Defense Demand Is Helping Finance America’s Industrial Rebuild

Erin Price-Wright frames the industrial problem in direct terms: the United States cannot have a defense industrial base without an industrial base. Conflicts abroad, she says, have made the consequences of decades of industrial hollowing visible on the battlefield, for American service members and allies. The lesson, in her view, is that the country needs to build again.

Connor Love sees the conditions for rebuilding as a convergence of geopolitical pressure, capital moving into physical industries, and entrepreneurs choosing to work in them. He argues that industrial capacity should be understood as a connected stack: materials and other inputs, the factories that turn them into products, and the systems those products ultimately power. A defense system depends on more than the company assembling it; it also depends on the components, materials, and production capacity behind it.

That connection helps explain how defense demand can support industries with broader markets. A new factory or production line can be expensive to establish before commercial demand is large enough to justify it. Love says a government customer willing to pay for an early system can change the initial economics: the company may be able to buy more machines and build capacity that commercial customers alone would not yet support.

Price-Wright uses a US printed-circuit-board factory to make the point. The machinery is expensive, and a defense buyer can help cover some of the initial investment. But printed circuit boards also go into robots and hearing aids. In this example, a defense customer helps establish production capacity that can serve other products too.

Love cautions against treating “dual use” as a startup’s strategy. Trying to serve too many markets at once can distract a young company. The broader point is about what production capacity can serve over time: factories, energy production, and materials do not necessarily have only one eventual use. Defense can be the first market that supports a capability without defining every market that capability may later reach.

Price-Wright points to DJI as an example of production knowledge moving between markets. She says the company began with consumer drones and later developed large-scale production for public safety and defense. The people behind DJI, she says, are now the world’s largest producers of robots. A drone motor is similar to a robot actuator, one of the hardest components to build into a robot. In her account, experience producing drones helped establish expertise relevant to consumer and industrial robotics.

The exchange describes a relationship between government demand and later commercial use. Price-Wright says the government can help subsidize the development of technology and production knowledge, while commercial customers push products to become more capable and cheaper. She sees that pattern across energy, electronics, and the wider industrial stack.

The capacity argument depends on more than installing machines. Connor Love says companies also need to rethink the processes that connect design to production. At Hadrian, he says, that includes quoting and taking an order, making a part, inspecting it, and moving it through the factory. There is no single software fix for a factory; the work involves examining many steps and asking whether each is still the right one.

Erin Price-Wright describes one opportunity in the way production lines are configured. Automation is easiest to justify when a line makes large quantities of a narrow range of products. When a factory must switch among different products or configurations, each change can make automation more costly. Price-Wright says manufacturing technicians who already understand a process can use software tools to reconfigure a line themselves. A change that might otherwise require a systems integrator, take three months, and cost hundreds of thousands of dollars could, in her example, take a manufacturing engineer about 20 minutes using Claude. The point is not that the technician becomes a software engineer; it is that software lets someone with practical knowledge of the line act on it more quickly.

More adaptable production can affect both iteration and scale. Love argues that if a company cannot change a line quickly, it can be slower to refine a prototype and may have to revisit its subsystems when it is time to produce at rate. Designing for manufacture from the outset means considering how a system will be made, not only whether it works. He points to Anduril facilities in Ohio and near Baltimore, and to Saronic’s ship, as examples of companies building physical production sites. In his account, real products are coming off lines, rather than the work remaining a theoretical exercise.

The same pressure to rethink production applies to the components inside a system. Love describes a supply chain in which suppliers often receive a specification and produce a part that meets it, rather than helping shape a new design. He uses landing gear as an example: a supplier may continue making a component that works because the customer has not asked for something different. That can be rational, but when a new system has new requirements, a component designed under old assumptions may no longer be the right fit.

Love calls the opportunity to change this arrangement a “manufacturing middle”: suppliers that work with manufacturers as co-engineers, rethinking parts and how they are made instead of simply receiving specifications. Price-Wright describes a parallel opening in energy. A grid built around static and mechanical equipment deployed at scale in the 1970s and 1980s is giving way, in her account, to modern power electronics and battery storage—an architecture that is more flexible and software-native. Because the technology is different, she says, the supply chain and the engineering behind it can be reconsidered rather than merely adapted from the old system.

