From $12.7B to $20B in Six Months: Shield AI Is in Talks for a Round That Would Crown Defense Tech's Next Decacorn

Six months ago, Shield AI closed a $1.5 billion Series G at a $12.7 billion post-money valuation — a 140% jump that made headlines in its own right. Now the San Diego-based defense tech company is already in talks to do it again, and at a number that would have seemed implausible for a military software startup just two years ago: a valuation of at least $20 billion, according to a report published by The Information.

If the round closes near that figure, Shield AI will have nearly doubled in value twice inside a single year — from $5.3 billion in March 2025 to $12.7 billion in March 2026, and now potentially past $20 billion before the leaves turn. It is the latest and starkest data point in what has become a full-blown repricing of the defense technology sector, where AI autonomy software, not fighter jets or missiles, is the asset investors are chasing.

What Shield AI Actually Sells

Founded by brothers Ryan and Brandon Tseng (the latter a former Navy SEAL), Shield AI has two lines of business that increasingly feed each other. The first is hardware: the V-BAT, a ducted-fan vertical takeoff-and-landing drone that the company has fielded with U.S. and allied forces for reconnaissance and strike-adjacent missions. The second, and far more valuable, is Hivemind — the company’s flagship mission-autonomy software stack, which allows teams of uncrewed aircraft to plan, coordinate, and execute missions with minimal human intervention, even in GPS- and communications-denied environments.

It is Hivemind that explains the valuation trajectory. In February 2026, Shield AI was selected alongside Collins Aerospace as the autonomy provider for the U.S. Air Force’s Collaborative Combat Aircraft (CCA) program — the service’s flagship effort to field thousands of low-cost autonomous wingmen that will fly alongside crewed fighters like the F-35 and the Next-Generation Air Dominance platform. Then in June, the company crossed the threshold that separates promising defense startups from institutional defense contractors: the Air Force awarded Shield AI a production contract to implement Hivemind’s collaborative combat autonomy for CCA, moving the software from demos into actual fielded capability.

Under that award, Shield AI is implementing behaviors that let multiple autonomous aircraft operate together under human supervision — the technical heart of what the Air Force calls “attritable mass,” the doctrine that future air combat is won not by a small number of exquisite crewed jets but by large numbers of cheap autonomous aircraft that can be lost without losing the war.

May brought another milestone: a Pentagon deal to power LUCAS — low-cost uncrewed combat attack systems — with Hivemind, alongside the company’s inclusion (at #49) in CNBC’s annual Disruptor 50 list.

Why the Valuation Is Running So Hot

The step from $12.7 billion to $20 billion-plus in six months is not primarily a story about Shield AI’s revenue, though defense analysts note its contracted backlog has grown sharply since the production award. It is a story about scarcity and positioning.

The Pentagon’s budget is tilting decisively toward autonomous systems across every domain, and CCA is one of the few programs with committed production scale in the near term. But very few companies have autonomy software mature enough to win a production slot. Anduril dominates one lane; Shield AI has effectively claimed another. In a market where the traditional primes — Lockheed, Boeing, Northrop — are still largely integrators of other people’s autonomy stacks, a startup that owns the software layer for a program of record looks less like a vendor and more like the next platform company.

There is also a portfolio logic at work. Shield AI’s $12.7 billion round in March was announced alongside its acquisition of Aechelon, a simulation company whose technology lets Hivemind-trained agents be tested against high-fidelity virtual environments before ever touching real hardware. Simulation-plus-autonomy is the same flywheel that made Anduril’s Lattice stack valuable, and investors have clearly decided Shield AI is the most credible independent holder of that pattern in the air domain.

The Defense Tech Supercycle Continues

Shield AI’s negotiations are the latest in a string of eye-watering defense and AI-infrastructure marks. Anduril has been reported at valuations well above $30 billion; Anthropic just picked Nasdaq for an IPO that could approach $2 trillion; Mistral raised €3 billion from Samsung last week. What distinguishes this wave from the 2021 vintage of unicorn inflation is that the money is chasing production contracts and government programs of record, not projections. The Air Force production award, the LUCAS deal, and the CCA selection are the reason underwriters can argue for $20 billion with a straight face.

That said, talks are talks. Rounds at this stage routinely reprice, stall, or get outbid by an IPO window — and Shield AI has itself seen earlier negotiations (a reported $1 billion raise at a $12 billion valuation floated in February) ultimately land at different terms than first reported. A $20 billion-plus mark would also raise the bar for an eventual public listing, which most observers now consider a question of when, not whether, for the company.

What is no longer in question is the direction. The Pentagon is buying autonomy software the way it once bought aircraft, the companies that build it are compounding in value every few months, and Shield AI — a startup founded a decade ago to put a Navy SEAL’s judgment into a drone — is now negotiating from a seat at the very top of the market.