Generated by Codex with GPT-5

This summary covers The Economist’s April 25th, 2026 International article listed in the contents as America's arms industry and published under the headline Shooting to prominence.

The article argues that America’s defence industry is entering a disruptive phase. The Pentagon still relies on giant contractors such as Lockheed Martin, RTX and Northrop Grumman, but war is changing faster than their procurement habits. Cheap drones, software-defined weapons and battlefield AI are making older, costlier systems look ill-suited to some of the conflicts the United States is now fighting or preparing for.

Into that opening have stepped the so-called “neo-primes”: Palantir, SpaceX and Anduril. They are not merely smaller versions of the legacy arms-makers. They come from Silicon Valley, build around software and data, and promise to move faster than the traditional military-industrial complex. The article’s core question is whether these firms will genuinely make American defence more agile, or whether they will simply become a new generation of entrenched contractors.

Why The Pentagon Is Looking Elsewhere

The immediate pressure comes from the economics of modern warfare. The article opens with the mismatch between expensive American munitions and cheap Iranian drones. A military that spends \$1m to destroy a \$50,000 target may win the engagement and still lose the economic argument. That problem is pushing the Pentagon toward cheaper, more numerous and more software-driven systems: drones, counter-drone tools, satellite networks, autonomous pilots and AI-assisted command platforms.

Palantir, SpaceX and Anduril fit that agenda. Palantir supplies intelligence and command-and-control software. SpaceX provides launch services and the Starshield satellite network. Anduril makes drones, sea systems and anti-drone weapons. Each has won important endorsements from the Trump administration. Palantir’s Maven system has been put on a more durable funding path, Anduril has consolidated army contracts worth up to \$20bn over ten years, and SpaceX has become a symbol of the Pentagon’s desire to copy Silicon Valley’s speed.

The legacy contractors remain much larger in defence revenue. The F-35 programme alone may cost more than \$2trn over its lifetime, and the biggest established firms still capture the bulk of procurement spending. Yet investors are betting that the challengers will shape the future. The article notes that Palantir, SpaceX and Anduril together are valued at more than three times the three biggest legacy contractors, even though their combined defence sales are far smaller. That gap says less about current scale than about expectations: markets think the old model is vulnerable.

A Different Contracting Logic

The article gives a practical reason for the excitement. Traditional defence procurement often uses “cost-plus” contracts, where the government reimburses expenses and adds a profit margin. That structure can make sense for vast, uncertain projects, but it also rewards delay and bloated cost. The challengers are more willing to use fixed-price deals, absorbing development risk themselves and earning more if they deliver efficiently.

That changes incentives. A company that profits by delivering usable products quickly has reason to reuse components, update software frequently and improve systems in shorter cycles. Anduril’s ambition to use common parts across different munitions is one example. The rapid deployment of SpektreWorks’s LUCAS suicide drone, adapted from Iran’s Shahed design, is another. The broader point is that battlefield innovation now rewards iteration as much as exquisite engineering.

For the Pentagon, this is attractive because it promises more competition and faster adaptation. The article describes a bureaucracy being cut back and an appetite for buying commercial or near-commercial technology rather than waiting years for bespoke systems. If that works, America could get a defence base that is less dependent on a handful of incumbents and better able to respond to wars where software, sensors and manufacturing speed matter as much as platforms.

The Risks Of Moving Too Fast

The article is not a simple celebration of defence startups. The first risk is execution. Palantir and SpaceX already have large non-defence businesses, which gives them scale and resilience. Anduril is more exposed to the difficulty of turning prototypes into mass production. It is building large factories, including a \$1bn facility in Ohio, but manufacturing weapons at scale is harder than impressing investors or winning early contracts.

The second risk is lock-in. The Pentagon says it wants interoperable systems rather than closed platforms, yet it could become dependent on SpaceX’s satellites or Palantir’s and Anduril’s battlefield software. If the new firms become gatekeepers for essential networks, the government may end up with a different kind of concentration rather than real competition.

The third risk is that today’s challengers turn into tomorrow’s incumbents. Anduril has bought smaller firms, and the article quotes concern that startups may need access and relationships as much as good technology to win large contracts. A neo-prime with lobbyists, acquisitions and privileged political ties could begin to resemble the old primes it set out to displace.

Politics And Autonomy

The political context makes the story more delicate. The article highlights close ties between the Trump family, parts of the administration and some of the defence-tech firms. That may help the neo-primes while Republicans control the federal government, but it could become a liability if Democrats return to power. A defence startup ecosystem that looks partisan would risk losing one of its most valuable assets: broad support for reforming procurement.

The ethics of AI weapons are another unresolved problem. These firms are enthusiastic about military AI, including target identification, data processing and autonomous systems. Their executives stress that humans remain involved in lethal decisions, but the boundary is contested. The article points to Anthropic’s refusal to allow its models to be used for autonomous weapons or mass surveillance, and the Pentagon’s reaction against it, as evidence of how tense the issue has become.

That debate matters because software is no longer just a support layer for weapons. It is increasingly part of how targets are found, decisions are accelerated and systems adapt in the field. The more central AI becomes, the more procurement reform becomes a question not only of speed and cost, but of control, accountability and public trust.

The Takeaway

The article’s judgment is that America’s military-industrial complex needs the disruption the neo-primes are bringing. Wars in Ukraine and the Middle East have shown that cheap drones, software updates, satellite links and AI-enabled systems can alter the battlefield faster than traditional procurement can comfortably handle. A defence base built only around legacy platforms is too slow for that world.

But disruption is not automatically reform. The Pentagon could move too quickly, overvalue fashionable tools, neglect conventional weapons needed for a conflict with China, or trade one set of monopolistic suppliers for another. The neo-primes could make American defence faster and more inventive. They could also become politically connected giants whose incentives drift toward the same habits that weakened the old system.

The article therefore treats the defence-tech boom as both overdue and fragile. America needs a more dynamic arms industry, but the point is not to replace yesterday’s primes with a new club of untouchables. The real prize is a procurement system that can absorb innovation without surrendering competition, accountability or strategic judgment.