29Dec, 2023
    Global AI Server Market Surges to $50 Billion in 2023, Expected to Exceed 50% Share by 2027

    Recently, market research organization IDC released its latest research report on the global server market. The report indicates that in 2022, the global server market size experienced a year-on-year growth of 20.1%, reaching $123.224 billion. It is anticipated that the market size will see a slight increase to $128.471 billion in 2023. The annual growth rates for the following four years are estimated to be 11.8%, 10.2%, 9.7%, and 8.9%, respectively. By 2027, the market size is projected to reach $189.139 billion.

    Specifically, in 2022, global x86 server sales amounted to $110.955 billion, marking an 18.8% year-on-year increase. In contrast, non-x86 server sales surged by 32.8% year-on-year to $12.269 billion. In 2023, the growth in the global server market slowed down, with x86 server sales experiencing a mere 2.3% year-on-year increase to $110.955 billion. However, non-x86 server sales maintained a high year-on-year growth of 22.1%, reaching $14.984 billion. According to IDC’s projections, by 2027, x86 server sales will escalate to $165.568 billion, while non-x86 server sales will also increase to $23.571 billion.

    Looking at the broad pattern, although non-x86 servers have sustained continuous rapid growth over the coming years, x86 servers remain the mainstream in the server market, maintaining a sales-based share of over 87.5%. Even by 2027, the market share of non-x86 servers will still be slightly under 12.5%, barely up from 11.66% in 2023, an increase of less than 1 percentage point.

    However, significant shifts are occurring within the non-x86 server market segment. Over the past decade, around half of the non-x86 servers were comprised of IBM’s Power Systems and System z mainframes, with the rest being a combination of other proprietary servers and Arm servers. But with the rise of Arm servers among hyperscale vendors and cloud builders, the portion led by Arm in the non-x86 server segment is growing. Furthermore, with the emergence of RISC-V servers, this market segment is expected to undergo further changes.

    Another dataset from IDC reveals that in 2020, IBM’s System z and Power revenues amounted to $4.98 billion. This suggests that in 2020, non-x86 server product sales, including Arm servers and other processor architectures, totaled $3.87 billion. Assuming certain scenarios for future IBM products, such as the upgrade cycles for Power10 and z16 machines in 2021 and 2022, and Power11 and z17 machines in 2025, it’s possible to anticipate a gradual decline in IBM server hardware sales from slightly below $5 billion in 2020 to $3.5 billion in 2026, potentially reaching $3.3 billion by 2027. Based on IDC’s data, other types of servers within the non-x86 segment, like Arm/RISC-V, are expected to grow at a robust pace, subject to investment cycles of hyperscale enterprises and cloud builders, maintaining a benchmark level. Inevitably, if adoption occurs, the annual sales of Arm and RISC-V servers (mostly Arm servers) could reach around $20 billion.

    Moreover, due to numerous hyperscale enterprises and cloud builders focusing on customizing Arm server CPUs and AI coprocessors, there’s a wide range of choices available. This trend faces pressure not only from using X86 server CPUs and Nvidia GPUs for AI and other compute-intensive workloads but also from the widespread adoption of customized Arm servers.

    Considering this, how AI server sales (primarily for training but also for inference) differ from sales of other types of servers?

    IDC mentioned in its report: “In each subsequent quarter of 2022, inflation had a stronger direct impact on servers, with the average selling price (ASP) year-on-year growth rate rising to 29% in the second quarter of 2023, while server shipments mostly hovered around low double-digit year-on-year growth for most of 2022 but declined by 1.4% year-on-year in the fourth quarter of 2022 and by 10% in the first quarter of 2023, and now in the second quarter of 2023, it has again fallen by 19.9% year-on-year.”

    Compared to the significant decrease in shipment volumes, the continuous rise in average selling price is primarily driven by very expensive AI training and inference nodes (each node costing hundreds of thousands of dollars with four or eight GPUs). AI servers and non-AI servers indeed need to be segregated as they represent very distinct segments in the market. Therefore, The Next Platform, based on IDC’s revenue forecast for servers from 2020 to 2027 (IDC data indicates the global AI server market size was $11.2 billion in 2021), attempted to classify them as follows:

    The Next Platform believes that unless something slows down the growth of AI models or makes AI training and inference computations cheaper, it’s reasonable to expect that by 2026 or 2027, AI servers will account for approximately half of the entire server market revenue.

