 |
 |
03 Jun 2025
Chip giant aiming to develop a variety of photonics and co-packaged optics solutions for future AI computing systems.
Advanced Micro Devices (AMD), one of the world’s largest providers of high-performance computer chips, has acquired Enosemi, a US-based startup that has been working on integrated photonics and co-packaged optics (CPO) for interconnect applications in artificial intelligence (AI).
According to a blog post from AMD's senior VP of technology and engineering Brian Amick the two companies had already been collaborating on photonics-related technology, while Enosemi’s advanced chipsets are fabricated by GlobalFoundries - the chip-making business spun off by AMD back in 2009.
“Now as part of AMD, the [Enosemi] team will help us immediately scale our ability to support and develop a variety of photonics and co-packaged optics solutions across next-gen AI systems,” wrote Amick.
“The elite team of experts and PhD-level talent at Enosemi, based in Silicon Valley, has a proven track record of building and shipping photonic integrated circuits [PICs] in volume, a unique feat that few select teams have accomplished.
“Their depth of experience, technical rigor, and track record of execution make them an ideal fit for AMD as we push deeper into high-performance interconnect innovation.”
Optical interconnects
The acquisition will see Enosemi CEO Matt Streshinsky take up a new role as AMD’s senior director of photonics. According to his LinkedIn profile, he co-founded Enosemi in 2023 with CTO Ari Novack and VP of engineering Shahab Ardalan.
Prior to Enosemi Streshinsky spent two years leading Nokia’s silicon photonics effort following its acquisition of Elenion Technologies, which was also co-founded by Streshinsky.
The three Enosemi co-founders had also worked together at Luminous Computing - a Santa Clara startup working on AI computing hardware. When it emerged in late 2023, Enosemi said it had an agreement to license and sell silicon photonics design IP originally developed at Luminous.
AMD reports that it has been building a “best-in-class” portfolio of technologies to address the challenging and rapidly evolving needs of AI, from foundational silicon through to systems-level integration.
“As AI models grow larger and more complex, the need for faster, more efficient data movement is accelerating,” it explained. “To meet these evolving demands, particularly at the rack scale, optical interconnects offer a compelling path forward.
“Co-packaged optics can deliver higher bandwidth density and better power efficiency than traditional approaches, representing a transformative step in system architecture where tighter integration between compute and networking is enabled to support the performance and scale that advanced AI workloads require.”
Jabil collaboration
The computing technology giant adds that future demands of AI systems will require not only powerful chips, but “full-stack innovation” across compute, networking, systems architecture, software and more.
“AMD is uniquely positioned to deliver across this stack, bringing together our industry-leading CPUs, GPUs, and adaptive SoCs with deep networking, software, and system integration expertise,” it claims.
As well as working with GlobalFoundries Enosemi has been collaborating with the contract manufacturing company Jabil, which provides it with advanced photonics packaging services to support higher bandwidth interconnects within AI compute systems.
Earlier this year the two firms announced a deal to package Enosemi’s 1.6 Tb/s photonic chiplets and related design IP incorporating key photonic components needed for AI interconnects such as high-efficiency modulators, modulator drivers, transimpedance amplifiers, and integrated control systems - all fabricated by GlobalFoundries on 300 mm-diameter silicon wafers.
“This collaboration with Jabil will demonstrate the suitability of Enosemi’s chiplet and design IP technology for use within AI compute systems,” said Streshinsky back in March.
 |  |  |  |  |  |  |
| © 2025 SPIE Europe |
|
