Capital Found Defense Tech. Proof Didn't.

Defense technology is having the best funding run in its history, and 2026 is on pace to outdo the year that set the last record. In the first six months alone, defense and defense-adjacent startups raised about $35.6 billion, according to PitchBook. That's roughly two-thirds of everything raised in all of 2025. The median round nearly doubled year over year, so the money is moving down the stack, not just into a few names at the top.

It isn't only venture funds anymore. The primes that startups were supposed to disrupt are now investing alongside them. Lockheed Martin expanded its venture fund from $400 million to a billion. Defense contractors joined a record $4.1 billion of venture rounds this year. When the incumbents start writing checks into the companies aimed at them, the category has arrived.

The money is ahead of the proof

A record year for capital is not a record year for capability, and this is where the story gets honest. Strategic acquisitions have shown that scarce defense capabilities can sell for billions. But the market still holds far more multi-billion-dollar private companies than it has liquidity events or fielded production to validate those prices. The likely winners are the ones that can make three hard transitions in a row: prototype to procurement, procurement to production, and a single product into a platform. Many of the funded companies haven't cleared the first.

You can see the uncertainty in the numbers themselves. Estimates of how big 2025 actually was sit anywhere between about $9.6 billion and $49.1 billion, depending on whose definition you use. That's roughly a fivefold spread, and it comes down largely to disagreement over whether a satellite operator or a dual-use AI company even counts as defense tech. When the market can't size the boom to within 5x, a lot of what looks like signal still depends on definition.

Validation unlocked the money. Now it's the scarce part.

It helps to remember what moved the capital in the first place. Investors didn't pile in because a pitch deck improved. They piled in because of validation. Ukraine turned drones and autonomous systems from slideware into demonstrated fact, and that real-world proof helped give generalist funds the confidence to write defense checks at scale.

The thing that unlocked the money is the thing most of the funded companies still can't show. Real proof means a system that survives a live operator, a contested RF environment, weather and dust, and the edge cases a lab and a demo reel are built to hide. And the incentives make it worse. In a hot market, the reward goes to whoever ships the most convincing demonstration fastest. A demonstration is not a deployment.

Why this is a buyer's problem, not just a builder's

For the people writing checks and the institutions buying hardware, the asymmetry is brutal. The cost of being wrong about a consumer app is a wasted seed round. The cost of being wrong about an unmanned system is measured in capital, in compromised security, and sometimes in lives. Premature deployment is not a soft failure. It's the failure mode the whole sector is structurally biased toward right now.

The traditional primes were supposed to be the filter, but their procurement and validation cycles can struggle to keep pace with how fast autonomy iterates. Agile startups have the speed but lack the testing access and the institutional machinery to prove battlefield efficacy on their own. So a gap opens between the people who can innovate and the people who can certify, and capital pours straight through the middle of it.

The scarce asset is trust

This is why the next decade of defense tech turns on a different question than who builds the most impressive system. It's who can be trusted to say, before the money moves and before a system is fielded, whether the thing actually works. In a gold rush, the scarce asset is rarely another shovel. It's a trustworthy read on what is real.

That's the layer UAX operates: operator-led validation of unmanned systems across air, ground, sea, and subsurface, run by people who have spent years in the domains they test, not analysts reading a lab report.

The record figures are real, and they're worth taking seriously. But the number that will matter more over the next decade isn't how many billions went in. It's how much of what got funded can survive the only test that counts.