Bringing the World the Latest in Optical Communications News
ISSUE 42 | Q3 2025
Future-Proofing Deployments | p26 RETHINKING LONG-HAUL FIBRE NETWORKS :
THE FUTURE OF COHERENT OPTICS IN EUROPE Helen Xenos | p18
THREE DECADES OF OPTICAL COMMUNICATIONS EVOLUTION John Edwards | p32
THIN TRANSPONDERS Fady Masoud | p8
REALTIME PHOTONIC SENSORS HIGH-SPEED FIBRE FOR HAWAII DFA & CIENA BREAK RECORDS
Media Partners to
CONTENTS
4 Industry News 8 800G Coherent Pluggables Fady Masoud 10 Flexible ROADM technology Kazuichi Ichikawa 14 The AI-led overhaul George Ashwin 18 Future of coherent optics Helen Xenos 20 The role of V-Grooves Focuslight 26 Long-haul connectivity Steve Roberts 29 Ultra long haul networks Eugene Park 32 Optical Evolution John Edwards
Welcome to the Autumn edition of Optical Connections, my first as editor. I want to begin by recognising the brilliant job Peter Dykes has done for the last seven years. I will do my best to emulate the high standards he set during his editorship. It is a very exciting time for the sector with the 30th anniversary of the ECOC conference and exhibition upon us. But as we look back, we also want to look into the future as I cannot recall a time when there were so many new technologies and innovations driving this industry forward and responding to the challenges it faces. My personal link with optical communications dates back to September 8, 1980, when I was a young press officer for British Telecom, helping to launch its first fibre optic cable between Walsall and Brownhills – coining the now- familiar phrase that the optical fibres were “thinner than a human hair.” I also remember visiting the Plessey research centre at Caswell, where the scientists were preparing what they claimed was the first ever optical switch to be demonstrated at Palexpo. My next major chapter leading to the role as Editor of Optical Connections was my involvement in ECOC, which stretches right back to 2002. The pace of the sector’s technological advances has been nothing short of remarkable to witness. And now, as the industry steps into the AI driven era with fresh challenges, it is rising with innovation to meet the surging demands for bandwidth. This Autumn edition of Optical Connections dives into the technologies redefining that landscape. From the evolving role of coherent optics and IP-over-DWDM, to scalable ROADM architectures and engineered PIC-fibre interfaces, the issue highlights the technologies supporting next generation optical networks to overcome the daunting challenges they face. We also explore alternative solutions for AI-ready data centres and the future of fibre for long-haul connectivity. It has been a real pleasure putting together my first issue of Optical Connections. I hope you enjoy reading it and I look forward to catching up with many of you in Copenhagen! OPTICAL CONNECTIONS – WHERE WOULD WE BE WITHOUT THEM TODAY?
34 ECOC preview 40 Product News
Brian Dolby Editor, Optical Connections
READ ONLINE/SUBSCRIBE: www.opticalconnectionsnews.com FOLLOW US @opconsnews EDITORIAL : editor@opticalconnectionsnews.com ADVERTISING: sales@opticalconnectionsnews.com DESIGN: Antonio Manuel
MEDIA PARTNERS
Optical Connections is published by NEXUS MEDIA EVENTS LTD Suite 3, Building 30, Churchill Square, Kings Hill, West Malling, Kent ME19 4YU t: +44 (0) 1732 752 125
www.opticalconnectionsnews.com
3
ISSUE 42 | Q3 2025
INDUSTRY NEWS
New European photonics project aims to prevent bridge collapses
A new €5.1 million European research project is exploring whether fibre- optic cables can serve as real-time sensors for hidden damage in critical infrastructure, including bridges, railways, tunnels and energy pipelines. The ECSTATIC project, coordinated by Aston University in the UK, is trialling this approach in a major UK city, using a heavily used Victorian-era railway viaduct as its first live test site. The goal is to detect subtle structural shifts, stress,
and vibrations in real time, using laser light pulses sent through fibre-optic cables already embedded right beneath our feet. Installing physical sensors across entire transport and energy networks would cost billions and cause major disruption according to the ECSTATIC project, which plans use existing fibre infrastructure.
As trains pass overhead, the fibres subtly flex and vibrate, changing how the light behaves inside the cable, altering the phase and polarisation of the light, creating a kind of optical “fingerprint” of the forces acting on the structure. By measuring these changes and interpreting them using a new dual- microcomb photonic chip and advanced AI signal processing, ECSTATIC aims to pinpoint early warning signs of damage or fatigue. This process is done
without any interruption to existing internet traffic or deployment of new fibre cables. The ECSTATIC project runs until July 2028 and is bringing together 13 partners from across Europe. This includes leading universities Padova, L’Aquila, Chalmers, Alcalá, and West Attica, alongside industry Telecom Italia Sparkle, OTE Group, Nokia, Network Rail, MODUS, and Swiss SME Enlightra SARL, as well as the Greek seismology specialists NOA.