    The model assumes a steep decline in non-AI server revenue by 11.2% starting in 2022, followed by moderate growth similar to GDP every year. By 2023, this situation is expected to improve, with only a 5% decline. Between 2022 and 2023, AI server revenue will experience an astonishing nearly fivefold leap, reaching around $50 billion in the global AI server market size by 2023. Subsequently, there will be a healthy and stable growth of around 20% in 2024, tapering off to 15% by 2027. According to this predictive model, AI server sales will surpass traditional server sales by 2027.

    Currently, AI applications are experiencing explosive growth. With the increase in Nvidia GPU supply and price reductions, as well as the entry of GPUs from other brands and other types of AI accelerators gaining traction in the market, everything is expected to stabilize and normalize, possibly reaching an entirely new level.

    One question remains: how much supply does the world’s AI need? Predicting the situation four or five years down the line is indeed very challenging. If the demand for AI accelerators continues to outpace supply, and prices remain high, revenue will stay elevated. If production doubles or triples and prices decrease by half or two-thirds, then revenue might stay unchanged.

    Clearly, no manufacturer would want to lower profit margins through excessive sales, but intense competition might force companies into that position.

    From 1985 to 2000, it took fifteen years for RISC/Unix machines and internet technology to capture a 45% share of server market revenue through active upgrades in mainframes and proprietary small-scale machines. It might take a similar fifteen-year span — from 2010 to 2025, or from 2011 to 2026 — for AI servers to account for around 45% of global server revenue, with AI workloads replacing or expanding into various applications you can imagine.

    26Dec, 2023
    Revolutionizing Data Centers: GIGALIGHT’s Advanced 800G Solutions Reshape Optical Connectivity for AI

    GIGALIGHT released a marketing report highlighting the imminent transformation in data center network architecture driven by the burgeoning adoption of AI applications recently. This shift has spurred investments in data center infrastructure, propelling growth in the 400G/800G optical module market.

    With the release of several 800G switch chips and the gradual deployment of 800G optical modules, the focus now lies in enabling data centers to seamlessly operate at this high 800G rate. This involves selecting the right optical components and cables while considering crucial factors such as bandwidth, distance, and cost.

    Given that AI heavily relies on parallel computing, the article proceeds to explore parallel wiring solutions for 800G data centers, specifically delving into DACs (Direct Attach Cables), AOCs (Active Optical Cables), and optical modules.

    800G Data Center DAC & AOC Solutions

    When it comes to ultra-short-distance applications, DACs and AOCs emerge as preferred choices. These modules directly link to cables, offering easy installation and cost efficiencies compared to optical module solutions. DACs, notably the most cost-effective, have shorter connection distances. Both DACs and AOCs are available in two variations: direct and breakout connections.

    Using DAC as an example, the maximum transmission distance is 2m (PCC)/5m (ACC), replaced by AOC extend up to 100m.

    800G Data Center Multimode Optical Module Solutions
    Multimode optical module connectivity mirrors AOC solutions, both capable of achieving connections up to 100m. The difference lies in the requirement for multimode optical modules to be paired with MPO multimode fibers.

    800G Data Center Single-mode Optical Module Solution
    For long-distance 800G data center setups, GIGALIGHT employs silicon photonics. Compared to conventional 8-lane EML-based 800G optical modules, silicon photonics modules require fewer lasers. For instance, the 800G DR8 utilizes just one laser to achieve lower power consumption without compromising performance. Besides the 500m reach of the 800G DR8 optical module, GIGALIGHT has also introduced the 800G DR8+ for 2km transmission and the 800G DR8++ for 10km transmission, available in OSFP and QSFP-DD packaging options.

    GIGALIGHT offers QSFP-DD packaging options for all the aforementioned solutions.

    The deployment of 800G data centers demands careful consideration of various cable types and their specific characteristics. Different 800G wiring solutions come with distinct advantages, and the choice should align with the specific requirements of the data center.

