Impact of AI and Deep Learning on Semiconductor Market Opportunities for Designs
Handel Jones - IBS, Inc., Los Gatos, CA
AI will be disruptive to many application segments, but with many growth opportunities for new semiconductor products, but where new approaches to designs are required because of the complexity of the applications that need to be supported. The emergence and adoption of 5G will provide the high bandwidth and low latency required for autonomous driving as well as the ability to support tens of billions of IoT devices. Many new applications and services will emerge because of the bandwidth provided by 5G.
The migration to ≤10nm will require the development of ultra-high throughput processors along with the supporting algorithms. The cost of designing these processors will be hundreds of millions of dollars, and the result is that with the present approaches to designs, the number of new products will be small. Approaches are needed in order to increase efficiency of designs, and new business models are also needed to obtain higher payback from the design.
There will also be an increasing number of designs in 28nm and derivatives where RF and eNVM functionality will be mainstream. There will also be increased adoption of multichip packaging which will also require continued enhancements in design implementation technologies. The breadth of applications that will need to be supported will increase, and a key limitation will be shortages of design engineers.
It is clear that the value of high-level design expertise will increase significantly during the next decade, but a critical need is also to use AI and computer power to increase productivity of the design engineers.