The defense funding environment has attracted serious capital, and with it, serious noise. The $7.7 billion that flowed into defense tech in 2025 landed heavily at the top of the stack: Anduril, Saronic, Helsing, Chaos Industries. The companies that received the most institutional attention are also the most expensive bets on the table. What I am looking for is further down the stack, where the problems are equally urgent, the technology is equally validated, and the price of entry has not yet reflected what the market will eventually recognize. Three companies fit that description.
Allen Control Systems — Austin, TX | Founded 2022
What they do: Allen Control Systems builds Bullfrog, an autonomous precision weapon station designed to detect, track, classify, and engage drone threats using AI-enabled computer vision and a fully passive sensor suite. Bullfrog transforms any legacy or modern firearm into an autonomous targeting system, operates without emitting radar signals that would reveal its position, uses standard NATO 7.62mm ammunition, and achieves a cost per engagement of approximately ten dollars. The company also develops Treillis, a hardware-in-the-loop test platform for tracker and fire-control system evaluation, and FAFOS — Friend and Foe Operating System — a cryptographically secure, radio-silent identification system built specifically for the drone warfare environment. CEO Mike Wior and the Austin-based team moved from founding to multi-sovereign contracts in under three years.
Revenue: Not publicly disclosed. Contracted revenue spans the U.S. Army ($1.5 million base with $4.5 million in options for five-vehicle combat platform integration), SOCOM via ManTech for maritime platform integration, Republic of Korea Armed Forces, United Arab Emirates Armed Forces, and a Romania co-production MOU.
Latest raise + backers: $30 million Series A in March 2025, led by Craft Ventures, with Inspired Capital and Rally Ventures participating. Total VC raised: $30 million. This is not an Andreessen Horowitz or Founders Fund portfolio company.
IP: ACS holds filed patent applications covering Bullfrog’s passive multi-sensor target acquisition architecture, the AI-enabled fire control computer vision stack, and the cryptographic IFF protocol underlying FAFOS. The Treillis HIL test platform carries trade secret protection. Specific granted patent numbers have not been publicly disclosed, consistent with the company’s approach to dual-use technology protection. The passive sensor design — which allows Bullfrog to operate without emitting any trackable signal — is both a patent position and an operational capability that active radar-based competitors cannot replicate by design change alone.
Top customers: U.S. Army Applications Lab (five-vehicle combat platform integration). U.S. Special Operations Command via ManTech. Republic of Korea Armed Forces. United Arab Emirates Armed Forces. Romanian Armed Forces co-production MOU. Pentagon demonstration on Abrams main battle tanks and Bradley infantry fighting vehicles.
Moat and defensibility: Counter-drone defense is solving a cost asymmetry problem. The drone threat costs its operator hundreds of dollars. Existing countermeasures cost tens of thousands to millions of dollars per engagement. Missile and directed energy systems are effective but priced in ways that make mass deployment economically unsustainable for most military applications. Bullfrog’s $10 cost per engagement is not just a product feature. It is a structural answer to the economic logic of drone warfare. The passive sensor architecture, which allows Bullfrog to operate without emitting any trackable signal, adds an operational dimension that active radar-based systems cannot match: Bullfrog can defend a position without revealing it. Multi-sovereign validation — South Korea, UAE, Romania, and SOCOM — achieved before a Series B represents commercial credibility that no marketing spend produces.
SWOT:
| Strengths | Weaknesses |
| $10 cost per engagement vs. $10,000+ for competing systems. Passive sensor architecture provides position stealth. Multi-sovereign customer validation within 18 months of Series A close. Romania co-production MOU opens NATO manufacturing footprint. xTechOverwatch competition winner. | Revenue not publicly disclosed, limiting visibility into commercial traction. $30M is thin for scaling hardware production at the speed military demand is accelerating. Program of record competitions require organizational scale ACS does not yet have. |
| Opportunities | Threats |
| U.S. Army plans to acquire at least one million drones annually within three years, creating proportional counter-drone demand. FY25 NDAA Section 1709 restricts foreign-manufactured drone components, creating domestic supplier preference. NATO counter-UAS programs expanding rapidly on Ukraine battlefield data. | Shield AI, Epirus, and CHAOS Industries compete in adjacent counter-drone categories with significantly larger capital bases. Export control complexity could constrain the pace of international expansion. |
Regulatory and compliance hurdles: Autonomous weapon systems carrying lethal capability require DoD legal review under DoDD 3000.09, which mandates human authorization for lethal engagement at current policy. Bullfrog operates in semi-autonomous mode with human-in-the-loop authorization required to fire, positioning it within current policy parameters. ITAR applies to all export of Bullfrog hardware and associated software. The Romania co-production MOU requires ITAR licensing coordination and Technology Assistance Agreements. Foreign Military Sales processes apply to the Korea and UAE contracts.
