AUTHOR=Atwany Mohammad , Pardo Sarah , Serunjogi Solomon , Rasras Mahmoud TITLE=A review of emerging trends in photonic deep learning accelerators JOURNAL=Frontiers in Physics VOLUME=Volume 12 - 2024 YEAR=2024 URL=https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2024.1369099 DOI=10.3389/fphy.2024.1369099 ISSN=2296-424X ABSTRACT=Deep learning has revolutionized all sectors of industry, but as application scale increases, performing training and inference with large models on massive datasets is increasingly unsustainable on existing hardware. Specialized hardware accelerators like Graphics Processing Units (GPUs) are now widely used to improve speed over conventional Central Processing Units (CPUs). However, Complementary Metal-oxide Semiconductor (CMOS) devices suffer from fundamental constraints, relying on metallic interconnects, which impose inherent limitations on bandwidth, latency, and energy efficiency. Indeed, by 2026, the projected global energy consumption of data centers fueled by CMOS chips is expected to increase by an amount equivalent to the annual usage of an additional European country. To overcome resource constraints, Silicon Photonics (SiPh) devices are emerging as a promising CMOS-compatible alternative to traditional deep learning accelerators, using the efficiency of light to compute as well as communicate. In this review, we examine the prospect of photonic computing as a pathway to accelerating deep learning applications. We present an overview of the photonic computing landscape, then focus in detail on SiPh integrated circuit (PIC) accelerators used for deep learning. We categorize different devices based on their operational principles and relative strengths and present open challenges and new directions for further research.