Industry News
The CPO market will generate $2.6 billion in revenue in 2033
Views : 768
Update time : 2023-03-31 09:55:14
According to data from a report, the revenue generated by the CPO market will reach approximately US$38 million in 2022, and is expected to reach US$2.6 billion in 2033, with a compound annual growth rate of 46% from 2022 to 2033. Even if CPO becomes a mainstream technology, demand for pluggable modules remains high for several applications where CPO is not technically or economically feasible, such as long-haul applications and edge data centers. Pluggable technology is not expected to be phased out in the next 10 years.
Co-encapsulated optics gain traction in high-performance computing
Over the past 50 years, a new generation of technology has emerged in mobile communications every decade. Mobile bandwidth requirements have evolved from voice calls and text messages to ultra-high-definition (UHD) video and various augmented reality/virtual reality (AR/VR) applications. While the COVID-19 pandemic has had a profound impact on the telecommunications infrastructure supply chain, global consumers and business users continue to generate new demand for network and cloud services. Social networking, business conferencing, ultra-high-definition video streaming, e-commerce and gaming applications will continue to drive growth.
With the arrival of 6.4T optical modules in 2029 at the latest, there may be a fierce competition between CPO and pluggable optics. Multiple technical hurdles in the CPO system are expected to be resolved by this time. However, the transceiver industry is constantly innovating to drive the pluggable optics market. Pluggable will take a co-package approach until CPO systems enable volume shipments for networking applications, and optical engines will become increasingly popular in high-performance computing and disaggregated future systems.
According to a report, the revenue generated by the CPO market will reach approximately US$38 million in 2022, and is expected to reach US$2.6 billion in 2033, with a compound annual growth rate of 46% from 2022 to 2033. Projections of rapidly growing training dataset sizes suggest that data will become the main bottleneck for scaling ML models, so we may see progress in artificial intelligence (AI) slow down. Using optical input/output (I/O) in ML hardware can help overcome this bottleneck. This bottleneck is a major driver for the adoption of optical interconnects in next-generation high-performance computing (HPC) systems.
Photonic integrated circuits enable CPO of low-power & low-cost optical interconnects
One expects 800G and 1.6T pluggable modules to be very popular, as they take advantage of 100G and 200G single-wavelength optics, thus enabling technology and cost efficiencies in QSFP-DD and OSFP-XD packages. Pluggable packages will be limited in their ability to support 6.4T and 12.8 densities in terms of required electrical and optical density, thermal management and energy efficiency. Due to the discrete electrical components, power consumption and thermal management are becoming limiting factors for future pluggable optics. Co-packaging using silicon photonics technology platforms aims to overcome the aforementioned challenges.
Fiber is getting closer and closer to the chipset. Bringing data into the point of centralized processing using light is one of the main goals of architects. This trend started a decade ago with the proprietary design of optical components mounted on printed circuit boards (PCBs). The Optical on Board Consortium (COBO) continues these Embedded Optical Interconnect (EOI) ideas and develops specifications that allow the use of optical on board modules in the manufacture of networking equipment. CPO is an innovative approach that tightly integrates optics and ASICs. With today's technology, it is challenging to surround a 50T switch chip with 16 3.2Tbps optical modules, so Near Package Optics (NPO) solves this problem by using a high-performance PCB substrate (an interposer) located on the motherboard, CPO, on the other hand, uses modules to surround the chips on a multi-chip module substrate.
As technology advances, communication and computing technologies are able to be more tightly integrated in business systems, and network hardware is adopting more common components. In addition, the size of AI models is growing at an unprecedented rate, and the chip-to-chip or board-to-board capabilities of traditional architectures (copper-based electrical interconnects) will become a major bottleneck for scaling machine learning. As a result, new extremely short-reach optical interconnects are emerging for high-performance computing (HPC) and its new disaggregated architectures. The disaggregated design differentiates the compute, memory, and storage components on server cards and pools them separately. Through advanced in-package optical I/O technology, optical-based interconnects are used for various processing units (xPUs), especially central processing units (CPUs), data processing units (DPUs), graphics processing units (GPUs), Field-programmable gate arrays (FPGAs) and ASICs, memory and storage, can help achieve the necessary transfer speed and bandwidth.
Data center operators will favor proven low-cost and flexible solutions
Today, the pluggable optical module market supply chain has been well established. It includes discrete or integrated component suppliers, optical companies producing transmitter and receiver optical assemblies (TOSA and ROSA), multiplexers, digital signal processors (DSP) and PCBs, and assembly/test integrators.
It should be pointed out that although mainstream (mainly large cloud operators) have deployed high-end CPO solutions, there are still many small enterprise data centers that have not adopted the latest interconnection technology, so the switching speed between technologies will be relatively slow. This means that even if CPO becomes a mainstream technology, there will still be high demand for pluggable modules for several applications where CPO is not technically or economically feasible, such as long-distance applications and edge data centers. Pluggable technology is not expected to be phased out in the next 10 years. However, the pluggable optics industry is likely to consolidate and the CPO market will develop multi-vendor business models.
