Defense venture capitalists are running a taxpayer-funded bezzle through the Pentagon.
It is true — the phrase is doing real work here — that the American weapons acquisition system is broken. The war in Iran has made the point with expensive clarity: the United States has brought to bear few weapons systems less than fifteen years old. The rest of the inventory is decades-old, priced accordingly, and insufficient to produce decisive victory against a smaller, lower-cost adversary. Reformers have said the acquisition bureaucracy needs overhaul since 1960, and they have been right since 1960. Pete Hegseth’s diagnosis is not novel, but it is not wrong.
The trouble is what Hegseth has built in the diagnosis’s name. He eliminated the requirements process for weapons purchases. He suspended costly cybersecurity requirements — for weapons systems, connected to networks, connected to targeting. He gave lower-ranked officers unilateral buying authority and stood up programs catering to startups, speeding up contracting and adding flexibility. The language is “innovation,” “competition,” “speed.” The structure is the deregulation of public procurement in the direction of private venture-capital portfolios.
The numbers tell the story the language is designed to obscure. Pentagon contract spending on the fifteen highest-valued defense-tech startups tripled between 2022 and last fiscal year, according to the Ronald Reagan Presidential Foundation & Institute — and still accounted for less than one percent of total contractor dollars. That rate has held for years. The startups are not displacing the incumbents. The Pentagon simultaneously doubled its spending on traditional prime contractors to $372 billion. The “disruption” is additive, not substitutive — a new spigot opened alongside the old one, not a replacement for it.
The venture capital flooding through that spigot is something else entirely. Investment in defense and aerospace startups reached $16.8 billion in the first half of 2026, exceeding any prior full-year total, according to PitchBook. Anduril Industries doubled its valuation from $30.5 billion to $61 billion in a year. More than four hundred drone companies now operate in the United States, where — the estimate comes cheerfully from within the industry — ten or fifteen will survive the next five years. Trae Stephens, Anduril’s own co-founder and a partner at Founders Fund, said recently: “I am very uncomfortable. Prices are untethered from reality.” When the co-founder of the company whose valuation just doubled tells you the prices are untethered from reality, the charitable interpretation is candor. The less charitable one is that he is describing the bezzle.
John Kenneth Galbraith coined the term in The Great Crash, 1929: the interval between the commission of a fraud and its discovery, when the embezzler has his gain and the victim feels no loss. Cory Doctorow has made it the governing theory of tech grift — the gravity-defying interval when the coyote is running on air and hasn’t looked down. The defense-tech bezzle operates on a specific mechanism. Venture capital firms invest at rising valuations. The Pentagon awards contracts that validate those valuations. The contracts are small in total — less than one percent of defense spending — but large relative to startup revenue, which is often near zero. The validation drives further investment. The investment drives further valuation. The VC firms that entered early cash out at the next round, at IPO, or in the secondary market. The taxpayer, who funded the contracts that validated the valuations, holds the bag when the bubble finds its floor.
The floor is being surveyed now. An analysis by Howe Wang at Frontier Optic tracked a cohort of 568 venture-backed defense startups and found they received $4 billion in Pentagon contracts last fiscal year, up from $1 billion in 2022 — out of $506 billion in total defense spending. Anduril and Saronic, the two deepest-pocketed winners, accounted for about a quarter of all Pentagon spending to the startup cohort. There is, of course, a charitable explanation: the Pentagon cannot evaluate and integrate hundreds of tiny suppliers, and concentrating contracts among a few scaled firms is operationally rational. But this is the bezzle eating its own tail — the procurement reality that drives consolidation is itself the product of a contracting apparatus built to inflate valuations, and the resulting concentration produces more lawsuits, more political risk, more reasons for Congress to slam the spigot shut. The consolidation is happening before the market matures: a select few companies are gaining bigger contracts while hundreds of others compete for scraps, producing lawsuits and protests by startups accusing the military of playing favorites. The pattern is not disruption. It is the familiar shape of a market that has already begun to concentrate, before it has produced anything of consequence on the battlefield.
And the battlefield is where the ledger balances. In Iran, the weapons doing the fighting are mostly old. The few new systems — attack drones re-engineered from the Iranian Shahed, drone boats from startup Saronic — are exceptions that underscore the rule. The rest of the arsenal dates to the last century, at last-century prices, and has not delivered decisive victory against a smaller, cheaper adversary. The acquisition system bought those old weapons too, at enormous cost and over decades of delay. But the argument for the startup pipeline was that it would produce better, cheaper, faster. The evidence from the only active theater is a handful of new platforms alongside an unchanged inventory, funded by a contracting apparatus whose main achievement has been to inflate the balance sheets of the firms that sell to it.
The historical pattern is not new, only accelerated. The DARPA-to-commercial pipeline has run for decades: the government funds basic research, a startup commercializes it, a tech giant acquires it. Siri came from SRI’s DARPA-funded CALO project; the DARPA Grand Challenges seeded the autonomous-vehicle industry that became Waymo and Cruise. In those cases, the public bore the research cost and the private sector captured the commercial upside. There is, though, a brutal difference that makes the current bezzle more fragile than the older one. Those earlier examples produced dual-use technologies that scaled into enormous civilian markets: consumer surplus, tax revenue, productivity gains. The Pentagon’s research bet paid for itself many times over in ways that did not show up on Lockheed Martin’s invoice. The current crop of defense-tech startups sells to a single customer with a single set of requirements and no commercial crossover to speak of. The “dual-use” pitch is the marketing. The business model is a sole-source government check. When this bezzle pops on a portfolio with no civilian market to cushion the fall, the taxpayer does not just lose the contracts. The taxpayer loses the contracts without ever having gained the consumer surplus that justified the research bet in the first place. DARPA funded research and let civilian markets scale it. The new model funds portfolios that depend on the Pentagon check staying open. What is new is the industrialization of this arrangement — the explicit intent to route Pentagon procurement through VC-backed startups as an asset class, with the venture firms positioned as the primary beneficiaries of the contract-driven valuation cycle.
Hegseth is requesting a $1.5 trillion defense budget, including $54.6 billion for an autonomous warfare unit that would largely be allocated to companies building drones and AI weapons. Officials from both parties are balking and demanding scrutiny of Pentagon spending on loans and equity stakes in startups backed by venture capitalists including Donald Trump Jr. The question is not whether defense technology should improve — it should, urgently — but whether the procurement apparatus should be structured so the loudest beneficiaries are portfolio managers and the quietest auditors are the soldiers who will have to use what was bought. Genuine acquisition reform would test startup weapons against the operational demands Iran has exposed, buy what passes through competitive bid, and subject equity stakes in defense companies to the same congressional oversight that governs every other line of the defense budget.
There is a saying in the trades: if someone offers to sell you a tool they have never used to build anything, ask them why they are selling it. Four hundred drone companies. Sixty-one-billion-dollar valuations. Cybersecurity requirements suspended. Weapons that mostly date to the last century still fighting the current war. The venture firms will have their returns long before the next battlefield audit. Should the $54.6 billion autonomous warfare unit fail to deliver operational drones in the next conflict, the after-action review will be deferred by the same budget cycle that funded the failure — and the taxpayer, as always, will be asked to fund both the tools and the lesson.