# Replicator Initiative The Pentagon program launched in **August 2023** by then–Deputy Secretary of Defense Kathleen Hicks to field "multiple thousands" of cheap, "attritable" autonomous drones within 18 to 24 months, mainly for the Pacific. It was meant to answer China's advantage in military mass with American numbers. A second round, Replicator 2, added counter-drone defense in 2024. ## What it bought, and why it failed Replicator was the defense establishment's first large-scale attempt at Silicon Valley–speed acquisition, and venture-backed defense firms were its main beneficiaries. [[Anduril]]'s Dive-LD undersea vehicle moved into mass production under it, priced at about **$2.5 million a unit** in 2024 reporting ([[Anduril Dive-LD]], [[Anduril Product Ecosystem]]). The underlying argument, that technological superiority in autonomous systems is a matter of national survival, is the one the [[Effective Accelerationism]] and defense-tech crowd had been making. By its two-year mark it had not delivered the promised swarm. Later reviews blamed the chosen drones' persistent technical problems, poor integration with command-and-control systems, and above all the failure to buy the software needed to orchestrate thousands of vehicles. The Pentagon dissolved it in **late 2025**. ## What replaced it: DAWG Replicator's work was folded into the **Defense Autonomous Warfare Group (DAWG)**. Adm. Samuel Paparo of Indo-Pacific Command described the change by saying Replicator was still "alive," with a shift toward larger drones and one-way attack systems. Under Secretary for Research and Engineering [[Emil Michael]] is one of its principals. A separate "Drone Dominance" effort covers small drones. The money changed by orders of magnitude. The **FY2027 budget request asks $54.6 billion for DAWG**, against $225.9 million in FY2026. Only about $1 billion of that is in the base budget; roughly $53 billion sits in a reconciliation pot with up to five years to obligate, which leaves it largely outside ordinary appropriations oversight. [[Shield AI]] is integrating its Hivemind autonomy software into the military's low-cost attack drone. ## Why it matters here The Replicator-to-DAWG story is the vault's defense-tech pattern at scale. A program built on the case for moving fast failed on the hardest part, the software. Its successor got about 240 times the budget, most of it in the least-supervised form of appropriation, with the same small group of venture-backed contractors positioned to win it ([[Defense Tech Conflicts of Interest]]). ## See also - [[Anduril]] · [[Anduril Dive-LD]] · [[Shield AI]] · [[Palantir]] · [[Founders Fund]] - [[Emil Michael]] · [[Defense Innovation Unit]] · [[Autonomous Weapons Debate]] · [[Defense Tech Conflicts of Interest]] - [[Effective Accelerationism]] · [[Golden Dome]] - [[Space as a Warfighting Domain]] — the doctrine of orbit as a battlefield, from Space Force to Golden Dome