When Deputy Secretary of Defense Kathleen Hicks first announced the Replicator initiative in August 2023, the headline number — thousands of attritable autonomous systems by August 2025 — was a deliberate provocation aimed at a Pentagon procurement machine optimized to deliver dozens of exquisite platforms over decades. Two and a half years later, Replicator 1.0 has met or exceeded most of its delivery milestones, and Replicator 2.0 is already in planning, with an order of magnitude expansion implied: hundreds of thousands of attritable systems by 2028 (U.S. Department of Defense, 2024).
What Replicator 1.0 demonstrated is that the Pentagon can scale. What Replicator 2.0 will reveal is that the supplier base behind it cannot — not at the depths the program now implies, and not with the dependencies that the U.S. small-drone industry has accumulated over fifteen years of low-cost Chinese component sourcing.
What Got Built in Replicator 1.0
Replicator 1.0 funded eight named system tranches: AeroVironment's Switchblade 600, Anduril Altius and Bolt, Performance Drone Works C100, Skydio X10D, Shield AI V-BAT, and two undisclosed programs. By the end of FY2025, the DoD had reported approximately 3,000 units delivered, with the bulk weighted toward the loitering munition and ISR ends of the catalog (USNI News, 2025).
On time, on count. The Pentagon's bureaucratic muscle, applied with sustained leadership attention, delivered. But the suppliers behind those eight tranches are, with rare exceptions, small. They were built to deliver hundreds, not hundreds of thousands.
Where Replicator 2.0 Hits a Wall
The transition from prototype-scale to mass-attritable-scale exposes four layers of the bill of materials simultaneously: motors and electronic speed controllers, batteries, autopilots and flight computers, and EO/IR sensor packages. Each is a different industrial base story, and three of the four lead back to China.
Motors and ESCs
Brushless DC motors and electronic speed controllers in the small-drone size range are dominated by Chinese suppliers — primarily T-Motor (Cuav), Hobbywing, and DJI's vertically integrated component lines. U.S. assembly does not equal U.S. content. A 2024 study by Govini estimated that more than 70% of small-drone motors and ESCs in U.S.-designed systems trace to Chinese manufacture, regardless of the prime contractor (Govini, 2024).
The Blue UAS list — DIU's qualified-vendor manifest — was created to mitigate this exposure. It has helped, but the list is itself capacity-constrained: roughly two dozen platforms across eight categories. Replicator 2.0's demand vastly exceeds the Blue UAS list's combined output.
Batteries
Lithium polymer and lithium-ion cells for small drones are even more concentrated than motors. CATL, BYD, and EVE Energy collectively control more than 80% of the global pouch and prismatic cell market in the 18650/21700 form factors that small drones depend on. The Departments of Defense and Energy have both moved to fund domestic cell production, including American Battery Factory in Arizona and KORE Power in Arizona (U.S. Department of Energy, 2024). Production starts in 2026 and ramps through 2028 — the same window in which Replicator 2.0 needs cells.
Autopilots and Flight Computers
Autopilot software has rapidly become a U.S. strength, with PX4-derived stacks from Auterion (Switzerland/U.S.) and ArduPilot from the open-source community now dominant. The underlying flight computer silicon, however, is mostly TSMC-fabricated and mostly assembled in China or Vietnam — a parallel structural exposure to the broader semiconductor story covered in earlier posts on the CHIPS Act.
EO/IR Sensors
Sub-pound EO/IR gimbals are a narrowing field, with U.S. options (Teledyne FLIR, Sionyx) and Chinese options (DJI Zenmuse, Yuneec) leading. Replicator's tier-two programs have leaned heavily on U.S.-built sensors. Cost-per-unit is climbing as demand scales, and U.S. focal-plane-array production capacity is the long-pole investment.
Skydio's Cobalt-Free Pivot — a Microcosm
Skydio's X10D pivot to cobalt-free battery chemistry, announced in 2024, is the clearest signal that drone primes now understand the supply chain dependency they inherited. Replacing cobalt with iron-phosphate chemistry trades a small range hit for a substantial supply chain hardening — cobalt's market is dominated by the Democratic Republic of Congo (mining) and China (refining) (Skydio blog, 2024).
The same pattern will play out across motors, ESCs, and sensors over the next two years. Defense primes that have not begun engineering changes for component diversification will lose Replicator 2.0 contracts to those who have.