    Explore more interconnect architectures for AI applications at https://www.gigalight.com/news-events/reports-7329.html

    21Dec, 2023
    GIGALIGHT’s Years of 8-Channel Optics Mastery: Elevating 800G SR8/DR8 Module High-Temperature Performance and Channel Consistency

    Shenzhen, China, Dec. 21, 2023 – As an expert in long-term research on 8-channel optics, GIGALIGHT has finally won an opportunity in the 800G optical module.

    It’s commonly understood that the efficacy of 800G optical modules hinges on the intricate interplay of 8-channel optical signal integrity technology, precise optical design, and meticulous process form factor. The realm of 8-channel optics has historically received less attention within the industry, primarily due to considerations such as productivity and cost. However, this technology represents a pivotal milestone in optical field advancement, serving as an unavoidable threshold toward the future. While attempts have been made, notably with technologies like 100G/200G PAM4, to circumvent 8-channel optics, such moves may not align with strategic foresight. Looking ahead, it’s important to recognize that wavelength division multiplexing in optics offers limitless potential, unlike the finite nature of wavelength division multiplexing in electronics. For nations grappling with underdeveloped semiconductor technologies, priority should be placed on the advancement of analog-dominant wavelength division optical technologies.

    The intricacies of 800G optical modules lie in ensuring channel consistency and uniform high-temperature performance. Targeted at AI-intensive applications, these 800G modules effectively return to the established benchmarks of telecommunications-grade reliability in optical modules. To ensure network robustness and mitigate potential issues, support for objective progress becomes pivotal. Past trends saw data centers substantially reducing reliability demands for data center interconnects—a logical move but one that also posed detriments.

    800G SR8 Parameters

    GIGALIGHT 800G SR8 supports QSFP-DD and OSFP two form factors. It integrates high-performance 112Gbps VCSEL lasers and 7nm DSP. The electrical host interface operates at 112Gbps PAM4 signal per channel, supporting the CMIS 4.0 protocol version. The module achieves a TDECQ of less than 3dB per channel, while the OMA RXsen sensitivity meets -5.2dBm @ 2.4E-4 Pre-FEC 53.125GBd, supporting transmission distances of 60 meters (OM3 MMF) and 100 meters (OM4 MMF).

    70 °C
    ChannelOptical PowerOuter ERTDECQSensitivityOMA Sensitivity
    CH13.23.041.89-4.6-6.32549390611992
    CH23.22.951.70-4.63-6.47771814241623
    CH34.022.931.65-4.6-6.47654142620682
    CH43.692.962.02-4.22-6.04319970127156
    CH53.453.061.56-4.88-6.58181898883739
    CH63.383.051.56-4.79-6.50100664410559
    CH73.332.931.31-4.83-6.69690734712414
    CH83.562.991.43-4.77-6.55808508046488

    GIGALIGHT advocated for the continued development of three technologies at the June 2021 Optical Communication Industry Summit Forum with the theme of “Optical Linking Everything, Colorful Future (click to review details, The 400G QSFP-DD LR8 Optical Module designed by GIGALIGHT won the CFCF 2021 Product Innovation Award ):

    • 8-channel optics
    • Optoelectronic collaboration in 100G PAM4 technology
    • Commitment to silicon photonics

    Currently, GIGALIHGT has put into small-scale production and deployment globally the following 8-channel optical products within segmented markets.

    This is the golden age of 800G. Regardless of the ultimate direction of AI, let us contribute the most credible products for computational interconnection!

    About GIGALIGHT

    As an open optical networking explorer, Gigalight integrates the design, manufacturing, and sales of both active and passive optical devices and subsystems. The company’s product portfolio includes optical modules, silicon photonics modules, liquid-cooled modules, passive optical components, active optical cables, direct attach copper cables, coherent optical communication modules, and OPEN DCI BOX subsystems. Gigalight focuses on serving applications such as data centers, 5G transport networks, metropolitan WDM transmission, ultra-HD broadcast and video, and more. It stands as an innovative designer of high-speed optical interconnect hardware solutions.

