Defense Tech Raised $12.3B in H1 2026. There Are 70,000 More Jobs Than Cleared Engineers.
Defense tech VC hit $12.3B in H1 2026. Here's how to source cleared engineers when LinkedIn is thin, clearances take a year, and Anduril already won the pipeline.
Crescendo AI's July 2026 tracker, built on PitchBook data compiled for the Financial Times, put H1 2026 defense-tech VC at $12.3B, nearly double all of 2025's $9.95B. Peregrine Technologies alone closed a $250M Series D at a $6.8B valuation on June 23, with the press release naming "expansion of Peregrine's engineering and implementation teams" as a specific use of proceeds. The money is already deployed. The engineers are the constraint, and your sourcing stack was built for a market that does not exist here.
The capital is fast. The clearance queue is not.
Anduril took $5B in May at a $30.5B valuation. Shield AI took $2B in March. Saronic took $1.75B. American defense-tech startups captured $11.4B of the global $12.3B total, with Anduril alone accounting for almost half. Crunchbase's broader national security bucket puts the number above $14.6B in just the first five months of the year.
Now the other side of the ledger. There are roughly 2 million U.S. persons holding active federal security clearances, less than 0.6% of the population. ClearanceJobs' own analysis says there are about 70,000 more open positions than there are cleared people to fill them. A Tier 3 background investigation runs six to twelve months. VP and C-suite defense searches routinely take 75 to 110 days on their own.
The asymmetry is the whole story. Capital deploys in weeks. Clearances do not. H1 2026's raise is effectively hiring against a 2024-vintage clearance pool, and the Q3-Q4 hiring war is already lost for anyone who has not started sponsoring, relocating, or mining alumni graphs by now.
LinkedIn is structurally wrong for this market
Every playbook a SaaS recruiter has ever run assumes the candidate's profile is a marketing document. In cleared work, it is the opposite. Engineers on SCIF-adjacent programs are actively discouraged, and in some cases contractually prohibited, from listing program names, agency affiliations, current employer specifics, or the letters "TS/SCI" on any public profile.
This inverts the entire sourcing heuristic. A thin LinkedIn on a cleared engineer is often a positive signal, not a negative one. The best people working on the hardest classified problems look, on paper, like they have done nothing for four years. Your Boolean strings for "TS/SCI clearance polygraph" return the candidates who need to advertise, not the ones who do not.
GitHub has the same problem in a different shape. A staff engineer at a service lab is not pushing weapons code to a public repo. Their commits, if any, are on ROS 2 forks, on obscure DSP libraries, on internal-facing tooling they touched in a previous life at a prime. The signal is there. It just does not respond to keyword search.
This is the specific friction we built Refolk for. You describe the person in plain English ("US citizen software engineer, 5+ years, in Huntsville or DC metro, background at a prime or service lab, likely eligible for Secret") and get a ranked shortlist that infers eligibility from career trajectory instead of demanding self-declared clearance strings that policy tells candidates not to write.
The Palantir mafia is now the hiring surface
Peregrine's founders, Nick Noone and Ben Rudolph, are ex-Palantir. That is not a footnote. It is the template.
The Palantir alumni tree, and now the Anduril alumni tree, is the single highest-leverage sourcing surface in defense tech right now. These people have been through DoD procurement, they understand what a JWICS terminal is, they already hold or have held clearances, and, critically, they will accept startup equity in a way that lifers at Lockheed and RTX will not. Anduril's own $20B+ ten-year DoD contract has minted the next generation of founders and early engineers who will leave to start or join the second wave. Mach Industries in Huntington Beach, which tripled its funding from $100M in 2025 to $300M in 2026, is a live example of that cohort hitting its own hiring wall.
Alumni-graph sourcing is not something ClearanceJobs' filters do well. It is not something a LinkedIn Recruiter seat does at all, because "worked at Palantir 2018-2022" returns thousands of profiles with no way to rank by proximity to the founding team, program exposure, or current willingness to move.
ITAR is not clearance. Clearance is not clearable. Stop conflating them.
This is the single most expensive tactical error in defense-tech recruiting, and almost every ATS filter makes it.
- ITAR eligible means U.S. person status: citizen or lawful permanent resident. That is it. No investigation, no adjudication.
- Secret clearance is a Tier 3 investigation, six to twelve months typical, granted or denied by the DoD Consolidated Adjudication Facility.
- Top Secret is Tier 5, twelve to eighteen months, deeper financial and foreign-contact scrutiny.
- TS/SCI with polygraph is separate again, agency-specific, and is the scarcest and most competed-for credential in the market per Christian & Timbers' 2026 defense recruitment brief.
Recruiters who filter their pipeline for "active clearance" cut out roughly 80% of qualified candidates who are perfectly clearable and could be sponsored. Given the 70,000-role gap ClearanceJobs identified, "clearable" is the pool that actually scales. The House Armed Services Committee's version of the FY26 defense authorization bill would legally enable this "pre-clearance" hiring at greater volume, letting companies build deeper benches on classified programs while investigations run. The Senate version would extend clearance eligibility to five years after DoD departure, up from the current 24 months. Both are policy admissions that the shortage is real.
The tactical read for a recruiter: run two pipelines. One for "active TS/SCI in this ZIP code, hire in 30 days." One for "cleared-adjacent, sponsorable, will accept a start date six months out." Do not merge them. Do not force one JD to do both.
Capital deploys in weeks. A Tier 3 investigation takes twelve months. You are hiring against a 2024 clearance pool.
