Anduril's $61B Round Needs Cleared Firmware. The Real Pool Is 1,334.
Anduril's $5B Series H, Shield AI, and Saronic are chasing the same 1,334 defense-embedded engineers. Here is why LinkedIn cannot see them.
Anduril closed a $5B Series H at $61B on May 13, 2026, double its June 2025 mark. Saronic plans to hire 3,200 people in 18 months. All the new-defense buyers are bidding for the same ~1,334 US embedded engineers who already sit inside a defense or aerospace employer, and standard LinkedIn and GitHub sourcing structurally cannot see most of them.
If you are running firmware reqs at a new-defense company in 2026, the problem is not that you are posting too little. The problem is that the funnel you are using was built for the 34% of the market you do not need.
Why cleared firmware roles sit open for 102 days
Firmware time-to-fill in defense tech averages 102 days and 80% of postings stay open for months because the pool of US embedded engineers who can plausibly ship into a defense-tech company is small, geographically clustered, and already employed by roughly ten primes.
Cleared firmware engineer sourcing is not a keyword problem. It is a supply problem stacked on a discoverability problem:
- The candidates who can do the work do not advertise clearance status on their profiles.
- They do not push public GitHub commits because their day-job code is classified or ITAR-controlled.
- They list employers as "Confidential Government Client" or omit them entirely.
- They are not on the job market in the LinkedIn sense; they move through referrals and IEEE chapters.
- The primes' internal-mobility programs retain them longer than a typical software engineer stays anywhere.
Defense-tech startups also post 50% more unique open roles per company than software startups (5.2 vs 3.5 on average). More reqs against a smaller, harder-to-see pool is how you get to 102 days.
The 1,334 number that reframes the search
In Refolk's index of professional profiles, there are ~2,020 US firmware, embedded software, and embedded systems engineers who list Embedded Systems as a required skill. Of those, 1,334 are currently employed inside Defense & Space or Aviation/Aerospace firms. That is ~66% of the entire US embedded-skilled workforce sitting inside about ten employers.
Two implications fall out of that number, and both matter more than the raw count:
- If you keyword-search LinkedIn without a company-history filter, you are effectively searching the wrong 34% of the market: roughly 686 people who are not currently in defense or aerospace and who every non-defense employer is also contesting.
- If you filter by prime employer history, you are searching the right 1,334, but you are competing head-to-head with Anduril, Shield AI, Saronic, Palantir, Scale AI, Impulse Space, Astrobotic, Intergalactic, and Shift5 for every profile.
Top current employers of the defense-embedded subset, per the same index:
- Lockheed Martin
- Anduril Industries
- Mercury Systems
- AeroVironment (AV)
- SpaceX
- Impulse Space
- Astrobotic
- Intergalactic
Those eight names, plus Raytheon (RTX), Northrop Grumman, L3Harris, and BAE Systems, are the alumni graph. Nothing else is close.
| Segment | Count | Source |
|---|---|---|
| US firmware/embedded engineers, Embedded Systems skill | 2,020 | Refolk index |
| Subset at Defense & Space / Aerospace employers | 1,334 | Refolk index |
| Share of US embedded talent inside defense/aerospace | ~66% | Derived |
| Share outside defense (recruitable, uncleared) | ~34% (~686) | Derived |
| Anduril employees added in nine months | 1,000+ | evotalents.com |
| Saronic 18-month hiring plan | 3,200 | evotalents.com |
| DCSA Tier 5 pending queue, mid-2025 | ~19,000 | kore1.com |
| Time-to-fill: firmware avg vs specialist channel | 102 vs 10-21 days | paraform.com / kore1.com |
"Cleared" is a red herring for most Anduril hiring
The majority of open engineering reqs at Anduril, Shield AI, and Saronic are uncleared roles with post-hire sponsorship, not day-one TS/SCI hires. That is the sector's quiet advantage over the primes, and it is the opposite of the loudest talking point in defense tech recruiting 2026.
