This is for a few reasons. Firstly, the big monopolies we use and rely on every day are talking a lot more publicly about it, but in the electronics world, we are starting to see a lot more dedicated semiconductor devices being released. Intel's Movidius, Google's TPU2 and similar AI / Deep Learning hardware accelerators were released in 2017, and within the last month or two ARM have announced the release of ML-optimized processor IP which we can expect to see in hard silicon by the end of the year or early 2019.
What we are beginning to see is that this hardware is getting into the hands of engineers and inventors and this generally sets a lot of minds on the topic, and by capillary effect, the rest of society. But for you and me in the PCB end of the electronics business, it means two things: a surge in unique designs as ML applications like vision and smart sensor networks proliferate and increase acceptance of machine learning-based algorithms in CAD.
This is one reason I've been writing lately about the evolution of AI/ML for routing and PCB design - it's not exactly a new application of Neural Networks, but it's now a time when we have the necessary computing resources to make it work well, really well, and be able to dynamically improve by continuous learning.
What do you see on the horizon?
On the horizon I see the application of Machine Learning not only to basic component placement and layout on the design side, but extensive use in predicting supply chain issues, DFM / DFA.
Technology-wise I see 3D printed PCB prototyping becoming more mainstream as competitors enter the market and drive costs lower, and along a similar vein 3D MID (molded interconnect devices) should become more accessible and have additional conductive layers made possible, with smaller trace geometries. So far, the tools available for designing these MIDs are not up to par with regular PCB design tools but as the demand increases the tools will be made available to make it a pleasant design experience. There's certainly no shortage of innovation projects available for us to choose from!
Ben Jordan is a leading EDA Product Marketer with Altium. He started his quest for knowledge in electronics as an 8-year-old with a Dick Smith "Funway 2" book and a soldering iron. He obtained his Computer Systems Engineering degree from the University of Southern Queensland with first class honors, and has a deep passion for electronics design, PCB, embedded systems and guitar.
Share an expert’s perspective - forward this message to your team members, customers and suppliers