by Sundeep Bajikar

Jan 23, 2019

Big Data and Artificial Intelligence (AI) are fueling a new computing revolution, which is transforming many industries from transportation to healthcare. In 2019, the pace of this change will accelerate, creating new challenges and also tremendous opportunities for innovation.

Virtually all electronic devices are becoming smarter and more connected. This is causing an explosion of machine-generated data that now exceeds data generated from humans. The challenge is how to capture, store and process the very large data sets at virtually real-time speed. This will take a lot of storage, along with new types of memory and order of magnitude improvements in computing performance per watt.

But just as momentum is building to drive the AI revolution, classic Moore’s Law scaling is slowing. While 2D shrinking previously delivered the required improvements in PPAC (performance, power and area/cost) for the PC era, the industry now needs new methods of driving chip improvements beyond Moore’s Law. This will involve a combination of approaches: new chip architectures, 3D design techniques, novel materials, new ways to continue shrinking transistors, and advanced packaging schemes to connect chips together in new ways. We call this the “New Playbook” for semiconductor design and manufacturing. Enabling this new playbook will require major advances in materials engineering and greater collaboration across the industry ecosystem.

One of the biggest challenges for the semiconductor industry is how to enable faster processing speeds without raising the power envelope. Today there’s a 1,000X gap in compute performance per watt that must be closed. Lowering power consumption will involve design changes at the chip- and system-level and innovations in materials and structures within the device. This is one of the critical hurdles that must be solved to unlock the potential of AI and Big Data.

New materials and architectures being developed focus on delivering PPAC solutions for the industry. For example, replacing copper with cobalt as a new material in the lower level wiring layers of a device reduces resistance and improves the RC delay problem, which results in higher speed and lower power consumption. A move to horizontal gate-all-around devices reduces transistor leakage providing further improvements in power and performance. Packaging is another key enabling area which provides systems solutions that bring logic and memory together for higher performance and lower power while also addressing form-factor requirements. Stacking DRAM chips together has brought about High Bandwidth Memory (HBM) which combined with logic in interposer designs has delivered major performance gains for AI computing.

With regard to materials, the days of working with individual materials from the periodic table are over. Continued advances in PPAC require new combinations of materials that work together to form novel structures with precise electrical properties. Assembling these new material combinations to achieve predictable properties and interactions is an increasingly complex task requiring Integrated Materials Solutions that can co-optimize multiple process steps, often within the same platform and under vacuum. More and more, engineering materials in vacuum permits precise engineering at the atomic-level.

Applied is expanding its capabilities to help enable the new semiconductor industry playbook. To collaborate with more customers on projects that can speed availability of new materials and process technologies, Applied announced plans for the new Materials Engineering Technology Accelerator (META Center) in November of last year. The META Center will complement and extend the capabilities of Applied’s Maydan Technology Center in Silicon Valley. Applied is also enabling advanced packaging capabilities with a full suite of 300mm advanced wafer level packaging equipment and personnel in Singapore. Our venture capital arm, Applied Ventures, is investing in AI chip startups with disruptive innovations in computing architecture. Last year Applied launched the AI Design Forum at SEMICON West to bring together design and manufacturing experts—from materials to systems—under one umbrella to collaborate at a higher level.

2019 is a pivotal year for semiconductors, where major collaborative efforts and innovations in manufacturing technologies will impact a wide range of industries. By providing breakthroughs in materials engineering, Applied is at the forefront of enabling the technologies shaping the future.

Tags: AI, artificial intelligence, Big Data, Moore's Law, META Center, Integrated Materials Solutions