AMAZEMET, a Warsaw-based developer of ultrasonic atomization and high-vacuum furnace systems for materials research and additive manufacturing, has unveiled an AI-controlled version of its rePOWDER ultrasonic atomization platform that enables fully autonomous metal powder production. The system allows researchers to operate unattended for several hours, accelerating alloy development across academic and industrial R&D environments. The innovation was announced on July 21, 2025, ahead of its debut at Formnext 2025 in Frankfurt, Germany.
AI process control for unattended atomization
The latest rePOWDER system integrates an artificial intelligence (AI) model that uses machine vision to analyze melt pool behavior in real time. Processing a live video feed every 120 milliseconds, the model autonomously adjusts torch position, plasma power, and material feed rate to maintain optimal melt pool conditions and achieve high yields in the desired Particle Size Distribution (PSD).
“In most institutions, it is far easier to buy new equipment than to hire new technical staff,” said Dr. Łukasz Żrodowski, CEO of AMAZEMET and Adjunct Professor at Carnegie Mellon University. “Our new AI process control delivers much more autonomy, allowing researchers to focus on discovery or supervise more devices and processes at the same time.”
Advanced control and performance benchmarks
The upgraded rePOWDER incorporates a newly designed Advanced Control Cabinet equipped with an industrial GPU for edge AI computing, a high-speed PLC, a redesigned plasma source, and an integrated gas recirculation and passivation system for increased safety. Connectivity via API allows for remote process monitoring and control, while connected feedstock feeders track the amount of processed material.
In benchmark tests using Ti-6Al-4V (Titanium Grade 5) wire, the system achieved production rates of up to 0.5 kg/h with four hours of unattended operation. AMAZEMET plans to extend this to eight hours and to support other high-value materials including NiTi shape-memory alloys and C-103 (Niobium) for high-temperature applications.
Toward autonomous materials discovery
AMAZEMET is developing additional feeder systems for bar, rod, machining chip, and powder feedstocks, with plans to introduce several in 2026, further expanding the system’s autonomy and material flexibility. The introduction of AI control aligns with the growing trend toward High-Throughput Materials Testing Facilities, which integrate machine learning (ML) and Integrated Computational Materials Engineering (ICME) tools to accelerate alloy discovery.
AI and data-driven materials discovery
AMAZEMET’s adoption of AI for process control follows a growing movement toward data-driven materials research and autonomous manufacturing. Recent work by researchers at the University of Toronto introduced a machine learning framework to optimize metal 3D printing, using predictive models to identify ideal parameters and feedstocks for advanced alloys.
Similarly, a recent review of AI-controlled 3D printing highlighted measurable gains from integrating reinforcement learning and model-predictive control into process monitoring systems. These developments reflect how artificial intelligence is transforming additive manufacturing, from predictive design to autonomous experimentation, and reinforcing the emergence of self-optimizing production platforms across the industry.
While you’re here, why not subscribe to our YouTube channel? Featuring discussion, debriefs, video shorts, and webinar replays.
Want to help shape the future of AM? Join the 3DPI Expert Committee.
Featured image shows New rePOWDER’s Control Cabinet with integrated recirculation system. Image via AMAZMET.
