The Money Has Arrived. Now Someone Has to Run the Factory.

On July 27, 2026, the first Fury autonomous fighter rolled off the line at Anduril's Arsenal-1 in Pickaway County, Ohio. The project went from announcement to first aircraft in 557 days. The industry norm for that sequence of permitting, building, equipping, hiring, and producing is three to five years. Anduril plans to add more than 4,000 jobs there by 2035.

An hour down Route 33

I went to college at Ohio University in Athens, about an hour southeast of Arsenal-1 on U.S. 33. In the early 2000s, the light industry that had once filled the beautiful Hocking Valley was mostly a memory. Rocky Boots was still headquartered in Nelsonville, and in the late 1990s it had been one of the last U.S. footwear companies still running a domestic plant. But most of its production had long since moved to Puerto Rico and the Dominican Republic. Up the road in Lancaster, Anchor Hocking kept its glass furnaces burning, one of the region's remaining big industrial employers.

Back then, nobody in southeast Ohio would have predicted that the most advanced aircraft factory in the country would go up a short drive away. The skilled-labor story in that part of the state was about leaving. Now it's about coming back. In the Ruhr Valley, Germany took coal country and rebuilt it, over decades and on purpose, into a center for engineering and the Mittelstand. The Hocking Valley never got that plan. Arsenal-1 is the first sign it might get a second chance.

It's a good story, and it's also the exception. For every Arsenal-1, dozens of well-funded companies are finding out that raising money was the easy part.

A flood of capital

U.S. venture investment hit a record $412.7 billion in the first half of 2026. Worldwide, physical AI (autonomy, robotics, aerospace) drew $47.4 billion, more than in 2022, 2023, and 2024 combined. Defense tech alone raised $69.5 billion over the past year.

PitchBook's latest report on defense technology names the test ahead: whether record valuations and contract vehicles turn into funded orders, scaled production, and recurring revenue. That is a question of operating execution. Capital is no longer the constraint.

The Tim Cook lesson

In 2017, Apple CEO Tim Cook said something at the Fortune Global Forum that still unsettles people. China, he said, "stopped being the low labor cost country many years ago." Apple builds there because of the skill, and the sheer quantity of it in one place. In the U.S., he said, you could hold a meeting of tooling engineers and might not fill the room. In China, you could fill several football fields.

So here's the natural follow-up: doesn't AI fix that now? If know-how is the moat, and AI can write the work instructions, run the simulation, catch the defect, and answer the engineer's question at 2 a.m., shouldn't the gap be closing? Partly. But the evidence says the knowledge that matters most isn't the kind that fits in a manual.

Why BYD is still cheaper

Rhodium Group recently took apart the cost gap between BYD and Tesla, both building cars in China. BYD builds a comparable EV for about $4,700 less. Subsidies, the usual explanation, account for only about 5% of that gap. The biggest single factor, about $2,400 per car, is vertical integration. BYD makes about 80% of its tier-one components in-house, compared with 37% for Tesla. Another $1,700 comes from lower overhead: engineering and administrative costs spread across enormous volume. BYD even gets a working-capital edge by taking about 155 days to pay its suppliers, against Tesla's 60.

Look at what that list contains. Vertical integration is an operating decision. Overhead discipline is a management system. Payment terms are a CFO's job. None of these is a labor advantage, and none is a technology gap that software can close. They are the result of thousands of operators making thousands of decisions, repeated for twenty years.

My grandparents' line

My grandparents worked for Timken in Canton, inspecting bearings as they came off the line. It was hard, repetitive work: hour after hour looking for flaws a customer would never see until something failed. Their eyes and hands held knowledge no manual fully captured. They knew what a bad part looked and felt like.

Timken is still headquartered in North Canton, and the steel business it spun off in 2014, now called Metallus, still runs three plants in Canton. But the job my grandparents did is exactly the kind AI now does well. Machine vision can inspect a roller in a fraction of a second, never gets tired, and never has a bad shift.

What AI can do, and what it can't

That's the promise. Machine vision catches defects people miss. Generative design shortens engineering cycles. Models can capture what a retiring inspector or machinist knows before it walks out the door. That matters when the Manufacturing Institute estimates U.S. manufacturers will need up to 3.8 million new workers by 2033, with 1.9 million of those jobs at risk of going unfilled.

But AI speeds up whatever operating system it's plugged into. If a company has no cost accounting, AI gives it faster bad numbers. If supplier strategy is incoherent, AI gives it more detailed incoherence. Tacit knowledge (how to ramp a line, when to bring a component in-house, how to negotiate terms with a sole-source supplier, how to pass a first DCAA audit) still lives in people who have done it. AI can take over my grandparents' inspection job. It can't decide what the factory should build, how to pay for it, or when to scale.

The founder's second job

Here's a familiar pattern. A defense hardware startup closes a $40 million Series B on the strength of a working prototype and a promising contract vehicle. Eighteen months later, the prototype is still excellent. But the bill of materials has grown 30%, the first production lot is late, two key suppliers want payment up front, and the program office wants cost data the company's accounting system can't produce.

The founder was the right person to invent the product. Scaling it is a different job, and it usually calls for people who have run production, managed a budget under government scrutiny, and built finance and operations functions from scratch.

Who runs the factory?

America has the capital, and it has the technology. What it's short on is operating capacity: experienced people who can turn money and prototypes into delivered units, on budget and on schedule.

That's the gap Nonlinear was built to fill. We give growth-stage builders senior CFO and COO leadership (production readiness, cost discipline, supplier strategy, compliance, and capital planning), sized to their stage and working alongside AI tools. The money has arrived. Now let's build the factory.

All insights