Co-Packaged Optics in the AI Data Center: 10-Year Market Forecast and Strategic Analysis: Evaluating Silicon Photonics, Switch ASICs, and Strategies of Nvidia, Broadcom, Intel, and Marvell
Dublin, Aug. 17, 2026 (GLOBE NEWSWIRE) -- The "Co-Packaged Optics in the AI Data Center: A Ten-Year Market and Technology Forecast" has been added to ResearchAndMarkets.com's offering.
New Co-Packaged Optics Market Report Maps the Transition to AI Data Center Deployment
A new co-packaged optics market report examines how CPO technology is advancing to address the bandwidth, latency, power and interconnect requirements of next-generation AI data centers. By placing optical interfaces closer to the electrical die, CPO shortens high-speed electrical paths and supports denser optical fabrics, greater aggregate bandwidth per accelerator and lower energy consumption per bit.
These performance advantages increasingly align CPO with the technical and economic demands of large-scale AI infrastructure. As AI clusters expand across multiple racks, data center operators, accelerator suppliers and system integrators must overcome escalating connectivity and power constraints. CPO is therefore being positioned as an important interconnect technology for scalable AI computing environments.
Despite growing industry momentum, co-packaged optics remains an emerging technology. Product architectures, manufacturing strategies and standards continue to evolve, while deployment schedules vary across suppliers and applications. The report evaluates multiple standards and product scenarios to show how different commercialization paths could influence the global co-packaged optics market.
The report's publisher first introduced a dedicated CPO market study in 2020. Its latest analysis projects that co-packaged optics will become a major interconnect technology within AI data centers worldwide beginning in 2026. The updated report provides a roadmap for the transition to CPO and identifies where business value may emerge across the developing ecosystem.
Key areas covered in the co-packaged optics market report include:
A central objective of the report is to update co-packaged optics market forecasts with detailed breakouts by application, data rate, technology, network segment and data center type. The projections incorporate the impact of generative AI, accelerated computing and rapidly expanding AI workloads on CPO adoption, product development and market dynamics.
The analysis also considers how CPO could provide a practical route for scaling multi-rack AI clusters as conventional electrical interconnects encounter power, reach and density limitations. This opportunity is attracting increased attention from semiconductor companies, optical component suppliers, networking vendors, AI accelerator providers and data center system integrators.
Major AI technology vendors are already developing CPO solutions and refining deployment strategies for AI data center infrastructure. Although initial market activity has focused on anticipated 2025 and 2026 deployments, the report assesses a range of adoption timelines to reflect continuing uncertainty around standards, product readiness, manufacturing complexity and operational requirements.
By combining market forecasts, supplier roadmaps, deployment guidance and application-level analysis, the report offers a comprehensive view of the global CPO market. It is intended to help AI data center operators, technology providers, investors and strategic planners evaluate commercialization opportunities and prepare for the expanding role of co-packaged optics in high-performance AI networking.
Key Topics Covered:
Executive Summary
E.1 Market Environment and Situations Influencing the Market
E.2 Potential CPO Benefits
E.3 Uncertainties Preventing a CPO Boom
E.3.1 Does AI Need CPO and is CPO Ready for AI?