    18Dec, 2023
    High-Speed Optical Module Demand Soars: AI Computing and Market Projections Drive Innovations

    Discovering the intersection of AI computing and escalating market trends, the reliance on optical modules has surged. From high-scale computational scenarios in AI-powered systems to market forecasts propelling technological advancements, the landscape is evolving. Let’s delve into the interplay of AI, market projections, and groundbreaking optical innovations, reshaping the future of computational networking

    Section 1, AI Computing Scenario Optical Module Application

    In the usage of NVIDIA’s A100 systems, such as within the NVIDIA DGX SuperPOD, concerning the application of optical modules required for computation and storage sides, all cables for both computation and storage sides utilize NVIDIA MF1S00-HxxxE series AOC active optical cables. Consequently, each port requires an optical module, meaning each cable corresponds to two optical modules. Therefore, the computation and storage sides total required amounts to (1120+1124+1120)*2 + (280+92+288)*2 = 8048 optical modules. In other words, the approximate quantity of 200G optical modules needed per GPU is about 1:7.2.

    DGX GH200 supercomputer, equipped with 256 NVIDIA GH200 Grace Hopper super chips, each of which can be regarded as a server, interconnected through NVLINK SWITCH. Structurally, the DGX GH200 adopts a two-tiered fat tree topology, with the first and second tiers employing 96 and 36 switches, respectively. Each NVLINK SWITCH has 32 ports operating at a speed of 800G. Additionally, the DGX GH200 is equipped with 24 NVIDIA Quantum-2 QM9700 IB switches for the IB network.

    For estimation based on ports, given the closer proximity of L1 tier, it’s assumed that copper cables are used for connectivity without involving optical modules. For the L2 tier with 36 switches in a non-converging fat tree architecture, each switch’s ports connect downwards to the uplink ports of L1 tier switches. Therefore, a total of 36 * 32 * 2 = 1152 units of 800G optical modules are needed. In the IB network architecture, the 24 switches require 24 * 32 = 768 units of 800G optical modules. Thus, the DGX GH200 supercomputer requires a total of 1152 + 768 = 1920 units of 800G optical modules, corresponding to 12 units of 800G optical modules per GH200 chip.

    Section 2, Optical Module Market Demand Forecast Driven by Computational Networking

    The significant surge in computing power demand driven by new-generation information technologies like cloud computing, artificial intelligence, and big data has hastened the development of cloud computing infrastructure. Concurrently, with the comprehensive launch of projects such as the “Eastern Data and Western Computing,” China’s computational infrastructure construction is rapidly advancing. Optical modules, as pivotal components in cloud computing data centers, will experience a continuous surge in market demand due to the substantial increase in data transmission volume.

    According to LightCounting’s projections, the global adoption rate of 800G technology is expected to be merely 0.62% by 2023. Large AI models like ChatGPT impose new demands on data flows within and outside data centers, potentially accelerating the transition of optical modules towards 800G. It’s anticipated that 800G optical modules will start dominating the market by the end of 2025.

    Based on LightCounting’s data, the global market size for optical modules increased from $58.6 billion in 2016 to $66.7 billion in 2020. Projections suggest that by 2025, the global optical module market will reach $11.3 billion, 1.7 times the size in 2020. Structurally, the data communication market is expected to hold around 60%, while the telecommunications market will encompass approximately 40% of the market share.

    GIGALIGHT Optical Modules for Computational Applications

    The core technology of GIGALIGHT’s silicon optical modules lies in its innovative packaging design with high-coupling efficiency in the free-space COB and the MZI software locking algorithm. Additionally, in terms of silicon waveguide cores, the company has engaged in collaborative design work for multiple silicon chip designs with partners.

    11Dec, 2023
    Cloud Storage: Transforming Data Management for Agility

    During the digital era, data is increasingly becoming the cornerstone asset for enterprises, experiencing exponential growth in volume. This surge in data has led to high costs for businesses using traditional storage technologies, prompting the emergence of a new approach to storage — cloud storage.

    Cloud storage involves enterprises and individuals leasing third-party storage space at a certain cost for data storage. Through network technology or distributed file systems, cloud storage aggregates various types of storage devices within the network through application software, collaboratively providing data storage and business access functions. Core technologies of cloud storage encompass virtualization, distributed storage, Software-Defined Storage (SDS), and Hyper-Converged Infrastructure (HCI) storage.

