Optical I/O for AI Chips Market Size & Share 2026-2035

Optical I/O for AI Chips Market Size

The global optical I/O for AI chips market was valued at USD 1.1 billion in 2025. The market is expected to grow from USD 1.5 billion in 2026 to USD 6.4 billion in 2031 & USD 14 billion in 2035, at a CAGR of 27.8% during the forecast period according to the latest report published by Global Market Insights Inc.

The growth of the market is attributed to the ever-expanding infrastructure of AI data centers, the growing implementation of co-packaged optics (CPO), government initiatives to promote domestic advanced semiconductor manufacturing and silicon photonics technologies, and higher acceptance and usage of high-speed optical networking standards and protocols. In addition, continuous advancements in silicon photonics-based photonic integrated circuits and optical transceivers to enhance scalability and power-efficiency are also likely to aid market expansion.

Optical I/O for AI chips market is driven by growing number of AI data centers, the exponential growth in AI data centers around the world necessitates the use of solutions that can deliver more bandwidth and less power compared to existing electrical interconnects. Major hyperscale cloud companies are investing more on optical interconnects to facilitate extensive AI clusters. For instance, investment in optical interconnect technologies is being driven by governments investing in leading-edge semiconductors. Specifically, U.S. department of commerce CHIPS for America initiative (CHIPS and Science Act) funds domestic semiconductor research, advanced packaging, and silicon photonics to invest in U.S. AI compute infrastructure and the next generation of chips, to be able to boost the speed of optical interconnect technologies for AI data centers.^{[1]US CHIPS, nist.gov}

Additionally, the market’s development is getting a significant boost from public investment in silicon photonics production and semiconductor ecosystem. Nations across the globe are ramping up funding for their homegrown semiconductor supply chains for AI hardware and photonics manufacturing capabilities. For instance, Europe is developing semiconductor ecosystem to support the optical I/O for AI chips market. In 2026, the European Commission endorsed state aid of euro 659 million for four first-ever semiconductor manufacturing projects across the EU under the EU Chips Act. To make matters clearer, the European Commission (EC) initiative aims to solidify Europe’s semiconductor supply chain, boost internal semiconductor output and facilitate innovative tech, such as silicon photonics as well as the next generation AI chips, which will spur the growth of the optical I/O solutions in the market.^{[2]European Chips Act, europe.eu}

The optical I/O for AI chips market increased steadily from USD 122 million in 2022 and reached USD 606.9 million in 2024, as a result of rapid growth of AI data centers, continued investment in HPC, higher investment in CPO, silicon photonics and newer optical interconnect solutions. The growth was also supported by increasing adoption of high speed optical networking by hyperscalers due to energy requirements, progress in PIC technology and government initiatives supporting photonic manufacturing as well as new use cases for ultra-high-speed optical connectivity.

Optical I/O for AI Chips Market Trends

• Co-packaged optics (CPO) adoption is revolutionizing AI infrastructure through the implementation of an optical engine directly into an AI processor and network switch, delivering substantial power reduction and enhanced bandwidth density. The development began gaining substantial traction around 2023 driven by a need from hyperscalers for a new path to augment existing electrical interconnects. CPO will see growth until 2035 as it expands commercially in AI data centers offering the needed scalability for next generation AI clusters dealing with the bandwidth and power limitations in future designs.
• AI chip makers and HPC platforms are also rapidly enabling the implementation of optical I/O due to advancements in silicon photonics. The use of optical I/O took hold around 2022 as chip makers and semiconductor foundries significantly ramped up their investments in photonic integrated circuits (PICs) and optical chiplets. These trends would take place through 2034 with rising capacity and governmental chip programs providing the opportunity for increased speed and reduced latency with better power efficiency.
• Ultra high-speed optical ethernet will transform AI network architectures, an area where 2023 stood out with major advancements occurring with ever increasing complexity of AI models and massive scale GPUs clusters. Expected to continue through 2035 with cloud service providers revamping networking architecture in 800G and 1.6T Ethernet, giving them higher bandwidth and alleviate network bottlenecks in order to boost performance of workloads.
• Growing adoption of optical chiplet architectures with next-gen AI accelerators chip-to-chip interconnect market trend began when the adoption of chiplet interconnect technologies with new age of next gen AI chip and the rise in the importance of connecting optical chiplet-based architecture with new age of AI accelerator systems began around the year 2024 as chip makers were concerned over the shortcomings of electrical interconnect. This would then expand up to 2035 with greater advancement in the space of the heterogeneous interconnect technology, packaging solutions and chip stacking technology and hence facilitate the modular AI system design, scalability and decrease the system power consumption.

Optical I/O for AI Chips Market Analysis

Based on product type, the optical I/O for AI chips market is divided into linear pluggable optics (LPO), co-packaged optics (CPO), near-package optics (NPO), in-package optical I/O (OIO) and on-board optics (OBO).

