Series 18 – Episode 6 – memory at the core of innovation

Shih describes memory as the "brain" of technological systems, emphasising its crucial importance in storing and processing data, especially with the rise of AI technologies. He highlights how memory is no longer just a storage component but a key enabler of innovation in autonomous driving, robotics, and industrial applications.

In discussing advances in NAND flash and DRAM technologies, Shih draws parallels between autonomous driving and robotics, noting that both require rapid decision-making, low-latency responses, and high-performance memory systems. He stresses the importance of memory that can quickly process complex AI models without creating bottlenecks in system performance.

The conversation delves into the challenging environments of automotive and industrial sectors. Shih explains that memory controllers must withstand extreme temperatures, operate continuously for potentially 20 years, and maintain consistent performance. Silicon Motion addresses these challenges through advanced firmware, proprietary IPs, and optimised technologies.

Three critical certifications are paramount in automotive memory design:

1. Automotive SPICE (process quality)
2. ISO 26262 (functional safety)
3. Cybersecurity ISO 21434

Shih notes that these certifications ensure rigorous development processes, safety mechanisms, and protection against potential security threats. Silicon Motion has achieved level three in Automotive SPICE and meets ISO 26262 level B requirements.

Looking toward future trends, Shih anticipates significant developments in AI, particularly vertical language models tailored for specific industrial and automotive applications. He predicts memory technologies will continue evolving to support more complex, data-intensive systems while maintaining low power consumption and high performance.

Silicon Motion positions itself at the forefront of this technological evolution, focusing on developing memory controllers that meet stringent certification standards while pushing the boundaries of performance and efficiency.

The podcast concludes with Shih's vision of using vertical AI models to create more efficient, intelligent industrial and automotive systems, ultimately contributing to technological advancement and improved global infrastructure.

To hear more from Steve Shih, you can listen to Electronic Specifier’s interview on Spotify or Apple podcasts.