    Traditional storage technologies evolved to meet the demands of enterprise data centers, databases, enterprise applications, and virtualization integration. They fulfilled traditional enterprise application requirements for reliability, performance, capacity, and business continuity. However, in the face of cloud computing and the big data era, scalability and concurrent processing performance have become somewhat limited.

    Active suppliers globally offering traditional storage solutions include DELL-EMC, NetApp, HPE, Huawei, HDS, IBM, Fujitsu, and others.

    The emergence of technologies like big data, cloud computing, and virtualization has made traditional IT architectures inadequate to meet enterprise data storage needs. Hence, SDS and HCI infrastructures have emerged and expanded from small and medium-sized enterprises to large corporations.

    According to Fortune Business Insights, the global cloud storage market reached $702 billion in 2021. As the cost advantages of cloud storage solutions become more apparent, it will drive enterprises to further adopt cloud storage. The global cloud storage market is estimated to reach $834 billion in 2022, with an expected growth to $376.4 billion by 2029, at a compound annual growth rate of 24.0%.

    Moving forward, Software-Defined Storage (SDS), distributed storage, and Hyperconverged Infrastructure (HCI) will continue their upward trajectory, becoming pivotal for the future of cloud storage.

    Software-Defined Storage:

    Firstly, the significant trend in SDS development revolves around the application of flash memory. SDS development relies on hardware advancements, wherein the rapid progress of generic hardware lays the foundation for SDS ascension. Utilizing and optimizing flash memory assists SDS products in handling large-scale workloads for enterprise users. To meet the digital demands of enterprise users with lower costs and superior performance, SDS solution providers enhance product performance through continuous technological advancements, leveraging innovative storage technologies like NVMe-oF and 3D XPoint.

    Secondly, SDS will decouple storage system software and hardware. This allows generic hardware to replace proprietary hardware, reducing the threshold for storage system utilization. The rise of open-source frameworks accelerates the expansion of open SDS projects, with open-source initiatives like OpenStack and RedHat leading the way in open-source SDS development. In the control plane, applications like OpenSDS and OpenStack Cinder have matured. In the data plane, high-performance, highly reliable, and scalable open-source software like Ceph, GlusterFS, and Swift find extensive industry applications.

    Faced with new storage demands like massive unstructured data, traditional storage systems exhibit drawbacks of low flexibility, inadequate scalability, and high costs. Users are gradually inclining toward open-source distributed storage.

    Open-source SDS possesses several advantages:

    1. Driven by the open-source community, developers and contributors strive for innovation, technological implementation, and building solution ecosystems, actively promoting the development of open-source SDS technology.

    2. Enterprises can use enterprise-grade technology with higher quality, stability, and security compared to proprietary systems at lower costs.

    3. Open-source SDS decouples storage systems from hardware, allowing users to freely deploy storage platforms and flexibly expand storage devices. Utilizing proprietary storage systems binds users to a single vendor, whereas the use of open-source software enables users to freely choose vendors, realizing the true potential of software-defined storage.

    Distributed Storage:

    Distributed storage systems disperse data storage across multiple independent devices. Distributed network storage systems use a scalable system structure, utilizing multiple storage servers to share storage loads and location servers to locate storage information. It not only enhances system reliability, availability, and access efficiency but also facilitates scalability. Characteristics of distributed storage systems include:

    1. Scalability: Distributed storage systems can scale to clusters of several thousand servers, with the overall system performance growing linearly.

    2. Low cost: The automatic fault tolerance and load balancing features of distributed storage systems enable them to be constructed on low-cost servers. Additionally, linear scalability reduces server costs, enabling automatic operation and maintenance of distributed storage systems.

    3. High performance: Both for individual servers and the entire distributed storage cluster, distributed storage systems require high performance.

    Moreover, similar to how not all SDS is necessarily distributed, distributed storage systems are not always software-defined; they might be hardware-bound. For instance, IBM XIV storage, essentially a distributed storage, is delivered through dedicated hardware, thereby retaining hardware bindings and presenting high-cost issues.

