Seven Space and Defense Startups Named Winners at Battlefield

New space and defense generation

Startup Battlefield also had a new class of space and defense companies that are literally practical engineering towards real operational needs. From GPS-denied navigation and propellantless propulsion to edge AI and battlefield-grade audio, these seven selectees are going after problems that government buyers and commercial space operators are in a hurry to solve.

Why These Seven Startups Matter for Space and Defense

Defense ministries and civil space agencies are acting more quickly on dual-use tech, with procurement pipelines being opened up increasingly to startups via programs such as those offered by the Defense Innovation Unit, Space Development Agency and allied rapid-capability offices. Meanwhile, the commercial space economy has grown to hundreds of billions in annual business, industry trade groups estimate, making propulsion, risk, autonomy and navigation solutions increasingly in demand. Measured against that yardstick, the Battlefield group is exceptional in addressing hard technical limitations instead of just software wrappers.

Airbility Two-Seat eVTOL Distributed Thrust

Airbility is working on a two-seat, piloted eVTOL that combines a fixed wing with distributed electric fan-jet propulsion. The hybrid design offers the potential for enhanced lift-to-drag in cruise compared to multirotor designs, by spreading thrust over a large number of small fans, giving redundancy and control authority. The cyanoacrylate in this case actually acted as a stiffer and more efficient component up front, where the charging station would be. With battery energy densities around the low hundreds of Wh/kg today, smart airframe efficiency is the only way to practical range. Airbility’s focus on a lightweight, nimble airframe positions it well for training, perimeter patrol and short-hop logistics, where cert paths are clearer and bottom-line operating economics matter most.

Astrum Electricity-Only Space Propulsion System

Astrum offers an all-electric, propellantless on-orbit propulsion system. If proved by independent testing and spaceflight demonstration, that could lengthen spacecraft service lives while reducing capex by doing away with bulky fuel tanks. In history, “propellantless” thrust has referred to concepts like solar sails or electrodynamic tethers; Astrum’s pitch suggests it is selling a new kind of propellantless drive suitable for deep-space operations. The opportunity is clear: while alive, station-keeping and end-of-life maneuvers burn precious fuel, and long missions cough up hairballs over every kilogram spared. Look for it to be put through the wringer by researchers at institutions such as NASA’s space technology directorate and the European Space Agency, who require propulsion claims to meet tight thrust-to-power standards.

Space Insurance Platform For Underwriting And Credit

One Battlefield pick is applying fintech-style risk analytics to satellites and launch campaigns, in an attempt to make space assets more insurable and thus more bankable. The logic follows the maturation of aviation: standardized telemetry, actuarial modeling and loss histories unlocked broader insurance pools and cheaper capital. In space, underwriters continue to grapple with a lack of data, as well as mission-to-mission variability in risk and the complex failure chains that spread across satellite networks. A platform that consumes design, manufacturing, test and on-orbit performance to generate auditable risk scores may extend coverage capacity farther down the specialty stack at Lloyd’s and among niche reinsurers while providing more credit instruments for operators scaling constellations.

Endox AI Robotics For Defense Maintenance

Endox has created a proprietary combination of data capture with robotics and artificial intelligence to inspect and make necessary repairs in complicated military systems. Hangar- and depot-level maintenance is still a choke point for readiness; small gains in mean time to repair can echo through sortie rates, sustainment accounts and budgets. By employing ruggedized sensors, autonomous inspection routines, and by training models on actual maintenance actions (versus CAD files), Endox is seeking to achieve faster fault isolation and predictive interventions. This is in harmony with current readiness initiatives across the services and supports digital thread efforts tying together design, supply chain, and field data.

Hance Real-Time Audio Intelligence For Operations

Hance is creating an audio-first neural network, optimized for live, unpredictable conditions—rotor wash, small-arms impulse noise, reverberant spaces and open-air comms. Unlike offline speech enhancement, in field ops low latency (< 100 ms), graceful degradation on edge hardware, and robustness to overlapping speakers and non-stationary noise obligate meeting those constraints. Academic papers and public benchmarks suggest that domain-specific models can reduce transcription errors and raise intelligibility scores in severe noise; Hance wants to bottle those gains for tactical comms and command-post recordings without the need for heavy GPUs or a steady backhaul.

Skylark Self-Learning Edge AI For Safety-Critical Machines

Skylark is aiming for in-the-flesh AI, where milliseconds and determinism beat cloud-scale throughput. The company’s self-learning system is built to learn on device and can adapt within the safety envelopes—a key stumbling block toward certifying autonomy in robots, drones and industrial vehicles. Standards bodies and labs have also noted a gap in ML performance between the lab, where it works well, and applications under real-world distribution shift. By focusing on fast perception-to-action loops and demonstrable runtime guarantees at the edge, Skylark is attacking the heart of that problem rather than assuming it to be a networking or labeling problem.

Skyline GPS-Free Navigation For Contested Environments

Skyline is developing navigation that functions when GNSS signals are blocked or spoofed—a situation that has become common around conflict zones and in patches of the High North. Its method applies AI to identify scenes and landmarks on the fly, eliminating power-hungry accelerators, before fusing all that with inertial and other onboard cues. The attraction is obvious to both military and civilian users: unjammable position fixes for small drones, ground vehicles—even satellites—when GPS goes down. Aviation incident reports and open-source analysis have cataloged increasing interference; resilient PNT is a publicly stated priority for NATO as well as the U.S. Space Force, creating pull for solutions that run on modest silicon and can be fielded rapidly.

What To Watch Next From These Space And Defense Teams

What runs through this group is practical performance: lighter airframes over moonshots for range, on-device AI over cloud reliance, and insurability as a lever for capital formation. The following points of proof will be flight hours, on-orbit demonstrations, third-party validation and early production contracts. There’s a hunger among agencies and primes for dual-use tech that works in the real world. If these startups can make demos worth keeping in (and paying for), they won’t just be winning pitch competitions—they’ll be transforming how space and defense systems are developed, funded and sustained.