Go-to-market: ACS is pursuing a multi-sovereign deployment strategy while simultaneously building the U.S. government program of record pathway. The Army vehicle integration contract is explicitly designed as a rapid prototyping vehicle that feeds into larger procurement decisions. SOCOM adoption via ManTech validates the maritime platform use case. International contracts with Korea, UAE, and the Romania co-production arrangement demonstrate that Bullfrog’s performance characteristics hold across different operating environments and threat profiles. The company tripled its Austin operations in February 2026, signaling production scaling ahead of anticipated demand.
GP Lens: The battlefield evidence from Ukraine is unambiguous. Cheap drones are destroying expensive equipment at a rate no modern army anticipated, and the cost asymmetry only becomes more favorable to the attacker as drone prices fall. The DoD has articulated this clearly through Replicator and the Army’s stated intent to buy millions of drones annually. The inverse problem — defending against that volume of drone threat — is where Bullfrog lives. What gives me conviction here is not the technology alone but the multi-sovereign customer validation achieved on a $30 million raise. South Korea, UAE, Romania, and SOCOM did not buy Bullfrog because of marketing. They bought it because the cost and performance metrics made the decision obvious. The next twelve months are about manufacturing scale meeting program of record timing.
ARX Robotics — Munich, Germany | Founded 2021
What they do: ARX Robotics develops the Gereon autonomous unmanned ground vehicle platform alongside Mithra OS, an AI-powered operating system that enables autonomy, remote control, and manned-unmanned teaming for legacy military vehicles without requiring full fleet replacement. The company’s central thesis is that European armed forces cannot afford to replace aging vehicle fleets on any realistic budget timeline, but can afford to make those fleets autonomous with a software-defined operating layer. Founded by Marc Wietfeld, Stefan Röbel, and Maximilian Wied — all former German Armed Forces officers — the team brings direct procurement relationship depth that no outside investor can replicate.
Revenue: Not publicly disclosed. Six European armed forces are deploying or actively testing ARX systems. Germany delivered 30 Gereon units to Ukrainian forces in November 2024 for casualty evacuation and logistics support. Europe’s largest defense robotics production facility opened in February 2025.
Latest raise + backers: €42 million total Series A — €31 million initial round in April 2025 led by HV Capital, with Omnes Capital, NATO Innovation Fund, and Project A; extended by €11 million in July 2025 led by Speedinvest with continued HV Capital participation. Total raised across all rounds: approximately €59 million (~$65 million). The NATO Innovation Fund specifically invests in dual-use defense technology. No U.S. Tier 1 generalist VC involvement.
IP: ARX holds filed patents on the Gereon UGV platform’s modular sensor and payload integration architecture, autonomous manned-unmanned teaming systems, and elements of the Mithra OS autonomy stack. Mithra OS itself represents significant trade secret IP in its sensor fusion algorithms, real-time command translation layer, and multi-connectivity protocol stack. Strategic partnerships with RENK, Daimler Truck, DEUTZ, and Helsing extend the IP ecosystem through co-development agreements.
Top customers: Six European armed forces (not individually identified publicly). Ukraine — 30 Gereon units delivered November 2024 for combat logistics and casualty evacuation. German Bundeswehr procurement evaluation ongoing. UK Army Warfighting Experiment participants.
Moat and defensibility: Mithra OS is the strategic moat. The software layer that autonomizes an existing vehicle fleet is stickier than the vehicle hardware itself. Once Mithra OS is integrated into a brigade-level vehicle fleet — with sensor inputs calibrated to specific vehicle types and autonomous behaviors validated in operational contexts — replacing it requires re-qualifying the entire autonomy stack from scratch. The Gereon hardware builds operational credibility, but Mithra OS is what converts ARX from a hardware company into an infrastructure company. The founding team’s active military service means procurement relationships were built before the company existed, not after.
