Co-Packaged Optics (CPO) Book – Scaling with Light for the Next Wave of Interconnect: Introduction

Scale-out and Scale-up CPO, CPO TCO and Power Budgets, DSP Transceivers vs LPO vs NPO vs CPO, TSMC COUPE, MZM vs MRM vs EAM Modulator Deep Dive, CPO Focused companies and CPO Supply Chain

Dylan Patel, Daniel Nishball, Myron Xie, and 3 others

Jan 02, 2026 Paid

Co-Packaged Optics (CPO) has long promised to transform datacenter connectivity, but it has taken a long time for the technology to come to market, with tangible deployment-ready products only arriving in 2025. In the meantime, pluggable transceivers have kept pace with networking requirements and remain the default path thanks to their relative cost-effectiveness, familiarity in deployment, and standards-based interoperability.

However the heavy networking demands that come with AI workloads mean that this time is different. The AI networking bandwidth roadmap is such that interconnect speed, range, density and reliability requirements, will soon outpace what transceivers can provide. CPO will provide some benefit and bring more options to scale-out networking, but it will be central to scale-up networking. CPO will be the main driver of bandwidth increases in scale-up networking for the latter part of this decade and beyond.

Today’s copper-based scale-up solutions, such as NVLink, provide tremendous bandwidth of 7.2 Tbit/s per GPU – soon to be 14.4 Tbit/s per GPU in the Rubin generation, yet copper-based links are limited in range to two meters at most, meaning the scale-up domain world size is limited to one or two racks at most. It is also increasingly difficult to scale bandwidth over copper. In Rubin, Nvidia will deliver another doubling of bandwidth per copper lane through bi-directional SerDes, but doubling bandwidth on copper by developing ever-faster SerDes is a highly challenging vector of scaling that is a slow grind. CPO can deliver the same or better bandwidth density and can provide additional vectors for scaling bandwidth, all while enabling larger scale-up domains.

A starting point for understanding the impetus for CPO is to consider the many inefficiencies and trade-offs when using a transceiver for optical communication. Transceivers can be used to achieve greater link range, but the cage on the front panel of a networking switch or compute tray that transceivers plug into is typically situated 15-30cm from an XPU or switch ASIC. This means that signals must first be transmitted electrically using an LR SerDes over that 15-30cm distance, with the electrical signal recovered and conditioned by a Digital Signal Processor (DSP) within the transceiver before being converted into an optical signal. With CPO, optical engines are instead placed next to XPUs or Switch ASICs, meaning that the DSP can be eliminated and that lower power SerDes can be used to move data from the XPU to the Optical Engine. This can reduce energy required to transmit data by more than 50% when compared to DSP Transceivers - with many aspiring to reduce energy requirements per bit by as much as 80%.

While scale-out CPO solutions like Nvidia and Broadcom’s are garnering more attention and are being closely looked at by end customers, major Hyperscalers are already starting to plan out their scale-up CPO strategy and are committing to suppliers. For instance, Celestial AI is estimating that they could generate a $1B revenue run rate by the end of calendar year 2028 - we believe this will primarily be driven by a CPO scale-up solution shipping with Amazon’s Trainium 4.

CPO focused companies are now well beyond papers, pilot projects and demonstrations and and are making key product decisions such as optical port architecture to solve for high volume manufacturing. CPO for scale-up is now not a matter of if and why, but when and how – how to bring these systems to volume production, and when key component supply chain companies like laser manufacturers can ramp up sufficient production.

This article will present an in-depth discussion on the benefits and challenges of CPO, how CPO architectures work, current and future CPO products, CPO focused companies, CPO-related components and their respective supply chains. This piece is intended to serve as a guide to practitioners, industry analysts, investors and everyone else who is interested in interconnect technology.

 Table of Contents and Guide to this Article

We have segmented the article into five parts – readers are welcome to focus on sections that are most interesting or relevant to them.

In Part 1: CPO Total Cost of Ownership (TCO) Analysis, we start off by analyzing how adoption of CPO changes the total cost of ownership for scale-out and scale-up networks. We think that TCO, reliability, and equipment vendor bargaining power will be the primary considerations for adopting CPO in scale-out networks. We will explore whether CPO is ready for primetime when it comes to scale-out, touching on the data we have so far on solution reliability such as Meta’s CPO scale-out switch study presented at ECOC 2025.

In Part 2: CPO Introduction and Implementation, we will dive deeper into how CPO works. This section will explore the evolution of the market from copper to co-packaged copper and from digital signal processor (DSP) optics to linear pluggable optics (LPO) to CPO and the impetus and arguments for the adoption of CPO. SerDes scaling limits and Wide I/O as an alternative to SerDes – particularly when used in conjunction with CPO – will also be discussed.

In Part 3: Bringing CPO to Market, we will describe critical technologies that will enable CPO to gain traction and come to market. We first discuss Host and Optical Engine packaging and explain TSMC COUPE in detail and why it is emerging as the integration option of choice. Fiber Attach Units (FAUs), Fiber coupling as well as Edge Coupling vs Grating Couplers will be explained thoroughly. We will cover modulator types such as Mach-Zender Modulators (MZMs), Micro-Ring Modulators (MRMs) and Electro-Absorption Modulators (EAMs). This section will end with explaining the core of why CPO is being adopted – the many different vectors for scaling bandwidth with CPO: More fibers attached, using wavelength division multiplexing (WDM) and higher order modulation.

In Part 4: CPO Products of Today and Tomorrow, we will analyze CPO products available on the market today and their associated supply chain. We will start with Nvidia and Broadcom’s solutions before discussing major CPO companies. We cover Ayar Labs, Nubis, Celestial AI, Lightmatter, Xscape Photonics, Ranovus and Scintil, describing each provider’s solution in detail and weigh in on important puts and takes for each company’s approach.

Finally, in Part 5: Nvidia’s CPO Supply Chain, we will conclude this report by describing in detail the supply chain for Nvidia’s CPO ecosystem, naming key suppliers for Laser Sources, ELS Modules, FAU, FAU Alight Tools, FAU Assembly, Shuffle Box, MPO Connectors, MT Ferrules, Fibers, and E/O Testing.