Defense tech is no longer a side bet for Silicon Valley. The money has moved into drones, autonomous ships and AI pilots, and you should take that as a real change in where venture capital thinks the next platform lives.
The numbers are hard to argue with. According to Crunchbase, defense tech startups raised $14.6 billion in venture capital through the first five months of 2026, more than the $9.6 billion they raised in all of 2025. Three rounds carried much of that weight: Anduril's $5 billion Series H in May, Shield AI's March financing package, and Saronic's $1.75 billion raise for unmanned naval vessels. These aren't odd little checks from specialist funds. They are mainstream Silicon Valley bets, written by firms that know exactly what kind of market they are trying to enter.
Anduril is the cleanest example. The Costa Mesa company, founded in 2017 by Palmer Luckey, Brian Schimpf and other former Palantir and Oculus hands, raised $5 billion at a $61 billion valuation, the Financial Times reported in May. Thrive Capital and Andreessen Horowitz led the round. The same FT report said Anduril's 2025 revenue reached $2.2 billion and that the company has a role in future U.S. Air Force unmanned aircraft programs. That is why the round matters. You are not looking at a PowerPoint company hoping the Pentagon eventually notices it. You are looking at a contractor with revenue, manufacturing plans and a valuation that now sits in the range once reserved for the most successful consumer software companies.
AI changed the math. For years, defense was easy for many venture firms to avoid: slow procurement, political heat, long sales cycles and returns that arrived on government time. AI doesn't erase any of that, but it gives investors a reason to tolerate it. Autonomous aircraft, battlefield software and sensor networks look more like compounding software platforms than like a single weapons contract. That is the part Silicon Valley understands.
Shield AI shows the point from a different angle. The San Diego company builds Hivemind, an AI pilot for drones and aircraft. Business Insider reported in February that Hivemind flew Anduril's Fury drone, also known as the YFQ-44A, over the Mojave Desert as part of autonomy testing tied to the Air Force's Collaborative Combat Aircraft effort. The company said the system handled required test points, including mid-mission updates and basic maneuvers. You don't need to dress that up. A software system flying a future wingman candidate is exactly the kind of fact that makes investors pay attention.
Its March financing was not a simple $2 billion common equity round, so the article needed tightening there. The New York Times reported that Shield AI was raising $1.5 billion in Series G funding at a $12.7 billion post-money valuation, plus $500 million in fixed-return preferred equity. That distinction matters because the capital stack tells you investors are not just chasing a story. They are structuring around risk while still accepting the premise that autonomous military software can become a giant business.
The government is pushing in the same direction. The Guardian reported in April that the Pentagon proposed $54 billion for the newly formed Defense Autonomous Warfare Group in fiscal 2027, a roughly 24,000 percent increase from the prior year. A proposal is not an appropriation, and anyone who has watched Washington budgets knows the difference. Still, the signal to investors is blunt: autonomous systems have moved from experiment to budget line.
Here is the thing. The old story was that startups could build clever prototypes but the primes would win the real work. That story is getting harder to defend. Saronic, based in Austin, announced a $1.75 billion raise in March at a $9.25 billion valuation, Barron's reported, with Kleiner Perkins among the backers. The company is building autonomous surface vessels and has talked openly about expanding U.S. shipbuilding capacity, including facilities in Texas and Louisiana. Kleiner Perkins did not wander from Amazon and Google into naval autonomy by accident. It followed the market.
There is still a hard limit in the data. Axios reported this month, citing the Silicon Valley Defense Group's NatSec100 work and the Defense Tech and Acquisition blog, that NatSec100 companies received only 0.5% of Defense Department contract obligations last year. That should keep everyone honest. Venture funding is moving faster than procurement, and a fat private valuation is not the same thing as durable Pentagon revenue.
Frankly, that gap is the story now. Silicon Valley has decided defense is investable before the Pentagon has fully proved it can buy from these companies at scale. The bet may be right. Russia's war in Ukraine, drone attacks in the Middle East and growing U.S.-China tension have all made cheap, autonomous and software-heavy systems look less optional. But you should not confuse a rush of capital with a finished market.
What you're watching is a reallocation, not a passing curiosity. Consumer social is mature. Enterprise software is crowded. Defense has a sovereign customer, visible demand and a long list of systems that need to get faster, cheaper and more autonomous. The next test is not whether venture firms will write checks. They already have. It is whether companies like Anduril, Shield AI and Saronic can turn those checks into production, contracts and machines that work outside the demo range.
Also read: Enterprise AI budgets hit a wall and the reckoning is reshaping how companies spend and how founders pitch; A Colorado startup's robotic spacecraft is days away from attempting to rescue a NASA telescope from falling to Earth; Nvidia's stock boom is quietly minting a new generation of AI startup founders