SWOT:
| Strengths | Weaknesses |
| Founded by former military officers with pre-existing procurement relationships. Europe’s largest defense robotics production facility operational since February 2025. Mithra OS creates software stickiness across hardware generations. Six armed forces deploying, including combat-proven deployment in Ukraine. Partnership network with RENK, Daimler Truck, Helsing, and DEUTZ creates industrial depth. | European defense procurement cycles are long. Revenue recognition from multi-year sovereign contracts can lag contract signing significantly. U.S. market access requires ITAR navigation and a domestic partner. |
| Opportunities | Threats |
| EU defense spending grew $145 billion between 2014 and 2024, with $110 billion of that in the last five years. Land operations represent 80% of military activity but are the most under-automated domain. Mithra OS addressable market spans all legacy NATO ground vehicles. European defense sovereignty as political priority creates tailwind for EU-manufactured systems. | Milrem and Forterra are direct competitors with longer operational histories. U.S.-made systems will compete for NATO contracts with political backing. Ukraine operational losses create reputational risk if systems underperform in combat conditions. |
Regulatory and compliance hurdles: European defense procurement requires compliance with EU Common Foreign and Security Policy frameworks. Autonomous weapon systems in EU member state militaries are subject to emerging European Commission regulation on AI in high-risk contexts. Exports to Ukraine are managed through German federal export licensing under the Foreign Trade and Payments Act. The NATO Innovation Fund investment creates an implied alignment with Alliance interoperability standards that should accelerate NATO member procurement processes.
Go-to-market: ARX operates a two-track commercial model: Gereon hardware sales for new capability deployment, and Mithra OS licensing for legacy fleet modernization. The hardware track generates immediate procurement contracts. The software track creates recurring revenue as vehicles are updated and use cases expand. The RENK and Daimler Truck partnerships provide distribution into existing military vehicle supply chains that ARX would take years to build independently.
GP Lens: The structural problem ARX is solving is not a technology gap. It is an economics and sovereignty gap. European armies cannot buy new fleets. They cannot achieve operational autonomy by continuing to rely on U.S.-sourced systems in a strategic environment where that dependency is increasingly questioned. ARX’s answer — an AI operating system that makes existing vehicles autonomous — is the right product for the moment. I am particularly interested in Mithra OS as a standalone asset. If that software layer achieves NATO-standard certification and becomes the reference architecture for European ground vehicle autonomy, the company’s valuation has very little to do with how many Gereon units are shipped. The six armed forces already deploying is the validation that matters.
Breaker — Sydney, Australia / Austin, TX | Founded 2023
What they do: Breaker builds Avalon, a platform-agnostic AI agent software system that runs entirely onboard any robot and enables a single military operator to orchestrate teams of up to 100 autonomous systems across air, land, and sea through natural voice commands over standard push-to-talk radios. No cloud dependency. No external networks. No GPS required. When communications are jammed or denied, Avalon continues operating autonomously, making mission-aligned decisions based on the mission brief loaded before the operation began. The core problem it solves is the operator bottleneck: the structural constraint that keeps battlefield autonomy at a one-human-per-robot ratio, making autonomous systems economically and operationally marginal at scale. Founded by Matthew Buffa (ex-Anduril), Michael Irwin (ex-DroneShield), and Vanja Videnovic (ex-Hargrave Technologies), who emerged from the UNSW Founders Defence 10X program.
Revenue: Not publicly disclosed. Completed paid demonstration contracts with both U.S. Special Operations Command and Singapore’s Defense Science and Technology Agency before closing the seed round. Avalon has been installed in the mission systems of Rheinmetall’s Boxer armored vehicles at the Australian test facility. In January 2026, the Pentagon’s Defense Innovation Unit launched a $100 million Autonomous Orchestrator Challenge — publicly naming the exact problem Breaker is solving.
