Anduril is building AI weapons to challenge America’s defence giants

A small tungsten warhead sits in Brian Schimpf’s hand in the middle of the Texas desert. It weighs just three pounds, small enough to disappear inside a palm, yet it is designed to tear through military targets after being carried in by an autonomous drone.

Around him, there is little except scrubland, trailers and open airspace stretching into the distance. Somewhere nearby, engineers launch Anduril’s Bolt drone vertically into the sky before it pitches forward and dives sharply toward distant targets at angles approaching 85 degrees.

For Schimpf, the engineer-CEO behind one of America’s fastest-growing defence startups, the demonstration is about far more than a single drone. It is about proving the US can still build weapons quickly enough for the wars military planners increasingly fear are coming.

A different weapons model

Over the last decade, Anduril has transformed from a controversial Silicon Valley startup into a central player in the Pentagon’s push toward autonomous warfare, AI-driven battlefield coordination and cheaper mass-produced weapons.

The company now builds loitering munitions, autonomous submarines, counter-drone systems, surveillance towers and cruise missiles, while also developing software designed to connect entire battlefields into a single operational network.

Its rise has been unusually fast for a defence company.

Anduril won a role in the US Air Force’s Collaborative Combat Aircraft program, secured a long-term autonomous submarine contract with Australia through its Ghost Shark system and recently received an expanded US Army agreement worth up to $20 billion over the next decade.

At the centre of nearly all of those projects sits Lattice, the company’s AI-powered command-and-control platform. Rather than functioning as a single weapon, Lattice acts more like a battlefield operating system, collecting information from radars, drones, cameras and sensors before turning it into a shared tactical picture.

That capability has become increasingly important as modern combat shifts toward dense drone warfare, electronic attacks and simultaneous engagements involving huge numbers of targets moving at once.

Anduril’s argument is that traditional defence procurement moves too slowly for that environment.

“Our munitions production rate and our stockpiles are at a dangerous level,” Schimpf told Fortune, warning that the US industrial base is struggling to keep pace with the realities exposed by the wars in Ukraine and the Middle East.

Cheap enough to lose

Much of Anduril’s strategy revolves around a concept now dominating military planning: attritable systems.

Instead of relying entirely on extremely expensive missiles, aircraft and drones designed never to fail, companies like Anduril are building systems intended to be cheap enough to lose in large numbers while still remaining operationally effective.

That thinking is visible across the company’s products.

Bolt carries compact warheads designed around strict aerodynamic and weight constraints. Barracuda is being developed as a lower-cost cruise missile intended for rapid production at scale. Altius drones are designed as relatively expendable loitering munitions capable of surveillance and strike missions.

The underlying assumption is that future wars may consume weapons faster than Western defence industries can replace them.

That concern has intensified following the conflicts in Ukraine and Iran, where drone saturation attacks, missile barrages and electronic warfare have forced militaries to rethink both stockpile requirements and production timelines.

According to Schimpf, recent US operations linked to Iran burned through roughly a decade’s worth of Tomahawk missile production in only days.

The solution Anduril proposes is not perfection, but speed.

Rather than spending years refining a platform before deployment, the company pushes for rapid iteration, battlefield feedback and continuous upgrades, borrowing heavily from Silicon Valley software culture.

“I always ask [customers]: ‘Do you want it where it’s at, or do you want to wait another year?’” Schimpf said. “Almost always the answer is: ‘I’ll take what you got right now.’”

Ukraine changed everything

Ukraine has become one of the defining laboratories for that philosophy.

The war has demonstrated how rapidly drones, countermeasures and battlefield software evolve under real combat pressure. Systems that work one month can become vulnerable the next as both sides adapt electronically and tactically.

Anduril’s own systems were not immune.

Reports from Ukraine suggested Altius drones struggled under heavy Russian electronic warfare, including signal jamming and disruption. Ukrainian forces reportedly shifted toward cheaper and more adaptable alternatives for some missions.

Schimpf openly acknowledged the mismatch.

Altius, he explained, was initially designed around US operational assumptions, particularly aircraft-launched missions conducted far from frontline positions. Ukraine instead required cheaper land-launched systems deployable in huge volumes and easily modified in the field.

With hindsight, Schimpf said Anduril’s Barracuda missile would likely have been more relevant to the conflict’s current realities.

The company argues those setbacks reflect the nature of modern warfare itself rather than a unique failure. Drone systems are now locked in constant adaptation cycles against radar systems, jammers, spoofing tools and electronic countermeasures.

“This stuff is the hardest of the hard,” Schimpf told Fortune. “It takes some iteration, and even then, it may work for a while until it doesn’t.”

That iterative battlefield cycle has become one of the clearest lessons of the Ukraine war. Both Russia and Ukraine continuously redesign drones, update software and alter electronic warfare tactics in near real time.

For defence startups like Anduril, the ability to adapt quickly may ultimately matter more than building flawless systems on the first attempt.

Silicon Valley meets the Pentagon

Anduril’s culture reflects that mindset.

Schimpf, a former Palantir engineering executive, runs the company with a heavy engineering focus rather than the finance-driven structure critics often associate with major defence contractors.

The company intentionally operates with what executives describe as controlled chaos to preserve rapid development and experimentation even as headcount has surged past 8,000 employees.

That approach has helped Anduril become one of the most heavily funded defence startups in the world, with projected revenue expected to grow from $2.2 billion in 2025 to roughly $4.3 billion in 2026, according to figures cited by Fortune.

Its systems are now being sold or developed for countries including Taiwan, Japan, South Korea, Australia and the United Kingdom as governments accelerate investment in autonomous systems and counter-drone technologies.

Yet the company’s ambitions stretch beyond individual contracts.

Anduril is effectively trying to build a new model for the Western defence industry itself: software-centric, rapidly iterative and capable of producing autonomous systems at industrial scale.

“The question isn’t whether we can build the next Lockheed Martin,” Schimpf said. “It’s whether we can avoid becoming the thing we’re trying to replace.”

Sources: Fortune, Anduril, Business Insider

This article is made and published by Asger Risom, who may have used AI in the preparation