At the project’s first demonstration site,
researchers will send ultra- precise laser pulses through buried fibre-optic cables.
Ocean Networks, Prysmian, IT, to boost high-speed fibre access for Hawaii
Ocean Networks has selected cable solutions provider Prysmian and submarine cable engineering and installation specialist International Telecom Inc. (IT) for the Hawaiian Islands Fiber Link (HIFL) project. The partnership is set to boost development
of Hawaii’s open-access, carrier-neutral inter-island fibre infrastructure, which is designed to improve and expand high-speed broadband internet throughout the state. The HIFL project, a key component of the State of Hawaii’s “Connect Kākou”
broadband initiative, aims to deliver robust and resilient digital connectivity across the Hawaiian Islands. Under the agreement, Prysmian will supply approximately 740 kilometres of state-of-the-art submarine cable, while IT will provide essential engineering and installation services for
the HIFL system.
Ocean Networks is responsible for the overall supply, construction, operations, and maintenance of the HIFL system, reinforcing its commitment to creating an advanced and equitable digital landscape for the state of Hawaii.
LINX announces major network upgrade at IXP in Riyadh
Nokia and MXFiber extend high-speed connectivity across Southeast Mexico
The London Internet Exchange (LINX) has announced a successful network upgrade at the center3 Internet Exchange Point (IXP) in Riyadh, Saudi Arabia. The upgrades will increase the availability of 100GE ports following customer demand and also to enable the interconnection point with 400GE connectivity services. The exchange is now set to meet the growing demand for high-capacity, low-latency interconnection services from global networks, content providers, and enterprises. The hub has become a
Nokia has deployed a new optical transport network backbone for MX Fiber to extend reliable, high- capacity connectivity across Southeast Mexico, one of the country’s most undeserved regions. The deployment is set to boost economic revitalisation and modern services for communities, businesses, and government projects in the area. Customers across Chiapas, Tabasco, and Quintana Roo, and soon Campeche and Veracruz, will benefit from faster internet, enhanced cloud access, and support for data-intensive applications. Spanning 1,800 km, the
cornerstone of Saudi Arabia’s digital infrastructure since its launch in 2024, with its latest enhancements supporting the region’s ambitious Vision 2030 agenda. The programme places digital transformation – from smart cities and AI-driven healthcare to fintech and cloud services – at the heart of national development. With over $3B already invested and an additional $10B planned, center3 is building a next-generation, carrier-neutral data centre
new network is built on Nokia’s Flex-Grid DWDM technology and 1830
Photonic Service Switch (PSS), delivering scalable 10G, 100G, and 200G services. This capacity enables customers to connect to modern data centres, industrial parks, and subsea transport hubs, critical for commerce, mobility, and digital inclusion. The Nokia 1830 PSS solution enables seamless upgrades to 400G and 800G without disrupting existing services and has a built-in dynamic network management and real-time performance monitoring via Optical Time-Domain Reflectometry (OTDR).
ecosystem that delivers unmatched scalability,
sustainability, and reliability for enterprises, hyperscalers, and government operations.
4
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
INDUSTRY NEWS
DFA and Ciena break fibre capacity records in world-first trial
Emtelle and GoFibre strengthens partnership to accelerate high-speed connectivity across Scotland
Dark Fibre Africa (DFA) and Ciena have set a new benchmark in fibre capacity with a world-first trial that delivered 1.6 Tbps over a single wavelength – more than four times the performance of previous tests. Powered by Ciena’s WaveRouter and WaveLogic 6 Extreme solutions, this breakthrough marks a major step toward future-proofing DFA’s network and driving economic growth in South Africa through advanced IP/ optical convergence. The trial was conducted over a 40 km stretch between Isando and Midrand, using DFA’s high-capacity core network. The route chosen replicates the previous trial that had demonstrated 400 Gbps capability.
Emtelle, a global supplier of passive network infrastructure for high- speed connectivity, has expanded its strategic partnership with ISP and network provider GoFibre following GoFibre’s latest Project Gigabit contract award. As key employers in the Scottish Borders, Emtelle and GoFibre will work together to deliver duct, sub-duct, and fibre infrastructure for new full-fibre networks, supporting Fibre to the Premises (FTTP) speeds of up to 10Gbps. This infrastructure will be provided to Edinburgh-based GoFibre as part of its fourth contract under the UK Government’s £5bn Project Gigabit, valued at £105m. It will see GoFibre deliver
Powered by Ciena’s WaveLogic capabilities and deployed by Ciena partner, Willcom, WaveRouter will allow DFA to deliver greater scalability, flexibility, and sustainability by integrating 400 Gbps, 800 Gbps, and 1.6 Tbps services within a single router. By combining Ciena’s optical and routing technologies in a unified, multi- layer architecture, DFA can fully harness the advantages of coherent routing to build a simpler, more robust network. Additionally, Ciena’s Navigator Network Control Suite (Navigator NCS) provides DFA with one point of control to automate multi-layer network operations and scale wide area network (WAN) traffic, delivering capacity precisely when and where it’s needed.