The Tier-2 Ecosystem Quietly Forming
Behind the headline primes, a Tier-2 ecosystem is emerging in Ohio, Oklahoma, North Carolina, and Texas focused on motors, ESCs, and small actuators specifically for attritable systems. Aetheros Robotics in Tulsa, RotorX in Ohio, and Verge Aero in North Carolina have all received DPA Title III or DIU Blue UAS investment in the past 18 months (DIU, 2025). None have reached steady-state production yet.
What Suppliers Should Be Doing Now
- Engineer for component substitution: design the autopilot, motor, and battery interfaces to allow drop-in replacement of foreign-sourced parts as domestic options mature. Single-source designs will fail Blue UAS qualification.
- Pre-qualify on Blue UAS standards: Blue UAS qualification takes six to nine months. Suppliers that wait until they have a customer will miss the customer.
- Stand up CMMC 2.0 Level 2 ahead of demand: drone primes are now requiring Level 2 from Tier-2 motor and ESC suppliers, especially those touching flight-control firmware (CMMC PMO, 2025). Compliance is now the price of admission.
- Invest in test ranges: the Pentagon's flight-test queue is a months-long bottleneck. Suppliers with their own restricted airspace or partnerships with DoD test ranges win short-list slots.
Autonomy and the Software Stack
The Replicator program has revealed an under-discussed truth about modern attritable systems: at scale, the difference between a useful drone and an expensive paperweight is the software stack. Flight control, sensor fusion, on-board mission planning, and contested-environment autonomy are now the binding constraint as much as the hardware. The Anduril Lattice mission autonomy stack, Shield AI's Hivemind, and Skydio's autonomy SDKs represent meaningfully different architectural choices that flow through to operational outcomes.
The Defense Innovation Unit's Government Reference Architecture for autonomous systems, published in early 2025, has begun to standardize interfaces between flight controllers, mission planners, and command-and-control systems (Defense Innovation Unit, 2025). Standardization is a double-edged sword for primes: it lowers integration cost across systems but commoditizes flight control firmware that was once a competitive moat. Suppliers that bet on proprietary stacks now face a strategic question they did not have to answer two years ago.
Contested Electromagnetic Environments
Ukraine's experience has been the most instructive operational data the small-drone industry has ever received. GPS jamming, RF link denial, and adversarial electronic warfare have shifted survival economics. Drones designed for GPS-permissive environments have proven mass-attritable in the harshest sense — not because of vendor planning, but because EW degrades performance below operational thresholds. Replicator 2.0 explicitly requires GPS-denied navigation, alternative time-and-positioning sources, and onboard inertial reference quality that pre-2024 designs frequently lacked.
The supply chain implication is that inertial measurement units, vision-based navigation modules, and the silicon that supports them are now mainline integration concerns. The MEMS IMU industrial base — Analog Devices, Honeywell, Safran's U.S. operations, and emerging entrants — is being asked to scale precision-grade output rapidly. The same IMU shortage that constrains PAC-3 and Tomahawk extends into Replicator-tier programs.
Magazine Math at Replicator 2.0 Scale
The honest arithmetic behind Replicator 2.0 is unsentimental: a peer-conflict planning baseline for attritable systems is measured not in thousands but in hundreds of thousands. Multiple unclassified wargame summaries have noted that combat-attrition rates for first-week unmanned systems exceed peacetime depot replacement rates by orders of magnitude. The industrial base does not yet produce at scale to absorb a hot-war demand signal of that shape.
The implication for the supplier base is the same implication that ran through the munitions blogs of this series: peacetime cadence and contingency cadence are different industrial systems. The companies that build for the latter — qualifying second sources, redundant tooling, and reserve capacity that costs money in peacetime — will be the suppliers that the Department turns to in a contingency. Those who optimize purely for peacetime cost will be the ones who cannot deliver.
Thousands to Hundreds of Thousands
Replicator was always going to test whether the Department of Defense could escape its own procurement gravity. The next 24 months will test whether the U.S. supplier base — primes, Tier-2, and component manufacturers — can escape its own dependence on a Chinese drone supply chain that was perfectly rational under peacetime industrial economics and is now strategically untenable. The orders are coming. The question is who can fill them.