• The linear pluggable optics (LPO) segment led the market in 2025, holding a 63.2% share. This segment gained traction owing to ease of integration with the current data center infrastructure and cost-effectiveness in migrating to CPO with reduced complexities of deploying such solutions compared to traditional pluggable optics. LPO offer the highest bandwidth with the least power consumption compared to other optical I/O and is well suited for AI workloads. High-speed and low-power consumption offered by LPO makes the LPO segment the preferred choice across hyperscale cloud providers and AI enterprises.
• The co-packaged optics (CPO) segment is anticipated to grow at a CAGR of 40.3% over the forecast period, owing to the increasing demand for ultra-high bandwidth, low latency and energy efficient interconnections to support next generation AI chips and hyperscale data centers for AI applications. CPO improves on the integration benefits of pluggable optical modules by integrating optical engines with switches, AI chips, and other devices at a chip package level.

Based on technology, the optical I/O for AI chips market is divided into silicon photonics (SiPh), indium phosphide (InP), thin-film lithium niobate (TFLN), heterogeneous / hybrid integration and others.

• The silicon photonics (SiPh) segment dominated the market in 2025 and was valued at USD 611.3 million. Silicon photonics have gained significant market traction over the years because of their CMOS process compatibility, higher integration capabilities, and their capacity to produce high-bandwidth and low-power optical interconnects cost-effectively at scale. The broad deployment of SiPH across hyperscale data centers, AI accelerators and high-performance computing systems has strengthened market demand and led to high investment by key semiconductor players.
• The heterogeneous / hybrid integration segment is expected to witness growth at a CAGR of 36.9% during the forecast period as co-integrate silicon photonics with different materials like indium phosphide, lithium niobate, and other semiconductors on a single package. Such integration allows the incorporation of better laser performance, offers high optical efficiency and a higher bandwidth density and reduced power loss. Increased development of the latest AI processors, optical chiplets, and co-packaged optics is also driving the demand for integration of this segment in the AI computing and networking infrastructure.

Based on application, the optical I/O for AI chips market is divided into AI training, AI inference, data center networking & switching and HPC & scientific computing.

• The AI training segment dominated the market in 2025 and was valued at USD 588.7 million. AI training has a larger market share of the market because massive bandwidth and ultra-low-latency capabilities are required to support LLM training, generative AI model training and hyperscale AI model training. This segment is growing because of higher installation of AI supercomputers, GPU clusters and high-speed optical interconnects in hyperscale data centers to support the booming of hyperscale AI applications.
• The AI inference segment is expected to witness growth at a CAGR of 33.5% during the forecast period. Factors such as fast commercialization of generative AI, real-time AI assistants, autonomous systems and edge AI applications with a low-latency requirement will contribute to the rapid growth of the market. Installation of inference accelerators, cloud AI applications, AI enterprise workloads and optical interconnects for the inference scale will fuel the market growth for AI inference segment.

North America Optical I/O for AI Chips Market

North America held a share of 57.8% of optical I/O for AI chips industry in 2025.

• In North America, the market is growing due to presence of major AI chip vendors, hyperscale cloud players and silicon photonics manufacturers. The strong demand for high-speed, low-latency interconnect for AI training and inferencing work load comes from companies such as NVIDIA, Marvell, Ayar Labs, Intel, Microsoft, Google and Meta that increase investments in AI infrastructure.
• With investments from the government and chip makers to promote state-of-the-art semiconductor manufacturing and photonics capabilities such as the US CHIPS and science act and Canada’s semiconductor innovation programs, North America is likely to hold leadership through 2035 with significant AI data center expansion, commercialization of co-packaged optics (CPO), significant R&D and silicon photonics adoption in future AI network infrastructure.

The U.S. optical I/O for AI chips market was valued at USD 71.6 million and USD 160.2 million in 2022 and 2023, respectively. The market size reached USD 602.5 million in 2025, growing from USD 339.7 million in 2024.

• In the U.S growth is notably accelerated by large investments in AI infrastructure, state-of-the-art semiconductor fabrication processes and silicon photonics technologies. This includes increased federal funding from the CHIPS and science act, continued investment in building hyperscale AI data centers, and commitments from key technology giants that continue to expand deployment of optical I/O technology in support of high-performance AI computing.
• For instance, U.S. department of defense led manufacturing innovation institute (USMII) AIM photonics has received recognition for enabling enhanced integrated photonics manufacturing capabilities in the United States. The institute collaboration across industry, academia, and government will foster the commercialization of silicon photonics, photonic integrated circuits, advanced packaging, and talent for leading edge AI, defense, and optical network capabilities.^{[3]US DOD manufacturing USA program, manufacturingusa.com}

Europe Optical I/O for AI Chips Market

Europe optical I/O for AI chips industry accounted for USD 339.3 million in 2025 and is anticipated to show lucrative growth over the forecast period.