E.4 Technical/Market Evolution of the CPO Market for AI Applications
E.4.1 Adventures in Cooling
E.4.2 CPO Test Equipment: Sales Prior to the CPO Market
E.4.3 Other Technical Issues
E.5 Why CPO is Likely to Succeed - At Least for Now
E.6 Summary of CPO Market Forecasts 2026-2034
Chapter One: Introduction
1.1 Background to this Report
1.2 Goals and Scope of this Report
1.3 Plan of this Report
Chapter Two: Co-packaged Optics: A Technology, Components and Business Opportunity Evolving
2.1 CPO in Traditional AI Data Centers
2.2 CPO for Next-Gen AI Data Centers
2.2.1 CPO Data Rates
2.2.2 Thermal Management
2.2.3 Electrical Interfaces and Channel Loss
2.2.4 Optical Interfaces
2.2.5 Integration Strategies, Manufacturing and CPO
2.3 CPO-related Test Equipment
2.4 Optical Engines
2.4.1 The Broadcom and NVIDIA Approaches
2.5 Fiber, Connector and Couplers for CPO
2.5.1 Thoughts from Broadcom and NVIDIA
2.5.2 Vertical Coupling
2.6 Fiber Solutions for CPO
2.6.1 Multi-core fibers and Fiber Pitch Reduction
2.6.2 Reducing the Pitch
2.7 Lasers for CPO
2.7.1 External Lasers and ELSFP
2.7.2 Module-integrated lasers
Chapter Three: CPO Standards and Implications
3.1 OIF and the Emergence of Co-Packaged Optics
3.1.1 Framework IA (2022)
3.1.2 OIF standards 3.2T CPO Module
3.1.3 External Laser Small Form Pluggable ELSP IA (2023)
3.1.4 CEI-112G / CEI-224G Electrical Interfaces
3.1.5 IAs for CPO Management Interfaces
3.1.6 Future IAs of the CPO Kind
3.1.7 Telemetry and Management
3.2 CPO Standardization in China
3.3 UCIe and CPO
3.4 CPO and Ultra Ethernet
3.4.1 Ultra Ethernet
3.4.2 Advanced Photonics Coalition
Chapter Four: CPO Markets and Forecasts
4.1 Thoughts on the Growth and Size of the CPO Market
4.2 Organic Traffic Growth for CPO Impact on CPO Demand
4.2.1 What Video Taught Data Centers: Bandwidth and Latency
4.2.2 CPO Market by Size of Data Center
4.3 CPO and Hyperscale Data Centers
4.3.1 Hyperscale Data Center Evolution
4.3.2 A CPO Inflection Point by 2032: A Forecast
4.3.3 Penetration of CPO into Switches
4.3.4 Penetration of CPO into Data Servers
4.3.5 Impact on Fiber and Data Center Rack Design
4.3.6 Pricing of CPO Products
4.4 CPO and Conventional Data Centers: Prophesies
4.5 CPO and Edge Data Centers
4.5.1 CPO at the Edge
4.6 CPO Data Center Interconnection
4.6.1 Role of CPO in DCI
4.7 Non-AI Applications for CPO
4.7.1 High-Performance Computing
4.7.2 CPO's Impact on IoT
4.7.3 CPO and Sensors
4.7.4 Disaggregated Compute Systems: Applications for CPO
Chapter Five: End-user Perspectives, Strategies and Guidance: Problems Still to Overcome?
5.1 AI Exaggerates the Importance of CPO!
5.2 Stage One: The Path to CPO - NPO, LPO and "Real" CPO
5.3 NPO (Near-packaged Optics)
5.3.1 LPO (Linear Pluggable Optics)
5.4 Stage Two: Rebuilding the Infrastructure for CPO: Practical Considerations for Processes and Components
5.4.1 Availability of Commercial Component Choices
5.4.2 CPO and the Old Copper vs. Glass Debate
5.4.3 Cooling, Power and Sustainability
5.4.4 ELSFP and the CPO Supply Chain
5.5 Stage Three: Pilot Project, Plugfests and Demos
5.5.1 Multivendor Plugfests
5.5.2 Hyperscaler Internal Trials
5.5.3 Vendor Demos
Chapter Six: Profiles: Suppliers and Influencers
6.1 AMD (United States)
6.1.1 Acquisition of Enosemi
6.2 Ayar Labs (United States)
6.2.1 Product Evolution
6.2.2 Alchip Technologies/TSMC alliance
6.2.3 Partnership with Quantifi Photonics
6.2.4 Partnership with HPE
6.2.5 Ayar the AI Infrastructure
6.2.6 Financing and Collaborators
6.3 Broadcom (United States)
6.3.1 Emerging CPO Product Range
