Columbia develops 3D photonics system with ‘record’ performance for AI…

26 Mar 2025

…while Rochester’s RAM Photonics launches fiber array for AI connectivity.

Artificial intelligence systems promise transformative advancements, yet their growth has been limited by energy inefficiencies and bottlenecks in data transfer, according to researchers at Columbia Engineering, part of Columbia University, NY. Now, a Columbia Engineering team has unveiled a solution: “a 3D photonic-electronic platform that achieves unprecedented energy efficiency and bandwidth density, paving the way for next-generation AI hardware.”

The development is described in the Nature Photonics paper 3D Photonics for Ultra-Low Energy, High Bandwidth-Density Chip Data Links, led by Keren Bergman, Charles Batchelor Professor of Electrical Engineering, published on March 21st. The research details a pioneering method that integrates photonics with CMOS electronics to redefine energy-efficient, high-bandwidth data communication.

“In this work, we present a technology capable of transferring vast volumes of data with unprecedentedly low energy consumption,” commented Bergman. “This innovation breaks through the long-standing energy barrier that has limited data movement in traditional computer and AI systems,” said Prof. Bergman.

The Columbia Engineering team collaborated with Alyosha Christopher Molnar, Ilda and Charles Lee Professor of Engineering at Cornell University, to develop a 3D-integrated photonic-electronic chip that boasts a high density of 80 photonic transmitters and receivers within a compact chip footprint.

Revolutionizing AI hardware

This platform delivers high bandwidth (800 Gbit/s) with exceptional energy efficiency, consuming just 120 femtojoules per bit. With a bandwidth density of 5.3 Tbit/s/mm², this innovation far exceeds existing benchmarks, says Columbia. Designed for low cost, the chip integrates photonic devices with CMOS electronic circuits and leverages components manufactured in commercial foundries, setting the stage for widespread industry adoption.

The team said its research “redefines how data is transmitted between compute nodes, addressing the long-standing bottlenecks in energy efficiency and scalability.”

The Columbia Engineering statement added, “By 3D integrating photonic and electronic chips, this technology achieves unmatched energy savings and high bandwidth density, breaking free from traditional data locality constraints. This innovative platform enables AI systems to efficiently transfer vast volumes of data, supporting distributed architectures that were previously impractical due to energy and latency limitations.”

“The resulting advancements make this technology a cornerstone of future computing systems across applications, from large-scale AI models to real-time data processing in autonomous systems. Beyond AI, this approach holds transformative potential for high-performance computing, telecommunications, and disaggregated memory systems, signaling a new era of energy-efficient, high-speed computing infrastructure.”

Automatically-produced fiber array for AI connectivity

RAM Photonics Industrial, an automated photonics manufacturer, based in Rochester, N.Y., has announced the launch of its Medusa Array range. This scalable 2D fiber array, produced in a high-speed, automated manner, eliminates any need for manual assembly and testing, the company stated.

RAM describes the array as “the first mass-produced optical interface that can be scaled beyond 1024 fiber ports with negligible insertion loss to meet next-generation requirements today. The family of products addresses the need for high-volume, 2D optical-to-electronic interfaces in optical circuit switching and co-packaged optics.

John Marciante, CEO of RAM, said, “AI, data center, and high-performance computing infrastructure expansion depend on industry’s ability to mass-produce photonics-to-electronic CPO and OCS connectors. Medusa meets this need by epoxy-free integration of submicron-precision 2D fiber arrays and semiconductor or MEMS backplanes.”

• The company will be exhibiting at OFC 2025, in San Francisco, between April 1–3 (booth #3357).