• Europe market is expanding due to government backing for the region chip manufacturing capabilities, research into silicon photonics and infrastructure for AI. Various initiatives like the European chips act, horizon Europe, and the chips joint Undertaking (Chips JU) are also fueling Europe investment in optical interconnects, PICs, and advanced packaging to foster its semiconductor industry.
• Europe increased adoption of hyperscale data centers, AI supercomputing, and other advanced optical networking infrastructure are fueling demand for optical I/O technologies in the continent. Investments in CPO, silicon photonics and high-performance, low-power optical interconnect by premier semiconductor organizations, cloud hyperscalers and the EU's leading research establishments are aiding in their commercialization.

Germany dominated the Europe optical I/O for AI chips market, showcasing strong growth potential.

• Germany dominated the European optical I/O for AI chips industry, given the nation significant semiconductor production capabilities, globally renowned research institutes specializing in optics and photonics and substantial government incentives to encourage microelectronics research. Fraunhofer, the Leibniz institute, and leading chip-makers, among others, are actively developing photonic integrated circuits (PICs), silicon photonics, and optical packaging technologies targeting AI and high-performance computing solutions.
• For instance, FMD expanded cooperative research and development activities, BMBF supported and based on its silicon photonics, photonic integrated circuit (PIC), advanced packaging, and heterogeneous integration programs for AI of the future and high-performance computing. The program helps FMD bolster the German semiconductor ecosystem and key technologies such as Optical I/O and Co-Packaged Optics.^{[4]German federal govt, bund.de}

Asia Pacific Optical I/O for AI Chips Market

The Asia Pacific optical I/O for AI chips industry is anticipated to grow at the highest CAGR of 31.2% during the forecast period.

• The Asia Pacific market is witnessing rapid growth due to significant investments being poured into AI infrastructure, semiconductor manufacturing, and hyperscale data centers in China, Japan, South Korea, Taiwan and India. Government incentives that boost the local production of silicon, silicon photonics and advanced packaging are also fueling the development and adoption of optical interconnect solutions for the advancement of AI workloads.
• The region is a hub to prominent silicon foundries, optical component vendors and AI hardware vendors that spearhead innovation in Co-packaged optics (CPO), photonic integrated circuits (PICs), and optical networking. The soaring demand for generative AI applications, cloud services and HPC coupled with increased adoption of 800G and 1.6T networks across Asia Pacific are key drivers propelling the optical I/O market forward.

India optical I/O for AI chips market is estimated to grow with a significant CAGR, in the Asia Pacific market.

• The Indian market presents one of the strongest growth opportunities in the optical I/O technology space given rising government backing for semiconductor manufacturing, the growing infrastructure to support artificial intelligence (AI) deployments and the electronics manufacturing segment. Support from government programs like the India semiconductor mission (ISM) and increasing investments in data centers in the country are driving innovation and investment into sophisticated semiconductor packaging and optical interconnect ecosystems.
• For instance, ministry of electronics & information technology significant initiatives to support AI data centers, semiconductor manufacturing and India semiconductor mission 2.0 to ramp up India capabilities to establish world-class domestic AI computing capabilities, chip design and semiconductor fabrication infrastructure have been provisioned in the union budget 2026-27, which are expected to generate significant demand for next-gen optical networking and optical I/O technologies in AI data centers.^{[5]MEITY PIB, pib.gov}

Middle East and Africa Optical I/O for AI Chips Market

Saudi Arabia optical I/O for AI chips industry to experience substantial growth in the Middle East and Africa.

• Saudi Arabia is witnessing notable growth in the market with increased adoption of AI, cloud computing and digital infrastructure through various reforms as part of Vision 2030. Driven by various initiatives, Saudi Arabia is paving its way to construct hyper-scale data centers, high-end semiconductor designs and AI compute capabilities, thus expanding the use of high-speed optical interconnections and silicon photonics solutions.
• Besides, the strategic collaboration between global tech organizations, coupled with continued investment in AI innovation is further reinforcing the digital and semiconductor ecosystem within the country. The rising demand for high performance computing (HPC) units, AI research centers, and next generation network infrastructure is likely to encourage the usage of co-packaged optics (CPO), PICs and the optical I/O devices.

Optical I/O for AI Chips Market Share

The optical I/O for AI chips industry is led by players such as Marvell Technology, Coherent Corp, Lumentum Holdings, Ayar Labs. These five companies cumulatively accounted for 93% market share in 2025. The major key players provide advanced technology leadership in areas such as silicon photonics, optical interconnects, CPO (co-packaged optics), and high-speed optical networking technology they have established robust partnerships with key hyperscale cloud service providers and the providers of AI infrastructure.