DCSA's 90th-percentile Tier 5 (Top Secret) processing sits around nine months, Tier 3 (Secret) at 60 to 90 days, with ~19,000 Tier 5 cases pending as of mid-2025. If you insist on day-one clearance, you are hiring from a fraction of a fraction. If you hire clearable candidates and sponsor them, your effective pool multiplies.
The real filter is not "has an active TS/SCI." It is:
- US citizenship
- Embedded, RTOS, or bare-metal C or Rust in production
- Willingness to work on weapons, ISR, or dual-use defense systems
- No foreign-national dependencies that will block sponsorship
Reframe the search around "clearable + embedded + mission-tolerant" and the addressable pool triples. That is the single highest-leverage change most defense-tech recruiters can make this quarter.
Stop sourcing for clearance. Source for citizenship, embedded chops, and mission tolerance, then sponsor.
Why LinkedIn structurally misses this cohort
LinkedIn under-indexes cleared and clearable embedded engineers because the platform's ranking signals are exactly the ones these engineers suppress. Search relevance rewards keyword density, project links, and public activity, and cleared engineers have professional reasons to publish none of that.
Common patterns that break LinkedIn search for this cohort:
- Employer listed as "US Government," "Confidential Client," or a staffing shell rather than the actual prime.
- Job titles genericized to "Systems Engineer" or "Principal Engineer" with no product context.
- Skills sections left sparse to avoid signaling program affiliation.
- No GitHub link, no personal site, no conference talks.
- Location set to a metro (Huntsville, Melbourne FL, Colorado Springs) rather than a specific employer.
The engineers most qualified to ship on Lattice, V-BAT, or Corsair are the least discoverable through the standard funnel. That is the gap Refolk closes: you describe the person in plain English ("US firmware engineers with 5+ years RTOS at a top-10 defense prime, open to relocating to Costa Mesa"), and Refolk pulls a ranked shortlist across LinkedIn, GitHub, and the open web without you having to guess the keyword the candidate did not use.
The Boeing BDS supply pulse, and why it does not repeat
Boeing's January 12, 2026 BDS supply-chain announcement pushed roughly 300 cleared workers into the market at once, the largest single liquidity event in cleared embedded this year and probably the last one for a while. Anyone building a 2026 pipeline strategy on "wait for the next Boeing" is planning for weather, not climate.
Lockheed, Northrop, and RTX had open reqs waiting when the announcement hit, and the prime graph reabsorbed most of the Tier 5-cleared portion quickly. Steady-state supply is still shrinking. DCSA processing is getting slower, not faster, and the primes are extending retention bonuses to hold their embedded staff against the new-defense poach. If you are waiting for another Boeing-scale release to fund your Q3 hiring plan, you will still be waiting in Q4.
The 40 to 100% comp premium is not what closes candidates
Anduril, Palantir, and Scale AI are openly paying 40 to 100% over prime baselines, and TS/SCI plus full-scope poly adds roughly $30K on top. Comp gets the meeting. It does not close the offer.
Prime cost structures, largely cost-plus government contracts, cannot match VC-backed equity plus cash. That is why the money gap exists. But 67% of defense firms still report they cannot fill cleared roles despite paying premiums, which means comp alone is not the lever. What actually closes cleared and clearable candidates:
- Modern engineering culture: Git, CI/CD, code review, no waterfall.
- Product ownership: shipping a system end-to-end vs owning one subsystem for four years.
- Mission clarity: a named adversary, a named platform, a named timeline.
- Autonomy on tooling: engineers pick their RTOS, IDE, and language.
- Equity that could be worth something: at Anduril's $61B valuation, early options actually cleared.
Recruiters who lead with base plus sign-on lose to recruiters who lead with "you will own the flight-control firmware on a program shipping in 14 months, and here is the tech lead you would report to." The comp premium is table stakes.