Closer collaboration can also improve quality. Price-Wright recalls a portfolio company that used to order twice as many printed circuit boards as it needed because it expected half the batch to fail testing. New suppliers emerged to address the problem, using software and modern manufacturing methods. The company’s pass rate later reached 70, 80, or 90 percent, she says. Her example connects co-development to a specific production problem: customers can identify what is failing, and suppliers can adjust how the product is made.

Love also sees a connection between manufacturing components for defense systems and building robots. Motors, actuators, valves, and switches used in military systems can be relevant to robotics. Producing them for one market may help develop capacity for another. He says robotics companies often buy much of what they need from China or elsewhere, and gives quality, cost, speed of iteration, and geopolitics as reasons to bring more purchasing closer to home.

That does not mean every company should make its own components. Price-Wright says companies should use high-quality suppliers where they exist. But when the required industrial base is missing, a company may have to build some of it itself. Love points to SpaceX and Tesla as examples of companies that developed production capabilities because the US industrial base they needed did not exist in the form they required.

The case for more industrial investment has a constraint that money and technology alone may not resolve. Asked about the gap between US and Chinese electricity generation, Erin Price-Wright says she does not see it primarily as a technology problem. She argues that the United States has to decide whether it wants projects to be built and whether it wants to grow. If it does, she says, the technology and capital will follow.

The chart shown in the discussion compares electricity generation in the United States and China from 1990 through 2024. It is attributed on screen to RBC Global Asset Management, the Energy Institute, and Macrobond.

Love is optimistic that investment and construction are beginning to build momentum, but he does not assume every projected need or planned system will materialize at its expected scale. Even partial demand, he argues, can start a cycle: once companies develop a design that offers better cost, efficiency, serviceability, or ease of manufacture, buyers may have reason to keep using it. At the same time, he emphasizes that physical projects take years. A factory cannot be built by putting engineers in front of computers; the work involves moving dirt, understanding a supply chain, and installing production equipment.

Capital is changing who can take that risk. Love contrasts the US market with a top-down system in which the state directs investment toward particular industries. In his view, US investors can follow signs of technological and production success and finance companies before a government customer has fully specified what it wants. That can let a company build in advance of a formal requirement.

He uses Castelion’s hypersonic missile as an example. The company, he says, spent years building before the government awarded contracts. Making a missile that flies at hypersonic speed and works reliably is difficult, even if it is not an especially elaborate system. Love says private investors and the entrepreneurs took the risk in the business’s first three years. The later contracts and purchases followed work that had already begun.

Price-Wright says that, in the first half of 2024, more risk capital went into the defense ecosystem than in all of the previous year combined. She describes a familiar “valley of death”: a company needs capital to build capacity but has not yet secured the government contract that might pay for it. In her view, that gap is starting to disappear as risk capital extends beyond defense into energy, robotics, and critical minerals.

That shift is not a guarantee of success. Love says capital will follow if companies show they can build and produce; he also stresses that investors need patience while physical capacity is constructed. The investment case depends on companies executing, not simply on the existence of capital or anticipated demand.

There is another limit: the communities where factories, mines, power plants, and data centers are built. Price-Wright contrasts these projects with software businesses, which can be concentrated in a few cities and experienced mainly through digital services. Physical industry is different: residents see the construction and live with its effects. She argues that people need to feel they have a stake in the economic growth that comes with it.

Love points to Amca as an example of a company that works with small machine shops to build new products rather than cutting management and employees. He presents that approach as one way to work with communities instead of against them. Price-Wright adds that industrial growth could mean less concentration of wealth in a small number of coastal enclaves.

The physical build-out also asks more of founders and workers than an investment thesis can capture. Price-Wright describes Mariana Minerals CEO Turner traveling to remote sites without cell service or electricity, spending weeks away from his family, and working directly with the land. Suppliers and operating cultures may differ from those of a typical startup. She says the work can sound romantic, but the travel and day-to-day demands are rough.

Site visits make the change in capacity visible. Price-Wright says she returns to some companies over the course of a year to see what has changed. An early visit may show an empty warehouse, a few pallets, and founders excited about a rented forklift. Months later, the factory may be crowded with machines and technicians, while the company tries to secure more power to run its equipment. Love describes sites as difficult to reach and involving earth-moving and large facilities; these are physical operations with physical constraints.