Skuld and DARPA develop AI-powered aluminum scrap casting for remote manufacturing

Skuld, a US-based advanced manufacturing startup, is leading DARPA's Scrap-to-Rocket (R2R) program, developing a process to convert aluminum scrap directly into high-performance structural components. The core innovation is a patented method that melts and casts wrought aluminum alloys such as 6061 and 7075 directly from scrap, eliminating the need for rolling mills. Skuld's approach combines AI-assisted spark testing for rapid alloy identification with a process called Additive Manufacturing Evaporative Casting (AMEC), which uses 3D-printed patterns and lost-foam casting to produce complex geometries without expensive tooling. The company has demonstrated crack-free parts with wrought-equivalent strength using only casting and heat treatment, working with Worcester Polytechnic Institute and MatMicronia on microstructure prediction.

This development fits the recurring pattern of defense-driven AM innovation that later finds commercial applications, similar to how early DARPA investments in metal PBF-LB eventually seeded aerospace production. Skuld's work addresses a fundamental gap in the AM value chain: the ability to use locally sourced, non-virgin feedstock in remote or contested environments. For defense logistics, this could reduce reliance on long supply chains for replacement parts. The technical challenge is significant — wrought alloys like 7075 are notoriously difficult to cast without cracking, and the AI characterization system must handle unknown scrap compositions reliably. If Skuld can validate this process at scale, it would represent a meaningful step toward distributed manufacturing for defense and potentially for civilian applications like disaster relief or remote infrastructure repair.

From a practical standpoint, Skuld now needs to demonstrate repeatability across a wider range of scrap inputs and part geometries, and to qualify the resulting material properties against aerospace or defense standards. The AMEC process is still early-stage, and the key question is whether the AI-driven quality control can compensate for the inherent variability of scrap feedstock. For potential users in defense logistics or remote operations, this is a technology to track for proof-of-concept results rather than immediate deployment.