Latest raise + backers: $9 million total — $2 million pre-seed led by Main Sequence (CSIRO’s deep tech venture fund), followed by $6 million seed led by Bessemer Venture Partners in February 2026. Bessemer’s prior portfolio includes Canva, Rocket Lab, Shopify, Anthropic, and Perplexity. David Cowan, Partner at Bessemer, wrote publicly that this represents “tackling one of the hardest and most important problems in defense technology.”
IP: Filed patent applications on the onboard AI agent architecture for multi-robot orchestration, the voice-to-action translation layer converting natural language operator commands into coordinated machine behaviors without external processing, and the mission-constraint enforcement system validating every action against loaded rules of engagement before execution. The OPORD parsing methodology — which extracts machine-readable mission constraints from unstructured military planning documents — carries trade secret protection. All processing runs on the robot’s existing compute hardware with no additional equipment required.
Top customers: U.S. Special Operations Command (paid demonstration contract). Singapore Defense Science and Technology Agency (paid demonstration contract). Rheinmetall (Boxer armored vehicle integration, Australia).
Moat and defensibility: SOCOM and Singapore’s DSTA do not sign demonstration contracts for software that has not been evaluated in realistic operating conditions. Both validations were achieved before a $6 million seed round, a signal quality that no marketing spend produces. The edge-native architecture compounds this: every deployment produces operational data from denied-environment missions that improves Avalon’s performance, and that data is inaccessible to competitors. The founding team’s backgrounds — Anduril, DroneShield, Hargrave — means the team has direct experience with the specific failure modes of military autonomy software that no outside hire can substitute.
SWOT:
| Strengths | Weaknesses |
| SOCOM and Singapore DSTA have paid before seed close — validated demand, not speculative. Fully edge-native: no cloud, no GPS dependency, runs on existing robot compute. Pentagon DIU $100M Autonomous Orchestrator Challenge explicitly names Breaker’s problem space. Rheinmetall Boxer integration opens NATO ground vehicle market. | $9M total raised is thin for competing against better-capitalized players. Pre-revenue on contracted basis. Australia-headquartered team faces additional friction in U.S. procurement relationships. |
| Opportunities | Threats |
| Army’s stated goal of one million drones annually creates proportional orchestration demand. Platform-agnostic design positions Avalon as the control layer for any defense robotics manufacturer. AUKUS technology sharing creates a natural three-market path across Australia, UK, and U.S. | Anduril, Helsing, and Shield AI have adjacent capabilities and could build or acquire an orchestration layer. DoD program of record competitions require organizational scale Breaker does not yet have. |
Regulatory and compliance hurdles: Avalon’s architecture maintains human authorization for lethal effects while enabling autonomous maneuver, keeping it within DoDD 3000.09 parameters. ITAR licensing is required for export to non-Five-Eyes partners. FedRAMP or equivalent authorization will be required for any classified network integration. Australian Defence export controls apply to transfers between the Sydney and Austin operations.
Go-to-market: Breaker’s immediate commercial path is embedding Avalon as the AI control layer in existing autonomous platform deployments, starting from validated SOCOM and DSTA relationships. Platform-agnostic design means Breaker is not competing with drone or robot manufacturers — it is a layer that makes those manufacturers’ platforms more deployable. The Rheinmetall Boxer integration demonstrates the model: hardware OEM provides the vehicle, Breaker provides the intelligence layer. The DIU Autonomous Orchestrator Challenge creates a formal procurement pathway Breaker is positioned to compete for.
GP Lens: Every conversation about military autonomy focuses on the hardware. Almost none of it addresses the fundamental economics problem: a robot requiring a dedicated human operator is not autonomous in any commercially meaningful sense. Breaker has identified this gap, built an operational solution, and validated it with SOCOM before raising $6 million. The $100 million DIU challenge launched three weeks before their seed close is not a coincidence. It is the government articulating the demand signal that Breaker is positioned to answer. I am including this company with a note on stage: Breaker closed a seed round in February 2026 and is pre-Series-A. Both SOCOM and Singapore’s DSTA have made paid commitments before institutional capital arrived — that is the manifesto’s definition of validated commercial demand. The stage is early. The demand validation is not.
The companies featured across this series represent my personal watchlist and research interest. I have not personally invested in any of them, and nothing written here constitutes investment advice. The views expressed are my own.
