Academia and industry have enjoyed long-standing collaboration in many scientific fields, especially disciplines of engineering. Despite much success, research conducted purely in industrial laboratories tends to favour patents over publications. However, although this tendency is still strong, it is not an absolute rule. Publishing research from industrial laboratories can directly benefit both industry and academic sectors by fostering collaborations, showcasing innovation, and providing a pipeline for identifying and recruiting top talent.

The act of publishing enables industrial laboratories to collaborate more closely with university researchers, as a finished publication aligns with several academic institutions’ key performance indicators. We believe that having shared goals in this way can bolster potential collaborations. Published works are the premier venue for receiving accolades for discovery, solidifying a laboratory’s reputation as innovation leaders to their communities and collaborators. Furthermore, many of the scientific staff working in industry come from academic backgrounds, and pursuing publications can ease the transition for any future workers from academia to industrial settings. Although preparing, submitting, revising and ultimately publishing a paper requires a considerable amount of time, we believe that these benefits justify the investment.

In this issue, we continue our aim to provide a platform for industrial research and development. Key to scaling up 3D nanodevices are solving challenges in heat dissipation and electron transport. A Perspective by Yuan-Chi Yang et al. highlights the importance of identifying the transport regime in nanoscale devices through statistical fields, which might be diffusive, ballistic or viscous. In addition, a Review by Wei-Yen Woon et al. proposes a roadmap for efficient heat dissipation solutions in 3D integration, exploring thermal management materials, integration challenges and characterization methods.

In addition to device-level research, end-to-end execution of deep neural network models requires a specially designed software stack. A Perspective by Corey Lammie et al. highlights the challenges associated with designing deep learning software stacks for analogue in-memory computing (AIMC)-based accelerators and future research directions to address them.

In optics, deep learning approaches require large training datasets and many computational resources. To mitigate these challenges, Arnaud Verdant and Pierre L. Joly suggest in a Comment that combining light sensing and modulation at the pixel level within a single device can reduce alignment and bandwidth limitations. They propose applications in free-space optical communications, lasers and medical imaging. A further Comment by Sarath Gopalakrishnan et al. outlines the hardware design challenges for 6G measurement tools. These include the incompatibility of hardware used for radiofrequency circuits, the unavailability of test bed environments for both cable-connected and over-the-air measurements, manufacturing and assembly feasibility, and the challenge of handling huge bandwidths in real time. If resolved, measurement tools for 6G communications could become both scalable and marketable.

Together, these works highlight our continued support for industrial research. In doing so, we aim to provide space for constructive multisector collaboration and the sharing of ideas (Nat. Rev. Electr. Eng. 1, 1; 2024).