Nano Dimension Acquires DeepCube for Machine Learning Integration in 3D Printers

The pair will be using DeepCube’s AI, ML and DL framework to optimize additive manufacturing processes.

Nano Dimension’s AME 3D printing technology. (Image courtesy of Nano Dimension.)

Nano Dimension’s AME 3D printing technology. (Image courtesy of Nano Dimension.)

3D printer and additive manufacturing electronics (AME) developer Nano Dimension recently announced that it is acquiring machine learning (ML) company DeepCube. This partnership will see the integration of DeepCube’s training platform and real-time inference engine into Nano Dimension’s AME 3D printers, to function as the printers’ AI control center as well as the smart nodes for its Smart Fabrication Network (SFN).

Nano Dimension has already signed a definitive agreement and will reportedly pay DeepCube $40 million in cash plus an additional $30 million in American Depositary Shares (ADSs). The ADSs will be retained over varying periods for three years. According to DeepCube, founder and CTO Eli David will be joining the Nano Dimension team as the new AI CTO. The company’s current technical staff, which includes ML experts, defense force veterans, homeland security professionals, former governmental agency employees and former academics, will also be joining Nano Dimension.

“We are excited to join forces with Nano Dimension and transform the AME industry to become fully AI-enabled, with efficiencies, quality, and innovation only possible with deep learning models,” shared David in a press release. “DeepCube’s technology, which accelerates neural networks by a 10x factor, is a natural fit for distributed edge nodes, which are self-learning and self-optimizing—all in real-time and on-demand. The team is incredibly excited for our next step, and we look forward to joining Nano Dimension in driving the Industry 4.0 evolution.”

DeepCube is known for its proprietary deep learning algorithms, which are used for enhanced data analysis and the deployment of complex AI systems. These algorithms have been used for deep learning model training to help accelerate multi-domain neural models—which is ideal for complex and real-time network edge problems. This means the technology can improve inference performance and the delivery of real-time metrics, allowing for automated self-learning and self-optimizing machine infrastructure. According to the company, the DeepCube framework can also be easily deployed through any existing hardware in data centers and edge devices.

Nano Dimension’s decision to acquire this technology primarily involves the company’s plans to develop a machine learning-based Distributed Electronic Fabrication network. This platform will allow manufacturers to shift into a fully digitized production system that lets them have full control over AME and high-performance electronic devices (Hi-PEDs) that are produced via Nano Dimension’s 3D printers.

“The move from 2D to 3D, from legacy processes to a fully digitized and real-time workflow, can only be done with AI as the driver of the full lifecycle process–deep learning models cooperating across the network to learn, optimize and automate the full cycle from design to production,” shared DeepCube CEO Michael Zimmerman. “Unlike the human- and labor-intensive processes of today [for 2D designs], DeepCube will create neural models, trained with varied edge sensor data, and offer customers out-of-the-box neural networks for manufacturing, which accomplish the full AME scale of benefits.”

In other words, DeepCube’s AI, ML and DL system will be responsible for managing the neural network of Nano Dimension’s next-generation edge devices. In addition, this will later allow Nano Dimension’s 3D printers to serve as contact points to a broader network where customers can 3D print electronic devices wherever and whenever they need them at optimized production speeds.

Machine learning models and algorithms have recently seen increased use in 3D printing applications. Besides improved workflow, this is allowing designers to easily print complex parts as well as automate self-learning processes for continuous optimization.

For more information, visit DeepCube and Nano Dimension.