There's a quiet arms race happening above the battlefield, and it has almost nothing to do with weapons. It's about who can see clearly, communicate reliably, and process data fast enough to act — all from orbit. The drone surge everyone is watching on the ground depends entirely on a satellite architecture most analysts barely mention. Get that layer wrong, and the $55 billion the Pentagon is betting on autonomous warfare becomes an expensive collection of very sophisticated paperweights.
Two developments from the past week make this concrete. They're easy to miss individually. Together, they sketch the outline of where military satellite communications are actually heading — and why the gap between who gets this right and who doesn't will define force effectiveness for the next decade.
The Drone Surge Has a Hidden Dependency
Start with the Pentagon's Defense Autonomous Warfare Group, which is seeking $54.6 billion in FY27 — the vast majority through reconciliation — to supercharge drone development and collaborative autonomy. Lt. Gen. Steven Whitney described the DAWG's philosophy as "incremental capability," not a fixed procurement of one platform forever. The department wants to spin, test, and iterate.
That's the right instinct. But here's what the DAWG briefing didn't dwell on: every one of those autonomous systems needs to communicate. Orchestration tools for collaborative autonomy — the software that lets swarms coordinate, share targeting data, and adapt to contested environments — require connectivity infrastructure. When that infrastructure is degraded or denied, the autonomy software either operates in a degraded mode or goes dark entirely.
The Navy's selection of Shield AI for an $800 million ISR contract vehicle is instructive here precisely because of how Shield AI designed around this problem. The V-BAT unmanned aircraft system integrates Hivemind autonomy software specifically to continue missions with "reduced reliance on continuous human control" in environments where GPS and communications links may be degraded or denied. That's not a feature — it's a confession about the state of battlefield connectivity. The system was built to survive the failure of the satellite layer, not to depend on it.
The pattern suggests something important: the most sophisticated defense startups are designing for connectivity degradation as a baseline condition, not an edge case. That's a significant shift from how military systems were architected even five years ago.
VLEO Is the Bet That Changes the Geometry
The more technically interesting development came from Canada. NordSpace secured a Canadian Department of National Defence contract worth around $183,000 — a seed-stage figure, not a program of record — to advance very low Earth orbit satellite technologies toward hardware tests. The company's proposed Kestrel constellation aims to begin deploying from 2028, offering 10-centimeter spatial resolution imaging from VLEO altitudes.
Ten centimeters. No commercial product currently offers that resolution from orbit, according to NordSpace. For context, that's the difference between seeing that a vehicle is present and identifying what type of vehicle it is. For targeting, logistics interdiction, and battle damage assessment, that distinction is operationally decisive.
VLEO — typically defined as altitudes below roughly 450 kilometers — offers two advantages that matter enormously for battlefield connectivity and ISR: lower latency and higher resolution. Physics is doing the work. Closer to Earth means signals travel shorter distances, and optical systems can resolve finer detail. The tradeoff is brutal: atmospheric drag at those altitudes degrades orbits quickly, NordSpace's initial Kestrel satellites have a projected lifespan of around three years, and maintaining a useful constellation requires continuous replenishment.
NordSpace is betting it can solve the replenishment problem with its own launch capability — the Tundra orbital rocket, slated for first flight in 2028, and a dedicated spaceport in eastern Canada. That vertical integration strategy (build the satellite, build the rocket, own the launch site) mirrors what SpaceX demonstrated is necessary for constellation economics to work. Whether a startup founded in 2022 can execute all three simultaneously is a genuine open question. But the technical direction is sound, and U.S.-based competitors Albedo and Redwire are pursuing similar VLEO capabilities, which suggests the market thesis is real even if the execution risks are high.
The defense funding angle matters beyond the dollar amount. Early government contracts in this space serve as technical validation and, more practically, as security clearance accelerants. A company with a defense contract has a reason to pursue the clearance infrastructure that eventually lets it compete for classified programs. NordSpace is playing a long game, and the Canadian DND is placing a small, smart bet on sovereign space capability.
The Allied Coordination Problem Is Getting Structural
Meanwhile, France and Poland announced a joint military satellite project involving Thales Alenia Space, Airbus Defence and Space, and Poland's RADMOR to develop a geostationary communications satellite for Poland's armed forces. The details available are limited, but the strategic logic is straightforward: Poland, sitting on NATO's eastern flank with a land border with Russia's Kaliningrad exclave and proximity to Ukraine, needs communications infrastructure it controls and that doesn't depend entirely on U.S. systems.
Geostationary orbit for military communications is a mature technology. What's new is the political architecture — France and Poland deepening bilateral defense-industrial ties through a space program, with Polish industry (RADMOR) included as a partner rather than just a customer. This is how allied nations build sovereign capability while maintaining interoperability: joint development rather than pure procurement.
For U.S. defense tech companies, this trend has a specific implication. Allied nations are increasingly willing to fund their own satellite communications infrastructure, which creates both competition and partnership opportunities. The question is whether U.S. startups pursuing VLEO and low-latency connectivity can position themselves as interoperable components of allied architectures rather than purely domestic suppliers. The companies that figure out allied market access — navigating export controls, security frameworks, and industrial partnership requirements — will have a structural advantage over those that treat the Pentagon as their only customer.
What the Connectivity Layer Actually Needs to Become
Pull these threads together and a coherent picture emerges. The Pentagon is spending aggressively on autonomous systems that are explicitly designed to operate when connectivity fails. Allied nations are investing in sovereign satellite communications to reduce single-point dependencies. And startups are racing to deploy VLEO constellations that could offer resolution and latency characteristics current systems can't match.
The gap in this picture is integration. High-resolution VLEO imagery is only useful if it can be processed, transmitted, and acted upon faster than the target can move. That requires not just better satellites but better ground architecture, better edge processing (NordSpace's Chronos system, which uses NVIDIA-powered edge AI to reduce downlinked data by 100x, is one approach), and better software-defined radio systems that can operate across multiple satellite constellations simultaneously.
I'd argue the next major procurement fight in this space won't be about which satellite constellation wins — it will be about which company can build the terminal and software stack that talks to all of them. The military doesn't want to choose between Starlink, a VLEO constellation, and a geostationary backup. It wants a system that automatically routes through whichever link is available, at the required classification level, with the lowest latency for the mission type. That's a hard software and systems integration problem, and it's largely unsolved.
Watch for two near-term signals: whether the DAWG's FY27 reconciliation funding survives the legislative process intact (the mid-term election dynamic makes this genuinely uncertain, and Hurst's own framing of it as a "generational investment" suggests the department knows it may be a one-year window), and whether NordSpace's Terra Nova LEO pathfinder — slated to launch this fall via SpaceX rideshare — successfully validates the edge-AI processing and station-keeping technologies that Kestrel depends on. A pathfinder failure would set the VLEO timeline back significantly. A success would accelerate every competitor in the space.
The satellite layer is where the autonomous warfare bet either pays off or doesn't. The Pentagon is spending heavily on the systems that need it. The infrastructure itself is still being built.