The leading players continue to expand competitive advantage through significant investments in photonic integrated circuits (PICs), high-speed optical transceiver technology, heterogeneous integration, and advanced packaging technology, these efforts are supported by a strong R&D portfolio, large-scale production capacities, and cooperation with top-tier semiconductor companies to cater to the rising demand for increased bandwidth, power efficiency and scalability for AI training clusters, inference systems, and hyperscale data centers.

Optical I/O for AI Chips Market Companies

Prominent players operating in the optical I/O for AI chips industry are as mentioned below:

• Ayar Labs
• Broadcom Inc.
• Celestial AI
• Ciena Corporation
• Coherent Corp.
• Furukawa Electric Co.
• Hamamatsu Photonics
• Intel Corporation
• Lightmatter
• Lumentum Holdings Inc.
• MACOM Technology Solutions
• Marvell Technology Group
• NewPhotonics
• Nexus Photonics
• PhotonicX AI Pte. Ltd.
• POET Technologies Inc.
• Ranovus Inc.
• Salience Labs
• Scintil Photonics
• Sumitomo Electric Industries

• Marvell Technology

Marvell is the top vendor in electro-optics silicon photonics and high-speed interconnects for AI. Marvell’s cutting-edge optical DSPs PAM4 and co-packaged optics solutions will provide ultra-high-bandwidth and low latency interconnect for AI cluster and hyperscale data centers.

• Coherent Corp

Coherent Corp. is an optical transceivers, photonic integrated circuits (PICs), lasers and silicon photonics suppliers for AI networking applications. It provides high-speed 400G, 800G, and upcoming 1.6T optical solutions designed for increasing bandwidth needs of AI training and inference workloads.

• Lumentum Holdings

Lumentum Holdings Inc. is an optical networking, lasers, and photonic components provider that serves the AI data center and cloud networking segments with high-speed networking solutions. Lumentum offers high performance optical engines and transceiver products that increase bandwidth density, energy efficiency and overall network scale and investments in optical network communication technologies.

• Broadcom Inc

Broadcom Inc is a leading provider of AI infrastructure networking solutions. With expertise in high-speed networking, silicon photonics, and optical connectivity technologies, including optical DSP, co-packaged optics (CPO) and optical interconnect technologies.

• Ayar Labs

Ayar Labs, is a leader in in-package optical interconnect and optical I/O technology, aiming to boost performance, efficiency, and scaling of AI accelerators and high-performance computing (HPC). By leveraging optical I/O, silicon photonics and an optical chiplet architecture, Ayar labs technology replaces conventional electrical interconnects with optics.

Optical I/O for AI Chips Industry News

• In April 2026, Ayar Labs introduced its TeraPHY optical I/O chiplet platform, the world’s first to enable large scale optical I/O, with initial samples available commercially to customers targeting AI accelerators and HPC systems, which dramatically reduces the power and latency in AI infrastructure.
• In December 2025, Marvell technology group bought silicon photonics firm celestial AI for USD 3.25 billion, with its photonic fabrics expanded with silicon photonics as it boosts its portfolio to accelerate the transition of the silicon in next-generation AI infrastructure.
• In March 2026, Lumentum Holdings Inc completed a strategic investment of USD 2 billion with NVIDIA that would increase the pace of developing and manufacturing of innovative technologies of optical interconnect for expanding the portfolio of Lumentum’s AI data centers and of the next-generation optical communication networks.

The optical I/O for AI chips market research report includes in-depth coverage of the industry with estimates and forecast in terms of revenue (USD Million) from 2022 – 2035 for the following segments:

Click here to Buy Section of this Report

Market, By Product Type

• In-package optical I/O (OIO)
• Co-packaged optics (CPO)
• Near-package optics (NPO)
• On-board optics (OBO)
• Linear pluggable Optics (LPO)

Market, By Technology

• Silicon photonics (SiPh)
• Indium phosphide (InP)
• Thin-film lithium niobate (TFLN)
• Heterogeneous / hybrid integration
• Others

Market, By Application

• AI training
• AI inference
• HPC & scientific computing
• Data center networking & switching

Market, By End-User

• Hyperscalers & cloud providers
• Enterprise data centers
• HPC & research institutions
• Government & defense

The above information is provided for the following regions and countries:

• North America
  • U.S.
  • Canada
• Europe
  • Germany
  • UK
  • France
  • Spain
  • Italy
  • Russia
• Asia Pacific
  • China
  • India
  • Japan
  • Australia
  • South Korea
  • Taiwan

• Latin America
  • Brazil
  • Mexico
  • Argentina

• Middle East and Africa
  • South Africa
  • Saudi Arabia
  • UAE