Geographic gravity is real and it is undefeated
Remote-first defense tech is a marketing posture, not an operating model. The work happens in SCIFs, and SCIFs live in specific ZIP codes.
Refolk's index of U.S. software engineers at Defense & Space companies returns roughly 865 profiles, and they cluster hard: Austin (Applied Research Laboratories at UT Austin, plus Saronic's own concentration), Huntsville, DC/Northern Virginia, Greater Boston, Colorado Springs, and the NAVSEA Warfare Center footprint. Not Miami. Not Boise. Not "distributed team, GMT-8 to GMT-2."
The implication for a defense-tech startup is uncomfortable but simple. Your first fifteen engineering hires need to live within commute distance of a facility you can actually clear them into. That means relocation budget, that means a physical office in one of those five metros, and that means your sourcing has to be geo-fenced from day one. A generalist AI sourcing tool that surfaces a brilliant Rust engineer in Lisbon is wasting your recruiter's afternoon. This is another place where the plain-English query pattern in Refolk earns its seat: "cleared or clearable backend engineers within 45 minutes of Redstone Arsenal" is a question a Boolean string cannot ask cleanly, but a semantic sourcing layer can.
The channels that actually work in Q3 2026
Ranked by leverage for a Series A to Series C defense-tech startup:
- Alumni graphs of Palantir, Anduril, Shield AI, SpaceX, and the tier-one primes. The Peregrine pattern is repeatable. Map the tree, rank by tenure and program adjacency, message with specifics about the mission and equity structure, not the job title.
- Service labs and FFRDCs. Applied Research Laboratories at UT Austin, Georgia Tech Research Institute, Johns Hopkins APL, MIT Lincoln Lab, the NAVSEA Warfare Centers. These people already hold clearances, are underpaid by market standards, and are one warm intro away.
- In-Q-Tel portfolio alumni and DIU/AFWERX accelerator networks. Small graph, high signal, most are cleared or eligible.
- Silent Professionals and other veteran-focused boards for transitioning military engineers, especially cyber and signals.
- ClearanceJobs. Yes, still. It hit 2 million registered candidates in June 2026, and the last 500,000 came in about three years versus a decade for the first 500,000. It is where active-clearance holders actually look. Use it for the "hire in 30 days" pipeline, not for the deeper bench.
- LinkedIn and the open web, but only through an inference layer. Because, again, the best candidates will not tell you what you need to know. Tools that treat a thin profile as a negative signal will miss the people you most want to hire.
An unfilled engineering role costs a company over $37,000 per month in lost output per ASME and Lightcast research. On a 12-person defense-tech team hiring against $250M of fresh capital, a six-month sourcing miscalibration is a $2.6M mistake before you count the runway you spent looking in the wrong ZIP codes. This is the argument for spending on a sourcing layer that actually understands the market's constraints instead of running the same GitHub-and-LinkedIn playbook you would use for a payments startup. Refolk exists because the plain-English question ("who left Palantir's Gotham team in 2023 and lives near Redstone") is the one your stack cannot answer today.
The 90-day move for a defense-tech founder
Three things, in order.
First, split your pipeline. Active clearance, hire in 30 days, is a different search from clearable, sponsor and wait. Different sources, different JDs, different comp bands, different close motions. Stop trying to run one funnel.
Second, buy a physical footprint in one of the five metros. Austin if you are unmanned and want the ARL and Saronic gravity. Huntsville for missile defense and Army adjacency. DC/NoVa for IC work. Colorado Springs for Space Force. Boston for MIT Lincoln Lab and the robotics cluster. Not two. One, first.
Third, map the alumni tree of your closest competitor or nearest analog, and go deep on the last 24 months of departures. That is where the founders' network and the recruiter's tool actually converge, and it is what wins the second-wave defense-tech hires that are being priced right now.
The $12.3B is not going to unwind. The engineers are going to stay scarce. The recruiters and founders who accept both facts and rebuild their sourcing motion around them will hire the next Peregrine. The ones who do not will lose six months, and in this market, six months is the whole round.
FAQ
Is a "thin" LinkedIn profile really a positive signal for cleared engineers?
Often, yes. Engineers working on SCI-level programs are told by their FSO and by DoD guidance not to publish program names, agency affiliations, or current-employer specifics. A candidate with four years of "Software Engineer, Confidential" and no public GitHub is behaving correctly. Judge them on career trajectory, prior employers, and warm-intro backchannels, not on the marketing polish of their profile.
What is the practical difference between ITAR eligible and cleared?
ITAR eligibility just requires U.S. person status, meaning citizenship or lawful permanent residency, with no investigation involved. A security clearance (Secret, Top Secret, TS/SCI) is a separate DoD adjudication that takes six to eighteen months depending on tier. Most ATS filters conflate the two and cut roughly 80% of qualified sponsorable candidates. Run "cleared" and "clearable" as two distinct pipelines.
How do I compete with Anduril and Peregrine on comp when I just raised a Series A?
You do not, not on cash. You compete on mission specificity, program ownership, and equity math against a smaller cap table. Anduril's alumni already believe in the sector; the people leaving in 2026 want to be employee number 15, not 1,500. Lead with the technical problem, the customer program, and the equity percentage, in that order. Comp bands come later in the conversation.
Where should I actually put my first defense-tech office?
Pick one of five metros based on your customer: Austin (unmanned, ARL/UT), Huntsville (Army, missile defense), DC/Northern Virginia (IC, DoD headquarters), Colorado Springs (Space Force, NORAD), or Greater Boston (MIT Lincoln Lab, robotics). Cleared engineering talent lives near SCIFs. Remote-first hiring in this sector is fighting gravity and losing.