Two tech companies, Lightelligence and Lightmatter, have recently introduced computer components that utilize laser light to process information. These processors are capable of solving real-world problems faster and more energy-efficiently than traditional computers. These advancements were featured in separate publications in Nature, demonstrating a significant step forward for light-based computing.
The introduction of light-based components signifies a shift in computer technology, as these devices can perform tasks more effectively than traditional electronic chips. Light is used to execute matrix multiplication within these devices, a critical operation in various areas of computing, particularly in AI processing. While other calculations take place in electronic components, the innovation lies in the use of light for computation rather than merely communication.
This development comes at a pivotal time as AI models are becoming increasingly complex, outstripping the capabilities of conventional chips. Moore’s law, which dictated the exponential growth of transistor density, has plateaued due to physical limitations. With the inability to further shrink transistors, there is a need for alternative solutions, such as photonic computing, to enhance computational capabilities.
Both Lightmatter and Lightelligence have unveiled devices that leverage light for processing tasks. To achieve this functionality, these companies have integrated photonic chips within their systems. Lightmatter’s device boasts four photonic chips that execute math operations using light, resulting in faster and more efficient matrix multiplication than electronic chips. Lightelligence’s PACE device combines photonic and electronic chips to accelerate computation for optimization problems essential in various industries.
Accuracy has been a challenge for experimental photonic processors due to the varied output values of light signals. Errors in data transmission can lead to significant discrepancies in calculations. By combining electronic chips with photonic components, Lightmatter aims to mitigate these issues and enhance the reliability of their processor. The practicality of these devices is highlighted by their compatibility with existing manufacturing processes for electronic chips, indicating a seamless transition to large-scale production.
With the potential for rapid scalability and integration into data centers within the next five years, the introduction of light-based processors represents a significant advancement in computing technology. These devices offer a promising solution to the limitations of traditional electronic chips, paving the way for enhanced performance and efficiency in computational tasks. Light-based computing is poised to revolutionize the field of computing, offering new possibilities for AI processing and other applications.
