The War Machine Economy, Part 2: The Price War

For three years, soldiers testing the Army's new augmented-reality headset complained of headaches, nausea, and blurred vision. Microsoft had spent nearly a decade and $22 billion adapting its HoloLens into a battlefield device, and the complaints never really stopped.

In February 2025, Microsoft did something a defense prime almost never does. It gave the contract away, handing a decade-long, multibillion-dollar Army program to Anduril Industries, a company that had existed for less than eight years.

That handoff wasn't an isolated embarrassment. It was the clearest sign yet of something bigger: the Pentagon's oldest, most protected business relationship, the one between the military and the handful of companies that build its weapons, is being priced out from underneath.

The Headset That Broke the Old Way of Building Weapons

The Integrated Visual Augmented System was supposed to give soldiers a heads-up display of maps, targets, and thermal imaging, built into a single headset. Microsoft won the contract in 2018 and spent years refining the hardware against a wall of soldier complaints about weight, battery life, and the kind of motion sickness that follows a bad virtual-reality demo.

Anduril didn't bid to build a better headset from scratch. It offered to take over production and keep developing the system using its own faster, more iterative engineering process, while Microsoft stayed on only to run the cloud backend. The Army approved the switch in 2025, effectively admitting that an eight-year-old software company could run a mixed-reality hardware program better than a trillion-dollar technology giant.

Did You Know?

Anduril is named after the sword wielded by Aragorn in "The Lord of the Rings," reforged from the shards of a broken blade. Founder Palmer Luckey has said the name was meant to signal that the company was rebuilding something broken, in this case, the pipeline between Silicon Valley and the Pentagon.

A Contract Type Built to Reward Running Late

To understand why that handoff mattered, it helps to understand how defense primes have been paid for seventy years. Most major weapons programs use cost-plus contracts: the government reimburses whatever the contractor actually spends, then adds a fee on top. If costs balloon, the contractor still gets paid for the balloon.

The incentive problem is structural, not a matter of any one company behaving badly. A 2008 assessment by the Government Accountability Office looked at a portfolio of 95 major defense programs and found their total acquisition costs had grown by 26%, a combined $295 billion over original estimates. The F-35 fighter jet breached federal cost-overrun limits so badly in 2010 that Congress had to formally reassess the entire program.

Cost-plus isn't purely a loophole, though. Program officials have long argued that contractors simply won't take on genuinely risky, unproven research without some guarantee they won't eat the loss if the science doesn't pan out. The problem is how far that logic got stretched, covering routine production decades after the risk had disappeared.

Fixed-price contracts aren't automatically a fix. Boeing accepted a firm fixed-price deal to build the Air Force's KC-46 aerial refueling tanker in 2011, and by 2021 its own cost overruns on that single program had passed $5 billion, more than the contract's entire original value. Boeing, not the government, absorbed the loss.

Selling Software Where Everyone Else Sells Steel

Anduril's bet, since Palmer Luckey founded it in 2017, was to build weapons systems the way a software company builds products: with its own money first, iterating fast on prototypes, then selling the finished thing at a fixed price instead of billing for every hour of development.

Its core product, a battlefield software platform called Lattice, fuses data from radar, cameras, and sensors into one real-time picture a commander can act on. In March 2026, the Army awarded Anduril a ten-year contract worth up to $20 billion, folding more than 120 separate, piecemeal procurement deals into a single enterprise agreement built around Lattice. That was on top of the IVAS handoff and a string of smaller counter-drone awards.

Investors took notice well before the Pentagon did. Anduril's private valuation went from $8.5 billion in 2022 to $30.5 billion in a Series G round in June 2025, then doubled again to $61 billion in May 2026 after a separate $5 billion funding round. It is, by some measures, one of the most richly valued defense companies in the world, without ever having gone public.

That kind of patient, platform-first bet is not unique to defense. It is the same playbook that let a graphics card company spend a decade quietly building the software layer that became the backbone of the AI industry long before its competitors realized the platform, not the chip, was the product.

Palantir's Bet: Sell the Software, Not the Steel

Palantir took a parallel route through the same door. In July 2025, the Army announced a ten-year, up to $10 billion enterprise agreement with Palantir that folded 75 separate existing contracts into a single deal for data integration and artificial-intelligence tools, cutting procurement overhead the Army said had been slowing deployment.

The payoff showed up fast in the company's numbers. By mid-2026, Palantir had raised its full-year revenue guidance to roughly $7.65 billion, about 71% higher than the year before, with government and commercial revenue now running close to even. Its market value passed $300 billion the same year, a scale the market has rarely assigned to a company that builds no hardware at all.

Palantir and Anduril aren't just parallel success stories. In December 2024, the two companies began organizing a consortium, reportedly including SpaceX and OpenAI, explicitly aimed at winning a larger slice of the Pentagon's roughly $850 billion annual budget away from Lockheed Martin, RTX, Boeing, General Dynamics, and Northrop Grumman. Palantir's data platform now runs directly on top of Anduril's Lattice sensor network in some deployments, giving the two companies a combined pitch neither could make alone.

Why the Primes Can't Just Copy the Playbook

The five legacy primes aren't standing still. Lockheed and RTX have both taken on more fixed-price deals in recent years, and the Air Force has said it has no plans to abandon that contract structure. But a contract type alone doesn't change a company's underlying cost structure.

Decades of cost-plus work built organizations optimized to document costs carefully, not to eliminate them. Factories, union labor agreements, and subcontractor layers built for an era of guaranteed reimbursement don't become lean overnight just because a new contract says "fixed price" at the top. Boeing's own KC-46 losses are the cautionary tale: the contract type changed in 2011, but the company underneath it hadn't, and it paid for that mismatch in billions of dollars of self-funded losses.

Anduril and Palantir never carried that legacy weight. They built lean, software-first organizations from day one, with venture capital absorbing early losses instead of the Pentagon. That is the actual advantage: not a clever contract, but a cost base built for fixed prices from the start.

Knowlegic Perspective

It is tempting to read this as a story about better technology winning. It is really a story about who gets to absorb risk, and who gets paid regardless of outcome.

Cost-plus contracting made sense when the United States was the only country attempting genuinely novel weapons engineering and needed contractors willing to try things that might fail. Over decades, that same arrangement calcified into something closer to a subsidy, one that rewarded documenting overruns more reliably than preventing them.

Anduril and Palantir didn't out-engineer Lockheed Martin or Boeing in any single technology. They out-priced them, by building companies whose cost structures could survive a fixed price and whose software could be resold, upgraded, and repriced the way hardware never could. The Pentagon's oldest contractors aren't losing a war. They're losing a price war they were never built to fight.

Sources & References

• U.S. Army Awards Enterprise Service Agreement to Enhance Military Readiness and Drive Operational Efficiency, U.S. Army (2025)
• Palantir lands $10 billion Army software and data contract, CNBC (2025)
• Palantir (PLTR) Q1 earnings report 2026, CNBC (2026)
• Anduril Doubles Valuation to $61 Billion With Latest Funding, Bloomberg (2026)
• Anduril raises $2.5B at $30.5B valuation led by Founders Fund, TechCrunch (2025)
• Anduril Takes Over US Army's $22B IVAS Program From Microsoft, The Defense Post (2025)
• Boeing's cost overruns on KC-46 now exceed initial contract with US Air Force, Defense News (2021)