Co-encapsulated optics gain traction in high-performance computing
Over the past 50 years, a new generation of technology has emerged in mobile communications every decade. Mobile bandwidth requirements have evolved from voice calls and text messages to ultra-high-definition (UHD) video and various augmented reality/virtual reality (AR/VR) applications. While the COVID-19 pandemic has had a profound impact on the telecommunications infrastructure supply chain, global consumers and business users continue to generate new demand for network and cloud services. Social networking, business conferencing, ultra-high-definition video streaming, e-commerce and gaming applications will continue to drive growth.
With the arrival of 6.4T optical modules in 2029 at the latest, there may be a fierce competition between CPO and pluggable optics. Multiple technical hurdles in the CPO system are expected to be resolved by this time. However, the transceiver industry is constantly innovating to drive the pluggable optics market. Pluggable will take a co-package approach until CPO systems enable volume shipments for networking applications, and optical engines will become increasingly popular in high-performance computing and disaggregated future systems.
According to a report, the revenue generated by the CPO market will reach approximately US$38 million in 2022, and is expected to reach US$2.6 billion in 2033, with a compound annual growth rate of 46% from 2022 to 2033. Projections of rapidly growing training dataset sizes suggest that data will become the main bottleneck for scaling ML models, so we may see progress in artificial intelligence (AI) slow down. Using optical input/output (I/O) in ML hardware can help overcome this bottleneck. This bottleneck is a major driver for the adoption of optical interconnects in next-generation high-performance computing (HPC) systems.
Photonic integrated circuits enable CPO of low-power & low-cost optical interconnects
One expects 800G and 1.6T pluggable modules to be very popular, as they take advantage of 100G and 200G single-wavelength optics, thus enabling technology and cost efficiencies in QSFP-DD and OSFP-XD packages. Pluggable packages will be limited in their ability to support 6.4T and 12.8 densities in terms of required electrical and optical density, thermal management and energy efficiency. Due to the discrete electrical components, power consumption and thermal management are becoming limiting factors for future pluggable optics. Co-packaging using silicon photonics technology platforms aims to overcome the aforementioned challenges.
Fiber is getting closer and closer to the chipset. Bringing data into the point of centralized processing using light is one of the main goals of architects. This trend started a decade ago with the proprietary design of optical components mounted on printed circuit boards (PCBs). The Optical on Board Consortium (COBO) continues these Embedded Optical Interconnect (EOI) ideas and develops specifications that allow the use of optical on board modules in the manufacture of networking equipment. CPO is an innovative approach that tightly integrates optics and ASICs. With today's technology, it is challenging to surround a 50T switch chip with 16 3.2Tbps optical modules, so Near Package Optics (NPO) solves this problem by using a high-performance PCB substrate (an interposer) located on the motherboard, CPO, on the other hand, uses modules to surround the chips on a multi-chip module substrate.
As technology advances, communication and computing technologies are able to be more tightly integrated in business systems, and network hardware is adopting more common components. In addition, the size of AI models is growing at an unprecedented rate, and the chip-to-chip or board-to-board capabilities of traditional architectures (copper-based electrical interconnects) will become a major bottleneck for scaling machine learning. As a result, new extremely short-reach optical interconnects are emerging for high-performance computing (HPC) and its new disaggregated architectures. The disaggregated design differentiates the compute, memory, and storage components on server cards and pools them separately. Through advanced in-package optical I/O technology, optical-based interconnects are used for various processing units (xPUs), especially central processing units (CPUs), data processing units (DPUs), graphics processing units (GPUs), Field-programmable gate arrays (FPGAs) and ASICs, memory and storage, can help achieve the necessary transfer speed and bandwidth.
Data center operators will favor proven low-cost and flexible solutions
Today, the pluggable optical module market supply chain has been well established. It includes discrete or integrated component suppliers, optical companies producing transmitter and receiver optical assemblies (TOSA and ROSA), multiplexers, digital signal processors (DSP) and PCBs, and assembly/test integrators.
It should be pointed out that although mainstream (mainly large cloud operators) have deployed high-end CPO solutions, there are still many small enterprise data centers that have not adopted the latest interconnection technology, so the switching speed between technologies will be relatively slow. This means that even if CPO becomes a mainstream technology, there will still be high demand for pluggable modules for several applications where CPO is not technically or economically feasible, such as long-distance applications and edge data centers. Pluggable technology is not expected to be phased out in the next 10 years. However, the pluggable optics industry is likely to consolidate and the CPO market will develop multi-vendor business models.
Related News
Read More >>
Optical Connectivity: The Engine of AI Scaling
May .08.2026
With the rapid growth of generative AI and large language models, data centers are quickly becoming intelligent computing centers. The growing demand for computing power from AI models is not just about performance.
5G Core & MEC: AI-Driven Growth Through 2030
Apr .17.2026
Global communications infrastructure is currently witnessing a long-awaited "second wave of explosive growth."
6G Outlook: The Future of Next-Gen Wireless Communication
Apr .03.2026
The transition from 5G to 6G is more than just faster internet speeds. 6G will bring new changes to network design, operation, and business models. Wireless communication will become smarter, more efficient, and more energy-efficient.
Ready or Not, Next-Gen Communication Is Here
Mar .20.2026
Standardization efforts for the next generation of mobile communication technologies commenced in 2025. They project commercial deployment to take place around 2030.