    Hyper-Converged Infrastructure (HCI):

    Hyper-convergence refers to the process of integrating storage, computing, and network connection resources into a single IT framework system, aiming to reduce data center complexity and enhance scalability. This infrastructure amalgamates many advantages of Software-Defined Storage (SDS) and distributed storage. It represents a novel deployment architecture in a virtualized environment by internalizing external storage arrays within servers, sparking an industry shift that profoundly influences the existing IT landscape. Embracing the “hyper-converged architecture” is envisioned as a rational mode for deploying fundamental systems. The hyper-converged architecture is more encompassing than the mere software-defined storage hierarchy, as it covers a wider spectrum and to some extent encapsulates software-defined storage.

    However, hyper-convergence is still in a developmental phase. Despite rapid market expansion, the involvement of numerous vendors and the absence of standardized hyper-converged products pose challenges. Many products currently in the market are relatively new and necessitate continual iterations to enhance stability and sustainability.

    The diverse needs of hyper-converged customers range from requiring substantial storage capacities to necessitating high-computing systems. Yet, existing hyper-converged products often remain too singular and fail to meet the needs of all customers. Moreover, integrating hyper-converged systems with traditional legacy devices poses challenges, with issues concerning the seamless integration of old and new systems remaining a critical concern.

    Conclusion

    The versatility of cloud storage has opened a new chapter in providing organizations with more efficient, scalable, and adaptable data management solutions. The continued evolution of cloud storage is not just a technological shift, it is a transformative force that is propelling organizations into a more agile and competitive future in a data-driven era.

    5Dec, 2023
    GIGALIGHT Showcases DWDM Transmission System and Industrial-Grade Triple-Proof Hardened Modules at National Defense Exhibition

    Shenzhen, China, Dec. 5, 2023 – The 11th Chinese (Shenzhen) Defense Information Equipment & Technology Exhibition and China (Shenzhen) military and civil dual-use scientific and Technological Equipment Expo will be held from December 6th to 8th at the Bao’an New Hall of Shenzhen World Convention and Exhibition Center. GIGALIGHT will showcase solutions such as Wave Division Multiplexing (DWDM) transmission systems and industrial-grade hardened optical modules at booth E702 in Hall 3. We cordially invite industry professionals to visit our booth for a closer look. We cordially invite all attendees to visit our booth for deeply coordinate and insights exchange.

    Military communication encompasses the comprehensive use of various methods to transmit information within the military domain. It serves as a pivotal support system during military operations and remains a crucial tool in military confrontations. Following dedicated years of research and development, GIGALIGHT has crafted several products designed specifically for military communication needs. The showcased products include:

    GIGALIGHT 1G/10G hardened optical modules have been manufactured and delivered in bulk quantities for low-speed product lines. The newly launched 100G QSFP28 SR4/PSM4/CWDM4 hardened optical modules are designed to support both industrial and commercial applications. They have undergone rigorous testing, including a 95% humidity reliability test, which they have successfully passed. Additionally, they have been assessed by a domestic authority,CTI, meeting several standards: the waterproof and dustproof property test meets the IEC 60529 standard IP67 level, the anti-mold property test meets the IEC 60068-2-10 standard, a duration of 28 days overall mold-free and salt-spray proof feature complying with GR 487 standards for neutral salt spray corrosion resistance.

    GIGALIGHT’s DWDM transmission series includes both coherent and non-coherent modes, providing a comprehensive solution with multiple options for ultra-long-distance transmission. The exhibited 6.4T DCI BOX 2.0 coherent WDM transmission subsystem can accommodate 400G QSFP-DD ZR coherent optical modules and multiple service cards, offering all-service access, high capacity, and simplified operations. The non-coherent 10G/25G DWDM transmission equipment exhibited presently boasts advantages, including cost-effectiveness, minimal power consumption, low latency, and simplified deployment.

    For more product information, visit our booth at the exhibition. We will have live demos of multiple products, and we sincerely invite industry peers to visit GIGALIGHT’s booth (Booth E702 in Hall 3) to experience our advanced technology and solutions firsthand. GIGALIGHT’s technical team will provide live demonstrations and solution consultations during the exhibition, sharing the latest trends and innovative applications in optical communication technology.

