimera engine is a turbine-based combined cycle — a turbojet for takeoff and low speed, a ramjet for high speed — and the turbojet at its heart is an existing fighter engine, a Pratt & Whitney F100, the same powerplant that flies the F-15 and F-16. Hermeus bolted a precooler and a ramjet flowpath around a proven core instead of spending a decade and a fortune developing propulsion from a blank sheet. The combined-cycle engine ran on a test stand in November 2022.

The discipline here is knowing which problem is yours to solve. Hermeus innovates at the seam — the transition from turbojet to ramjet, the thermal management, the integration — and buys the mature, capital-heavy core from a supplier who has already spent fifty years de-risking it. Reinventing the engine would have been the most impressive thing on the roadmap and the fastest way to run out of money.

Takeaway: Don't reinvent the most expensive component. Integrate the proven part and put your scarce engineering where the actual novelty is — at the seam nobody else has solved.

Hermeus runs the method the launch industry made famous: hardware-rich, test-often. Build real metal fast, fly it, break it, learn from the flight rather than the simulation, and fold the lesson into the next airframe. Quarterhorse went from a non-flying prototype to a supersonic aircraft in a handful of years, across four distinct vehicles, because each one was cheap enough to risk and fast enough to build that flying it was the test plan.

Legacy aerospace measures a high-speed program in decades because it treats every flight as a bet-the-company event, so it flies almost never and simulates endlessly. Hermeus inverts that: cycle time in the air is the whole advantage. An organization that can put a new airplane in the sky every year learns an order of magnitude faster than one that flies once a decade, and in a field this unknown, learning rate is the only moat that compounds.

Takeaway: In an unsolved field, your flight rate is your learning rate. Build hardware cheap enough to risk, and let flying be the test — the team that iterates in the real world outruns the team that perfects on paper.

Hermeus's staircase is not paid for entirely by venture capital. It is co-funded by the customer it is building for. The relationship started small — a $1.5 million Air Force contract in 2020 to study a high-Mach aircraft, including a presidential-transport application — and grew into roughly $60 million by 2021 from the Air Force and investors together to fund the Quarterhorse flight-test program. The military wants a fast, uncrewed aircraft; Hermeus wants to climb its staircase. The same test flights serve both.

This is the quiet financial engine under the whole company. Government development contracts pay Hermeus to build the exact testbeds that also advance its commercial ambition, which means the hardest, most expensive phase — proving the technology in flight — is partly de-risked by someone else's budget. Find the customer who will pay you to build your own future, and every dollar does double duty.

Takeaway: The best R&D money is a customer's. Structure the near-term contract so the thing you build to satisfy it is the thing you needed to build anyway.

Hermeus is really two companies stacked on one technology. Darkhorse is the near-term one: an uncrewed, mass-producible military aircraft, the mission the Air Force will pay for first. Halcyon is the dream: a passenger aircraft that flies civilians at Mach 5. The military program is not a distraction from the commercial vision — it is how the commercial vision gets funded, because the defense mission has a paying customer today and the passenger jet does not.

The sequencing is the strategy. A pure commercial supersonic play has to raise against a market that won't exist for fifteen years. By putting the defense aircraft first, Hermeus gives investors a revenue path measured in years, not decades, while the underlying engine, airframe, and thermal work carry straight over to the passenger plane. The near-term mission bankrolls the long-term one on the same core technology.

Takeaway: When your dream market is a decade out, find the adjacent mission that pays now and runs on the same technology. Monetize the near term to fund the moonshot — don't ask investors to wait fifteen years on faith.

The April 2026 Series C — $350 million, structured as roughly $200 million of equity and $150 million of debt, led by Khosla Ventures at a $1 billion valuation — was explicitly a manufacturing raise, not a runway extension. The stated purpose was to build a fleet of aircraft at once and stand up production, moving Hermeus from a company that builds one prototype at a time to one that builds several in parallel. The Atlanta site shifts toward production; a new El Segundo headquarters puts the company inside the densest aerospace talent and supplier cluster in the country.

The logic tracks the moment. At the transition from prototype to product, capital buys parallelism: three airplanes built simultaneously generate three times the flight data in the same calendar, and in a business where learning rate is the moat, compressing time is the entire point of the money. Raising to build many rather than to survive longer is a bet that the constraint is now flight hours, not cash.

Takeaway: At the prototype-to-production seam, raise to build in parallel, not to extend runway. When your moat is learning rate, capital's job is to buy calendar — many units at once, not one unit for longer.

Don't attempt the impossible product — build the cheapest flyable rung toward it, then the next, so a moonshot becomes a staircase a customer can watch you climb. Buy the most expensive component off the shelf and spend your engineering only at the seam nobody has solved. Fly early and fly cheap, because in an unsolved field your flight rate is your learning rate. Let a government customer fund the R&D by building the testbed you needed anyway. Stack a paying defense mission under the commercial dream so the near term bankrolls the moonshot on one core technology. And when you cross from prototype to production, raise to build in parallel, not to survive — buy calendar, not runway. The last archetype won by refusing to build the hardware at all; this one wins by refusing to build the final thing first.

This issue's company wins by refusing to build the final product first — it climbs a staircase of cheap, flyable prototypes toward an impossible one. The next archetype picks a different number to attack entirely: not the speed of the machine but the cost of the shot. A company that bet on a threat years before the market priced it, and won by flipping the exchange rate of the fight until every engagement cost the defender pennies. No company names yet. Collect the set: each issue is one operating archetype, and this one — sequence the impossible, fly your way up — sets up the archetype that wins by changing the math.

— COMMANDER

A new Playbook every Monday, the Weekly Transmission every Friday. If it earned its place in your inbox, forward it to one operator who's building something — that's how Commander spreads.