A short clip posted to Instagram this week shows something that would have been unthinkable a decade ago: a jet-powered combat aircraft, no cockpit, no pilot, settling onto a desert runway on its own. The aircraft is Anduril’s YFQ-44A Fury. The company that built it did not exist in 2016.

That is the story behind the footage. The Fury is the most visible product of a wave of venture-funded defence start-ups that have, in the space of a few years, gone from Silicon Valley curiosity to holders of multi-billion-dollar Pentagon agreements. Anduril, the largest of them, was valued at $61 billion in May 2026. Helsing in Germany, Shield AI in San Diego and Saronic in Austin have each raised more than a billion dollars this year. None of them builds a crewed aircraft. All of them are betting that the next generation of air power will be autonomous, cheap and built in quantity.

This is what that wave looks like from the flight line.

The clip, and what it shows

The reel, posted on 3 September 2026 by the aviation account Uncanny Expeditions with the one-word caption “Fury,” runs just over twenty seconds. A YFQ-44A comes down the approach with its gear extended, flares and touches down, then taxis past a hangar. Nothing dramatic happens, which is rather the point. Landing a fighter-sized jet without a pilot on board is, for this programme, now routine.

The aircraft has been busy. In April, the Air Force’s Experimental Operations Unit flew Fury sorties at Edwards Air Force Base to test how the type could be deployed and sustained from forward bases under threat. On 20 July, the same unit took a Fury to Creech Air Force Base in Nevada for an Agile Combat Employment exercise, loading an inert missile under its wing while US and Dutch personnel looked on. The Fury has also fired a live AIM-120 AMRAAM in testing, the first CCA to do so.

From first flight to a production line in nine months

The Fury first flew on 31 October 2025. On 17 June 2026 the Air Force selected it, together with General Atomics’ FQ-42A Dark Merlin, for Increment 1 production, with a combined 150 or more aircraft due by 2030. Anduril had already started building. The company announced its Arsenal-1 factory in Pickaway County, Ohio, in January 2025; production of the Fury began there in March 2026, and the first Ohio-built Fury was completed on 27 July. The plant is a $1 billion project planned to grow to some 5 million square feet and more than 4,000 jobs, with a stated capacity of up to 150 Furies a year.

The aircraft itself is deliberately modest. It is powered by a single Williams FJ44-4M business-jet turbofan of about 4,000 pounds of thrust, tops out around Mach 0.95, and carries two AIM-120 missiles. The Air Force has talked about a unit cost well below that of a crewed fighter; our earlier reporting put the target under $30 million, roughly a third of an F-35A. The idea is not to out-fly a Su-57. It is to give a crewed fighter several expendable wingmen that can carry sensors and missiles forward, and to be able to lose some of them.

A company that did not exist in 2016

Anduril was founded in 2017 by Palmer Luckey, the Oculus virtual-reality founder who had left Facebook the year before, together with Brian Schimpf, now chief executive, Trae Stephens and a small group from Palantir. Its first products were border-surveillance towers. Its central bet was software: an operating system called Lattice that fuses sensor data and lets one operator control many machines. The hardware, drones, interceptors, submarines and now a fighter-sized aircraft, came later and fast.

The money came faster still. On 13 May 2026, Anduril closed a $5 billion Series H led by Thrive Capital and Andreessen Horowitz at a $61 billion valuation, double the $30.5 billion it commanded in June 2025 and bringing its total raised past $11 billion. TechCrunch reported 2025 revenue of about $2.2 billion, twice the previous year. In March, the US Army signed an enterprise agreement with the company worth up to $20 billion over ten years, folding some 120 existing contracts into a single pre-negotiated framework covering Lattice, hardware and support.

Whether a company with a $61 billion price tag and, by Sacra’s estimate, an operating loss of over a billion dollars this year is still a “start-up” is a fair question. In Pentagon terms it is. The traditional primes, Lockheed Martin, Boeing, Northrop Grumman, RTX and General Dynamics, are the products of a century of mergers. Neither Increment 1 CCA winner is among them: Anduril is nine years old, and General Atomics is privately held and best known for the Predator and Reaper. That, more than any single aircraft, is what has the industry’s attention.

The rest of the wave

Anduril is the largest of the new defence companies, but not the only one raising at extraordinary scale in 2026.

Helsing, founded in Munich in 2021 by Gundbert Scherf, Niklas Köhler and Torsten Reil, closed a $1.8 billion Series E on 13 July at an $18 billion valuation, led by Lightspeed and General Catalyst, the largest defence-technology round in European history. It builds the HX-2 strike drone that has been fielded in Ukraine, the Altra battle-management software, and is proposing an uncrewed fighter of its own, the CA-1 Europa. It works alongside, rather than against, Rheinmetall, Saab and Kongsberg.

Shield AI of San Diego raised $1.5 billion in March at a $12.7 billion valuation, more than double its previous mark, and told Fortune it expects more than $540 million in revenue this year. Its Hivemind autonomy software flies the V-BAT surveillance drone in Ukraine, and its runway-independent X-BAT combat drone, which the US Navy has already put money into, is due to fly for the first time by the end of 2026.

Saronic in Austin, which builds autonomous surface vessels, raised $1.75 billion at a $9.25 billion valuation this spring according to industry trackers. In Germany, Quantum Systems raised $1.2 billion at an $8 billion valuation and Stark, another drone maker, was valued at €3.2 billion. Further down the list are counter-drone specialists, hypersonic engine developers, satellite builders and factory-automation firms, most of them less than ten years old.

Why now, and what could go wrong

Three things converged. The war in Ukraine demonstrated that cheap, software-defined drones could change a battlefield, and that the side able to iterate fastest gained an edge. Western governments, alarmed by that war and by China’s build-up, loosened procurement rules and started buying from companies that had never held a prime contract. And venture capital, having watched SpaceX and Palantir turn government work into enormous valuations, decided defence was investable after all.

The risks are equally clear. Valuations of this size assume the orders keep coming and that Congress keeps funding programmes like CCA at scale; Increment 2 is still being defined. Analysts at Value Add VC have warned that defence-tech pricing increasingly mirrors that of AI infrastructure, resting on expectations of future scarcity rather than current revenue, and that a pullback in risk appetite could hit both at once. Anduril has said it does not expect to be profitable on an adjusted basis before 2030. And building aircraft in volume is a discipline the primes learned over decades, often painfully.

None of that is visible in twenty seconds of a grey jet landing in the desert. But the jet is real, it is in production, and the company that built it is worth more than several of the firms that have supplied the US Air Force since the Second World War. For the primes, the new competition is no longer hypothetical. It has a tail number.

Sources: Uncanny Expeditions on Instagram; TechCrunch; Fortune; Defense News; Nextgov; The War Zone; The Aviationist; DVIDS; Wikipedia (Anduril FQ-44); Sacra; Value Add VC; Afterburner reporting