Aluminum Prices Surge, AI Steps In
Aluminum prices have jumped 20% after the Trump administration's conflict with Iran disrupted supply from the Gulf region, which produces about 10% of the world's aluminum. The U.S. has already designated aluminum a critical mineral, and recycling startups are using AI to capture more of it from waste streams.
Sortera Doubles Capacity
Sortera, a metals recycling startup based in Indiana, just opened its second facility in Tennessee. The new site doubles the company's processing capacity to 240 million pounds, of which 90% to 100% is aluminum. That's a significant chunk of the 4.3 million metric tons the U.S. consumed last year.
Sortera's approach focuses on sorting aluminum scrap. Their system uses a combination of sensors—lasers, cameras, and X-ray fluorescence—to feed AI algorithms that classify each potato chip-sized piece of scrap by its specific grade of aluminum. By separating grades with higher accuracy, Sortera commands a higher price per pound.
Amp's AI Sorting Hits Over 90% Accuracy
Amp takes a different tack, using AI to sort through both recycling and general waste streams. Their system employs visible light and infrared cameras to identify materials like wrappers, foil, and plastics. As the waste flows on conveyor belts, robotic arms and air puffs pluck or blow items into different bins.
Amp claims over 90% accuracy at recovering specific materials, including aluminum. "Half of the aluminum in a metro area—in places with successful recycling programs—are just in the garbage, not even touching the recycling system," said Matanya Horowitz, Amp's CTO.
The Technical Details
The AI models at both companies are trained on labeled images of scrap. For Sortera, the key is distinguishing between different aluminum alloys (e.g., 6061 vs. 5052), which have different market values. They use hyperspectral imaging and XRF to get elemental composition, then a neural network classifies each piece in milliseconds.
Amp's system uses a combination of CNN-based object detection and material identification. The infrared sensors pick up spectral signatures that differentiate aluminum from plastics and paper. The system then triggers pneumatic actuators or robotic arms to sort items at line speeds of several feet per second.
Why It Matters for Developers
If you're working in computer vision, robotics, or industrial IoT, this is a real-world application of AI that directly impacts commodity supply chains. The sensor fusion (visible light + IR + XRF) and real-time classification at high throughput present interesting engineering challenges. Both companies are likely using custom-trained models, possibly with PyTorch or TensorFlow, deployed on edge devices with GPUs or TPUs.
Bottom Line
Recycling startups are proving that AI can unlock value from waste. With aluminum prices high, the economics work. Sortera and Amp are scaling up, and their technology could help the U.S. reduce reliance on imported aluminum.