Where to actually find Shield AI and Saronic engineers
The fastest defense-embedded recruiters are placing in 10 to 21 days by working prime-alumni networks and IEEE chapters, a 5 to 10x speedup vs the 102-day benchmark. The channel mix that works looks nothing like a LinkedIn Recruiter dashboard.
High-yield channels for the 1,334-person pool:
- Prime-alumni graphs: everyone who left Lockheed, Raytheon, Northrop, L3Harris, BAE, Mercury, or AV in the last 36 months.
- National labs and FFRDCs: MIT Lincoln Laboratory, Johns Hopkins APL, Sandia National Laboratories, DARPA program alumni.
- IEEE Embedded Systems chapters, especially in Huntsville, Boston, DC, LA, and Denver.
- Embedded Online Conference attendee and speaker lists.
- Hack-A-Sat and DEF CON Aerospace Village participants.
- ClearanceJobs forum long-tenure posters, the answerers rather than the applicants.
"Everyone who left Lockheed in the last 36 months who did embedded work" is not a query LinkedIn Recruiter answers well. It is a query about employment history intersected with skill inference intersected with tenure, and it is the shape of query Refolk is built for. Ask in plain English, get the alumni slice ranked by fit, and skip the two weeks of Boolean tuning.
What a 2026 defense-tech sourcing plan should actually look like
Stop optimizing the job post and start optimizing the pool definition. Given ~1,334 defense-embedded engineers, ~686 non-defense embedded engineers who could be sponsored, and roughly ten new-defense competitors bidding on all of them, the winning plays are structural.
The five moves that beat 102-day time-to-fill:
- Split reqs into "cleared today" (small, precious, sponsor retention) and "clearable plus embedded" (10x larger, sponsor post-hire).
- Index candidates by employer history first, keyword second. The prime graph is your search space.
- Build an alumni-tracking pipeline for the ten primes and refresh it monthly.
- Show up in person at IEEE chapters, Embedded Online, and Hack-A-Sat. Public engineers do not know you exist.
- Lead outreach with the program and the tech lead, not comp. Comp is table stakes at $61B valuations.
Anduril, Shield AI, and Saronic all pull from the same 1,334-person concentration. The team that indexes on employer history and clearability, not clearance status and keywords, gets the offers signed. Everyone else waits 102 days.
FAQ
How large is the actual cleared firmware talent pool in the US?
There is no single public number for cleared firmware engineers because clearance status is not published. What is measurable: Refolk's index shows ~2,020 US firmware and embedded engineers with Embedded Systems as a required skill, and 1,334 of them currently work at defense or aerospace employers where active clearances are common. That 1,334 is the realistic upper bound of the cleared-plus-clearable embedded pool, and most sourcing plans should treat it as the denominator.
Is clearance really required for most Anduril or Shield AI roles?
No. The majority of open engineering roles at Anduril, Shield AI, and Saronic are uncleared with post-hire sponsorship. The binding constraints are US citizenship, embedded and RTOS skill, and willingness to work on defense systems. Recruiters who filter for active TS/SCI on day one shrink their pool 5 to 10x without a real programmatic reason in most cases.
How long does a new clearance actually take in 2026?
DCSA's 90th-percentile Tier 5 (Top Secret) processing sits around nine months, and Tier 3 (Secret) runs 60 to 90 days, with roughly 19,000 Tier 5 cases pending as of mid-2025. Plan sponsored hires around a Tier 3 timeline for anything you can scope at Secret, and only push Tier 5 for roles that genuinely require it.
What is the fastest way to source into this pool?
Work the prime-alumni graph directly. Specialist staffing firms hit 10 to 21 day time-to-fill by mining Lockheed, Raytheon, Northrop, L3Harris, BAE, Mercury, and AeroVironment alumni combined with IEEE chapters and events like Embedded Online and Hack-A-Sat. Tools that let you query employment history and skill inference in plain English, rather than tune Boolean strings, compress that further because "ex-prime embedded engineers open to new-defense" is not a query LinkedIn Recruiter answers cleanly.