Co-packaged optics (CPO)

Co-packaged optics (CPO) is a system architecture where optical transmitter and receiver components are integrated directly onto the same substrate package as the host networking ASIC (typically a switch chip or AI accelerator). This contrasts with the dominant pluggable architecture, where the ASIC drives a SerDes electrical interface across a printed circuit board to a separate pluggable optical module.

Why CPO matters. As datacom data rates scale beyond 800G per port, the electrical interface between switch ASIC and pluggable becomes the dominant power consumer. A typical 800G pluggable dissipates about 15 W, and much of that goes to the DSP and retimer electronics that recover the signal after its trip across the board, on top of the host SerDes power spent driving the trace. As speeds rise to 1.6 Tb/s and 3.2 Tb/s, the electrical interface power scales superlinearly with speed (due to the need for higher-order modulation, more aggressive equalization, and more retimer chips), while pluggable thermal envelopes cannot grow proportionally.

CPO eliminates the long electrical trace by placing the optics directly next to the ASIC, so the retiming electronics can be dropped and the host SerDes can be a short-reach design. Vendors quote large savings in optical-interface power; NVIDIA, for example, states 3.5 times better power efficiency for its photonics switches than for pluggable optics. The saving matters in AI training clusters, where networking is a substantial share of facility power.

Standard architecture. A CPO module typically includes:

• Host ASIC at center (switch chip or accelerator), $\sim$ 800 mm² die
• 8 – 16 optical engines placed around the periphery of the host ASIC, each containing a silicon photonic chip with multiple lasers, modulators, photodetectors
• Substrate or bridge providing high-bandwidth electrical signaling between optics and host
• Common laser-light source (sometimes external, sometimes integrated) delivering CW laser light to each optical engine via fiber
• Fiber array exiting the package: at 51.2 Tb/s, from 256 fibers in total (128 each way, with four 100G wavelengths per fiber) to 1,024 (one wavelength per fiber), plus the fibers that deliver laser light

Industry status (2025-2026). CPO is in the "early production" phase:

• Broadcom has shipped a 51.2 Tb/s CPO switch platform (Bailly) and in 2025 announced Tomahawk 6 – Davisson, a 102.4 Tb/s CPO switch, both with silicon photonic optical engines
• NVIDIA announced Quantum-X and Spectrum-X Photonics CPO switches in 2025
• Marvell and Cisco / Acacia have announced CPO product roadmaps
• Intel has demonstrated CPO prototypes but not yet productized at scale
• Open Compute Project (OCP) has a CPO working group defining standardized interfaces

Key technical challenges.

1. Optical alignment in manufacturing: bonding the fiber array to the silicon photonic engine with sub-μm accuracy at high yield. Active alignment is too slow for CPO production volumes; passive alignment with precision micro-features is the target.
2. Laser placement: lasers are heat-sensitive; placing them next to a 100+ W ASIC creates thermal challenges. Remote laser sources (off-package laser modules feeding the CPO via fiber) are common in production designs.
3. Field replaceability: pluggable modules fail and get hot-swapped in seconds; CPO failure modes require ASIC replacement, dramatically increasing the cost of any single fault. Robust optical engine design with redundancy is essential.
4. Standardization: CPO breaks the pluggable interoperability model. Industry is still negotiating how to specify optical I/O at the package level.

Comparison to pluggable.

Light source options. Two architectural choices:

• Lasers on the optical engine: simpler signal path but bigger thermal challenge
• Remote laser source with CW light delivered to the optical engine via fiber: better thermal management, but adds an external module that itself fails and must be hot-swappable

The remote-light approach is the common choice in current production designs because the laser is one of the most failure-prone components in a photonic system.

Long-term outlook. How quickly CPO displaces pluggables in the largest datacom and AI fabrics is still debated. Pluggables keep the advantages of field replacement and multi-vendor supply, and they will continue to dominate metro, access and lower-end datacom.

The light-source side of the architecture, external laser shelves, wavelength grids, and the comb-laser contenders, is covered in Light sources for co-packaged optics; the architecture ladder that ends at CPO is compared in Pluggable, LRO, LPO, CPO.

The distinction between optical interconnect, which co-packaged optics delivers, and optical computing, which is a separate and less settled bet, is argued in Why light moves data and electrons still compute.

References: C. Minkenberg et al., "Co-packaged datacenter optics: opportunities and challenges," IET Optoelectronics 15, 77 (2021); OIF Co-Packaging Framework (2022) and 3.2T Co-Packaged Module (2023) implementation agreements; Broadcom and NVIDIA product announcements for the 51.2T and 102.4T CPO switches (2024 – 2025).