6.3.2 The Bailly Switch
6.3.3 Long-term CPO Strategy at Broadcom
6.4 Ciena/Nubis (United States)
6.4.1 Ciena's Long-term Participation in the CPO Market
6.5 Cisco (United States)
6.6 Coherent (United States)
6.6.1 CPO Product Evolution
6.7 Corning (United States)
6.8 DuPont (United States)
6.9 Furukawa Electric (Japan)
6.10 Google (United States)
6.11 Hengtong Optic-Electric (China)
6.12 Huawei (China)
6.12.1 Huawei Thinking on CPO
6.13 IBM (United States)
6.13.1 IBM CPO-related Innovations in 2024-2025
6.14 Intel (United States)
6.14.1 First Bi-Directional Compute Interconnect
6.14.2 CPO and Multiwavelength Integrated Optics
6.14.3 The Future of CPO at Intel
6.15 Kyocera (Japan)
6.16 Lightmatter (United States)
6.17 Lumentum (United States)
6.17.1 CPO Related Activity
6.18 Marvell (United States)
6.18.1 CPO and the Marvell XPU Architecture
6.18.2 CPO and Marvell Switches
6.18.3 Acquisition of Celestial AI
6.19 Meta/Facebook (United States)
6.20 Microsoft (United States)
6.21 Micas Networks (United States)
6.22 Molex (United States)
6.23 NVIDIA (United States)
6.23.1 Silicon Photonics: Strengths and Architectures
6.23.2 The NVIDIA Switch Family
6.23.3 Emerging Markets for NVIDIA CPO Switches
6.24 POET Technologies (Canada)
6.24.1 The POET Interposer
6.24.2 Light Sources and Collaboration with Semtech and Sivers Semiconductors
6.24.3 Recent Alliance with Quantum Computing Inc. (QCI)
6.24.4 POET Products for AI Data Centers
6.24.5 Manufacturing and Foreign Subsidiaries
6.25 Quantifi (New Zealand)
6.26 Ranovus (Canada)
6.26.1 ODIN Optical Engine and its Versions
6.26.2 Collaborations with AMD
6.26.3 Collaborations with IBM
6.26.4 DARPA, Cerebras Systems and Ranovus
6.26.5 Alliance with Jabil
6.27 SABIC (Saudi Arabia)
6.27.1 EXTEM Resin
6.27.2 ULTEM Resin
6.28 Senko Advanced Components (United States)
6.28.1 Acquisition of Cudoform
6.29 Skorpios Technologies (United States)
6.29.1 The Skorpios Technologies Chip
6.29.2 Applications
6.30 Sumitomo Electric (Japan)
6.30.1 External Lasers
6.30.2 Electronic Integration on Glass
6.30.3 Fiber-array Interconnects for CPO
6.31 TE Connectivity (United States)
6.31.1 The TE Path to CPO
6.32 Teramount (Israel)
6.33 Third-party Transceiver Suppliers and Distributors and the Future of CPO
6.34 Startups, Investors and Pure Players in CPO
6.34.1 Product/Technology Focus of Future CPO Start-ups
Acronyms and Abbreviations Used in this Report
List of Exhibits
Exhibit E-1: Summary of CPO Markets by Type of Data Center ($ Millions)
Exhibit 2-1: AI Data Center Switches Today
Exhibit 2-2: Thermal Management for Future CPO: Technology and Opportunities
Exhibit 2-3: Current Use of Silicon Photonics for CPO
Exhibit 2-4: CPO Testing Functionality
Exhibit 2-5: ELSFP: Possible Project Range
Exhibit 3-1: Future IAs of the CPO Kind
Exhibit 4-1: Types of AI Data Centers in 2025
Exhibit 4-2: CPO Markets by Product Type in Hyperscale AI Data Center
Exhibit 4-3: Switches in Hyperscale Network Racks
Exhibit 4-4: How CPO Changes the Physical Rack Layout
Exhibit 4-5: CPO Markets by Product Type in Non-Hyperscale AI Data Center
Exhibit 5-1 NPO vs. LPO
Exhibit 6-1: CPO Start-Ups
Companies Featured
For more information about this report visit https://www.researchandmarkets.com/r/4h4fo
About ResearchAndMarkets.com
ResearchAndMarkets.com is the world's leading source for international market research reports and market data. We provide you with the latest data on international and regional markets, key industries, the top companies, new products and the latest trends.