Matt Grimm and the Factory That Thinks

The vehicle was called Titan, because student engineering teams have never been shy about naming a difficult thing before making it work. In the autumn of 2005, it sat in the California desert with a military chassis, a thicket of sensors and an assignment that bordered on rude: drive 175 miles over unpaved roads, sand, water, rocks, underpasses and whatever else the desert could arrange, without a human touching the controls.

Matt Grimm was a Cornell engineering student and leader of the business team behind the attempt. The computer scientists taught Titan to see. The mechanical team prepared it for punishment. Grimm helped persuade sponsors to supply the cash and equipment that allowed the machine to reach the starting line. Nearly $450,000 arrived in money and donated hardware, including $116,000 in cash. A robot may steer itself, but it cannot charm a donor.

Grimm also understood what the contest was really buying. The course was so severe that failure was not a scandal but a method. He called it an event with “wait till next year!” built in. Make the test difficult enough, he reasoned, and the people who returned would bring better ideas. The desert was not scenery. It was quality control.

A machine, then a mission

The useful line through Grimm’s career begins there: between an impressive machine and the system required to make it useful. He earned a bachelor’s degree in mechanical and aerospace engineering from Cornell in 2006, followed by a master’s in systems engineering. Then the abstractions acquired customers. At Booz Allen Hamilton, his consulting work connected him with the Defense Intelligence Agency and the Navy. In 2008 he joined Palantir, then a young company trying to make complicated data usable to people with complicated jobs.

Grimm worked in operational and technical roles and deployed to Iraq and Afghanistan. “Forward deployed” has since become a fashionable phrase in software, polished by a decade of sales decks. In its plainer form, it means the product and the problem occupy the same room, and excuses have nowhere comfortable to sit. A tool built far away meets the person expected to trust it. The gap between demonstration and use becomes visible at once.

He spent roughly six years at Palantir, overlapping with Brian Schimpf and Trae Stephens. The relationships mattered. So did the education in government procurement, demanding users and software that could not wait for an ideal environment. Grimm then crossed the table and became a principal at Mithril Capital Management, looking at growth-stage companies as investments rather than deployments. Engineering had taught him how systems fit together. Palantir showed him how systems meet reality. Venture capital showed him what the bill looks like.

Dinner, a deck and an inconvenient industry

Anduril began with the charming informality that precedes expensive consequences. In April 2017, one week after Palmer Luckey left Facebook, a small group gathered at his Orange County home. They ate Chick-fil-A and heard a pitch for a new defense company. The ambition involved autonomous systems, software and a willingness to work in a field much of Silicon Valley preferred to regard from a safe conversational distance.

Stephens had suggested Grimm as a co-founder. Luckey brought in Joseph Chen. Grimm, Luckey and Stephens later presented the idea to Palantir’s directors, where Schimpf was in the room. Schimpf joined and became CEO; Grimm became COO. The old Cornell teammates were now building another autonomous-systems organization, except the sponsors were investors, the customers were governments and “wait till next year” had lost its collegiate sweetness.

The division of labor suited the cast. Luckey became the exuberant public inventor. Stephens connected strategy, capital and policy. Schimpf took the chief executive’s job. Grimm occupied the place where grand plans encounter purchase orders, hiring plans, security requirements, facilities and supply chains. His remit has included procurement, manufacturing, facilities, finance and people. It is a portfolio of nouns that sound harmless until any one of them fails.

“We think of the manufacturing process of the products as the product itself.”Matt Grimm, on Anduril’s production philosophy

The sentence is pure operator: mildly inelegant, extremely consequential. Most companies allow a product to become emotionally complete before manufacturing spoils the mood with questions. Can this material be obtained? Can this shape be made repeatedly? Can a worker reach that fastener? What happens when one supplier misses Tuesday? Grimm’s formulation moves those questions to the beginning. The factory does not reproduce the invention. The factory is part of the invention.

No monuments on the floor

That idea now has a physical address in Ohio. Arsenal-1, Anduril’s manufacturing campus near Rickenbacker, began preparing its first Fury autonomous aircraft line in 2026. The opening building is arranged around 22 workstations. The equipment is meant to move. The line begins with human assembly and adds automation when the work itself reveals what should be automated. This is not aesthetic modesty. Fixed machinery can become a steel monument to yesterday’s assumption.

Grimm describes design choices that begin with production: aluminum instead of titanium where possible, commercially available components when they fit, and fewer castings or forgings that invite delay and specialized suppliers. The cheapest possible way to build something is considered while it is still being shaped. Cheap here does not mean flimsy. It means refusing to let exquisite complexity become a substitute for performance.

At full capacity across three shifts, the first line is designed for 150 aircraft a year. Fury is only the opening argument. Roadrunner and Barracuda lines are planned for the same campus, alongside a classified platform. The company has described an eventual workforce of 4,000. Numbers like these are not trophies for Grimm. They are variables in a systems problem: people, stations, parts, training, quality and time.

There is wit in the contrast. Anduril borrowed its name from a sword in The Lord of the Rings, a weapon reforged for a king. Its COO spends his days worrying about whether the sword should use common screws. Myth supplies the logo; operations checks the torque.

The quieter kind of futurist

Grimm’s public profile remains restrained beside the company’s louder personalities. He speaks in interviews, leads facility tours and comments on manufacturing, but rarely behaves as though attention itself were an operating metric. Even his most revealing comments point away from him and toward a process. Think about manufacturing on day one. Preserve the ability to reconfigure. Avoid the hard-to-build part. Put capability in the field before the calendar turns the idea antique.

A 2025 walk through Anduril’s Costa Mesa headquarters made the progression almost comic in scale. The company’s physical story had begun in a storage garage troubled by mold. The tour moved instead through a 200,000-square-foot research-and-development building, past machine tools, composites work, metrology and an electronic-warfare test environment. These are not the customary props of a software startup, though the company still talks with software’s appetite for iteration. Grimm’s job is to hold those worlds together: code that can change tonight and hardware that must survive tomorrow morning. A release can be rolled back. A badly chosen material has already arrived on a truck. The distinction explains his insistence that manufacturing decisions belong near the birth of the product, when engineers still possess the freedom to choose differently.

This makes him a particular kind of futurist. Not the person who draws tomorrow most vividly, but the one who asks whether tomorrow has a supplier. His education in this craft began when Titan rolled into the desert and every donated sensor, line of code and sponsorship promise had to survive dust and heat. It continued in rooms where soldiers used software, in investment meetings where ambition met capital, and in a startup where five founders divided a forbidding problem according to temperament.

The moral of the story is not that execution beats vision. That would flatter them both by pretending they can be separated. Grimm’s career suggests something less tidy: the production line is a form of thought. A movable workstation contains an opinion about change. An ordinary component contains an opinion about supply. A simple assembly contains an opinion about time. Factories argue in metal.

Two decades after Titan, the desert lesson remains intact. A difficult course produces fewer illusions. The machine either moves or it does not. The factory either delivers or it becomes very large architecture. Grimm has kept choosing the part of the future that must answer to the ground.