    About GIGALIGHT

    As an open optical networking explorer, Gigalight integrates the design, manufacturing, and sales of both active and passive optical devices and subsystems. The company’s product portfolio includes optical modules, silicon photonics modules, liquid-cooled modules, passive optical components, active optical cables, direct attach copper cables, coherent optical communication modules, and OPEN DCI BOX subsystems. Gigalight focuses on serving applications such as data centers, 5G transport networks, metropolitan WDM transmission, ultra-HD broadcast and video, and more. It stands as an innovative designer of high-speed optical interconnect hardware solutions.

    1Dec, 2023
    800G Innovation Reshapes Network Dynamic

    The initial wave of switches featuring 800G ports has entered the market, facilitating a comprehensive evaluation of opportunities and strategic deployment planning by service providers. Accompanying this release are 800G client-side optical modules, designed to support either two 400GE connections or eight 100GE connections. Furthermore, as industry standards continue to evolve, an 800GE version is poised for introduction. The advancement from 400G to 800G technology streamlines this transition, simplifying the adoption of dual-density optical modules and branching connection methodologies.

    The emergence of 800G switch ports, optical modules, and Direct Attach Cables (DACs) presents an enticing prospect for service providers aiming to enhance network performance without delay, bypassing the need to await the formalization of the 800GE standard. In parallel, the ongoing efforts of the IEEE 802.3df 800G and 1.6T working groups, coupled with notable industry progressions, ensure the formulation of a robust roadmap for forthcoming upgrades. The swift deployment of 400G technology signifies the industry’s agility in delivering cutting-edge solutions, thus aligning with the imminent shift towards 800G as service providers expand and modernize their data center infrastructure.

    The release of 25.6T switches with 800G ports creates new opportunities for data center operators, leveraging networks that significantly enhance performance while reducing complexity, costs, and power consumption.

    As depicted in Figure 1, a single 25.6T switch can replace six 12.8T switches. Each 800G port on the 25.6T switch comprises 8x 100G PAM4 channels and will support 8x 100GE connections, two 400GE connections, or a single 800GE interface. By utilizing 100G PAM4 serial interfaces, the number of signal channels on the 25.6T switch matches each one of the six 12.8T switches, reducing the overall complexity of the solution. Replacing six switches with one at the same 25.6T network capacity significantly decreases power consumption and costs.

    Figure 1, Decreasing Complexity, Cost and Power Consumption with 25.6T Switches (Resource from Earlswood Marketing)

    Figure 2 depicts a 25.6T platform equipped with 800G ports, featuring 32 ports in a compact 1RU form factor, supporting 64 400GE links or 256 100GE links using breakout cables. Several use cases exploring this flexibility are discussed below.

    Figure 2: 25.6T Platform with 32 800G Ports

    Figure 3 illustrates an 800G port supporting two IEEE 400G BASE-FR4 connections extending up to 2km, offering high-density 400G interfaces. Compared to 12.8T systems, the configuration using breakout cables enables a doubling of the 400GE link count supported by the 1RU 25.6T switch system. Deployments with this versatile configuration are progressing in various applications, including high-density AI/ML clusters and ultra-high-definition video processing in data centers.

    Figure 3, 2x400G QSFP-DD to 400GBASE-FR4 Breakout

    Figure 4 illustrates an 800G port supporting 8 IEEE 100GBASE-DR (up to 500m)/IEEE 100GBASE-FR (up to 2km)/IEEE 100GBASE-LR (up to 10km) connections, offering high-density 100G interfaces. The figure displays branching into 8 independent 100G links, branching into two 400G connections, and an 800G direct connection. These configurations are highly suitable for upgrades in peer/hosted networks and distributed data centers requiring numerous 100GE network connections.

    Figure 4, 8x100G to 100GBASE-DR/FR Breakout

    800G technology undoubtedly stands as a significant driving force in the field of communications, leading us towards an unprecedented future of high-speed, high-efficiency, and highly interconnected networks in the upcoming digital era. Its outstanding transmission speed and vast bandwidth enable various industries to embrace new application scenarios and innovative opportunities.

    Reference: “800G Client Optics in the Data Center — A Heavy Reading white paper produced for Cisco,” SIMON STANLEY, ANALYST AT LARGE, HEAVY READING