Beyond Electrons: The Dawn of Optical Computing

As electronic engineering transitioned from vacuum tubes to semiconductor chips, a similar shift is underway in optical systems. Silicon Photonics, a technology leveraging light for computation, is rapidly gaining momentum as a potential successor to customary electronic processing.

Professor han Sang-yoon, a researcher at the Daegu-Gyeongbuk Institute of Science and Technology (DGIST), is at the forefront of this revolution.His work focuses on ultra-low power optical circuit technology, merging Silicon Photonics with Micro-Electro-Mechanical Systems (MEMS).

Unlocking the Potential of Light-Based Circuits

Ultra-low power optical circuit technology manipulates light within silicon-based semiconductor processes.This is a critical step towards next-generation optical computing, which promises to surpass the capabilities of existing electronic Graphics Processing units (GPUs). The current global GPU market is projected to reach $185 billion by 2027, highlighting the potential impact of a disruptive technology like Silicon Photonics.

It is a technology that integrates the ‘circuit that works with light’ in the semiconductor chip for silicon photonics.

Professor Han Sang-yoon,DGIST

Rather of bulky lenses,mirrors,and optical fibers,this approach uses silicon-based CMOS processes to implement large-scale optical circuits directly on a chip.

addressing the Aggregation Challenge

One of the primary hurdles in optical circuits and silicon photonics is achieving high aggregation,or circuit density.Current technologies achieve less than 1% of the theoretically possible density due to limitations in energy efficiency, signal loss, and physical area.

The biggest limit of optical circuit and silicon photonics is the low aggregation. The current technology, which is less than 1%of the theoretically implemented circuit density, is blocked by various constraints such as energy efficiency, loss and area.

MEMS: A Key to Enhanced Refractive Index

Professor Han is employing MEMS technology to overcome this challenge. MEMS allows for critically important changes in the refractive index with minimal energy expenditure, a crucial factor in driving silicon photonics circuits.

The driving principle of the silicon photonics circuit is to adjust the refractive index,and the MEMS technology can create a very large change of refractive index with a very small energy.

Professor Han Sang-yoon, DGIST

This approach was detailed in his November 2023 paper in Nature Photonics, where he demonstrated a 100-fold reduction in standby power compared to existing technologies, along with a significant increase in photonics circuit size. This work has been recognized as a paradigm shift in Silicon Photonics, earning Professor Han a young scientist award at the Piers Symposium of The Electromagnetics Academy.

Commercialization and the AI Revolution

Professor Han’s research is paving the way for commercial applications.He is currently involved in international collaborations with European IMECs to develop silicon photonics-based light GPUs. The project aims to achieve a 100-fold enhancement in performance and energy efficiency compared to electronic GPUs. This technology is particularly relevant given the rapid growth of Artificial Intelligence (AI) models and the increasing energy demands of data centers. The global AI chip market is expected to reach $83.25 billion in 2030,further emphasizing the importance of energy-efficient computing solutions.

Optical Switches: Enabling AI Supercomputers

Another request of this technology lies in large optical switches. AI supercomputers require high-speed, reliable switches to dynamically manage data flow between tens of thousands of GPUs. Current commercial technologies often require light-conversion steps,which can be inefficient. Professor Han’s large-scale optical switch technology, developed since his time at Dr. Berkeley’s program, offers world-leading integration and size.

A New Era of Computing

Professor Han argues that the progress of electronic circuit technology is slowing down,with innovation primarily focused on recombining existing technologies.He believes it’s time to move beyond electron-centered computing.

AI is a field that was born on the exponential development of electronic engineering. But now, the progress of electronic circuit technology has been noticeably slowed down, and innovation is staying at the level of recombination of existing technology.

Silicon Photonics offers a wholly new computing architecture based on light, enabling fundamentally different physics phenomena.

It is time to get out of the electron -centered computing. Silicon Photonics enables a completely different physics phenomenon, that is, a completely new computing architecture based on light.

Professor Han sang-yoon, DGIST

Complementary to Quantum Computing

While quantum computing is often touted as the future of computing, Professor Han emphasizes that it is designed to solve different types of problems than classical computing. Silicon Photonics, therefore, represents a complementary approach to advancing computational capabilities.

Many people say that quantum computers will be the leaders of the next -generation computing, but it’s a tool to solve a different problem than classic computing.

Professor Han Sang-yoon, DGIST

Korea’s Chance in silicon Photonics

Professor Han believes that Silicon Photonics presents a unique opportunity for Korea to establish a leading position in the semiconductor industry. While the existing electronic semiconductor market is largely dominated by established players, Silicon Photonics is still an emerging field where Korea can make significant contributions.

The existing electronic semiconductor market is already being fixed, and Korea is losing its place in the industrial logic of the great powers, but Silicon Photonics is still being made.

Professor Han Sang-yoon, DGIST