Chip Design: Emerging Technologies and Trends to Watch

Semiconductors play an immeasurable role in our daily lives and are essential components in appliances, transportation, infrastructure, and computational devices. Furthermore, chip performance and energy efficiency demands continue to increase with the rise of the cloud, AI, and IoT, creating ongoing advances in the semiconductor industry. Here, we’ll cover some of the latest chip design technologies and tech-sector trends you should know.

IBM’s New Two-Nanometer Chip Technology

IBM recently unveiled the world’s first 2-nanometer chip technology, kicking off a new era of semiconductors. This essential innovation helps advance the state-of-the-art in the semiconductor industry and could achieve 45 percent higher performance with 75 percent lower energy use compared to today’s 7 nm chips. This technology helps increase phone battery life, reduce the carbon footprint of data centers, accelerate processing, allow faster reaction times in autonomous vehicles, and assist in AI tasks such as language translation. 

IBM’s “nanosheet technology” fits up to 50 billion transistors on a surface the size of a fingerprint. The transistor designs are called nanosheets, or gate-all-around transistors, and have been in development since 2017. Nanosheet transistors contain a channel of stacked layers surrounded by a gate on all sides. This architecture provides better current control through the channel while preventing leakage. Smaller transistors allow for faster, more reliable, and more efficient devices that enable core-level innovations for processor designs and allow for AI and cloud computing workloads while supporting hardware-enforced security and encryption. 

Application-Specific SoC Developments

We have seen Apple and Google release custom SoCs—Apple’s M1 and M2 chips in their larger personal computers and Google’s Tensor Processing Unit (TPU)—that offer more efficient AI workloads. Many other industry giants are following similar custom SoC trends. 

Microsoft has invested in chip designs for data centers—among them a class of programmable chips founded on Intel FPGAs for running AI tasks. Furthermore, Microsoft has been developing an ultra-secure coprocessor called Pluton that focuses on security and adds another layer of hardware and software protection. 

Amazon, on the other hand, has been designing a wide range of networking chips and specific processors for its data centers, as cloud computing services based on in-house ARM-server CPUs cost significantly less than the frequently used Intel’s.