November/December AMP_Digital

A D V A N C E D M A T E R I A L S & P R O C E S S E S | N O V E M B E R / D E C E M B E R 2 0 1 8 3 1 power propelled by AI/ML will require the most advanced semiconductors, packaging approaches, and manufac- turing prowess ever developed in order to reach the high density interconnect required. To enable AI and its building blocks—machine learning, deep learn- ing, and neural networks—new chips (processor and memory) and new ar- chitectures will be needed. In addition, system designs that deliver low power consumption, high performance, low latency, high bandwidth, and fast speed will be required. Inference processing in lieu of traditional program processing will only be achieved by meeting these performance targets. Equally demand- ing is the ability to assess and opti- mize for different types of AI workloads, which provides a business case to jus- tify building custom-designed chips (e.g., ASICs). Recently, the Defense Advanced Research Projects Agency (DARPA) cre- ated a $1.5 billion fund to work on ad- vances in chip technology, dubbed the Electronics Resurgence Initiative. The funding is an exorbitant increase in the hardware budget, focusing on chip design, architecture, materials, and in- tegration. The project also plans to leverage machine learning to substan- tially speed up new chip design. MATERIALS INNOVATION To face the challenges of the AI era, new semiconductor technology will call for materials innovation to de- velop a wide array of new processor ICs and memory chips. This is a demanding technology area and a growing space in business. In the AI era, as global compe- tition heats up and new technologies become available, who will have the upper hand remains to be seen. Up to now, the U.S. semiconductor industry is still in a leading position in AI hard- ware. This is an ongoing global compe- tition among scientists and engineers, companies, and countries. There is a long way to go before reaching the full potential of AI to truly mimic human cognitive capabilities and functions, i.e., asking the right questions at the right time, and solving the right prob- lems in real-time. AI is creating a new paradigm. Ultimately, to best team up human- machine intelligence, we can expect synergistic performance by integrating judgment-focused humans and predic- tion-focused AI agents. AI should aim to augment human cognition and capabil- ities without causing ethical and social issues. That is the real value and virtue of human-machine intelligence team- ing. ~AM&P For more information: Jennie S. Hwang is the CEO and Principal of H-Technologies Group. She may be reached at 216.577.3284 or jennie. hwang@h-techgroup.com .