high-speed broadband connections and download speeds to around 63,000 residents and businesses to parts of Aberdeen City, Aberdeenshire, Angus, Dundee, the Highlands, Moray and Perth and Kinross. The first connections are set to be delivered by summer 2026. Emtelle has partnered with GoFibre on its commercial network build, and to deliver previous contracts under Project Gigabit, which prioritise areas that are not commercially viable for broadband providers, including rural communities. Together GoFibre’s network now covers more than 120,000 premises across over 30 hard-to-reach areas in both Scotland and the North of England.
www.opticalconnectionsnews.com
5
ISSUE 42 | Q3 2025
INDUSTRY NEWS
ESA announces Europe’s first deep-space optical communication link
Nokia establishes subsea cable system across Europe and North Africa
The European Space Agency (ESA) has marked a historic milestone by establishing its first optical communication link with a spacecraft in deep space. The link was made with NASA’s Deep Space Optical Communications (DSOC) experiment aboard its Psyche mission, currently at a distance of 1.8 astronomical units, around 265 million km. It is the first of four planned links occurring this summer. The achievement marks yet another milestone in the long history of cross-support between space agencies, said ESA, demonstrating the potential for interoperability between the organisation
Nokia has announced that it will power a new subsea fibre-optic network will connect the Atlantic coast, Mediterranean Sea and the Red Sea, creating a new high- capacity digital corridor in the region, driving connectivity, innovation, and economic growth. The Medusa Submarine Cable System, a project owned by AFR-IX Telecom, is a significant step toward closing the digital divide between Europe and North Africa, connecting countries such as Morocco, Tunisia, Libya, Algeria, and Egypt with high-capacity fibre-optic links.
and NASA in the realm of optical communications, something previously only achieved with radiofrequency systems. “The first successful demonstration of deep- space optical communication with a European ground segment marks truly a leap step towards bringing terrestrial internet like high- speed connectivity to our deep-space spacecraft. This joint achievement together
Designed as an open- access system, Medusa provides telecom providers across the region access to advanced connectivity services, supporting the rollout of 5G, the growth of cloud infrastructure, and the increasing bandwidth demands of AI and future technologies. Leveraging Nokia’s 1830 GX Series platform and advanced ICE7 coherent optics, capable of transmitting tens of terabits per second per fibre pair, the Medusa Submarine Cable System is equipped to deliver high-capacity, low-latency connectivity with optimal cost and power efficiency per transmitted bit.
with our colleagues and partners in industry and
academia, ESA’s Directorate of Technology and NASA/JPL underlines the importance of international cooperation”,
says Rolf Densing, ESA’s Director of Operations.
Spirent and Telescent partner to deliver enhanced test lab automation
EXA Infrastructure announces 1,200km fibre route from London to Frankfurt, Amsterdam and Brussels
A new partnership between Spirent Communications and Telescent is set to provide a boost for companies seeking to optimise AI/ML resource connectivity and automate test infrastructure. AI and machine learning workloads demand optimal connectivity through optical interconnections to maximise resource utilisation and investment returns. The collaboration addresses these challenges by bringing together Telescent’s innovative optical circuit switch (OCS) and automated fibre patch-panels solutions with Spirent’s Velocity test lab automation portfolio. Telescent’s robotic fibre optic cross-connect systems enable automated and remote provisioning of fibre connections, dramatically reducing the need for manual patch panel operations and minimising human error. “Our partnership with
Telescent represents another great example of how Spirent is bringing innovation to automated test solutions to help companies ensure reliability, security, and performance in their operational networks,” said Anil Kollipara, VP of Product Management at Spirent. “Furthermore, by integrating Telescent’s advanced optical switching technology with the proven test automation capabilities of our Velocity portfolio, we’re delivering a solution that will transform how organizations approach lab automation and AI/ML resource optimization. When integrated with the Velocity Test Automation
EXA Infrastructure has announced the deployment of a new fibre route connecting London to Frankfurt, Amsterdam and Brussels. The project aims to enhance Europe’s digital infrastructure with two scalable cable landing stations, modern high-fibre- count cables, and upgrades to existing In-Line Amplifier (ILA) facilities across the UK, Belgium, and the Netherlands. Integral to the project is a new consortium submarine cable and EXA Infrastructure is the sole telecom consortium member. It is responsible for providing landing party and backhaul services. EXA Infrastructure claims to be the largest dedicated digital infrastructure platform throughout Europe. The new 1,200 km route includes 1,085 km of new low-
loss G.652D terrestrial fibre for end-to-end connectivity and a 115 km subsea build from Margate, UK to Ostend, Belgium, utilizing ultra-low- loss G.654C cable. The two new landing stations – EXA’s 21st and 22nd globally – further strengthen its extensive network spanning the U.S. East Coast, Western Europe, and the Mediterranean. Steve Roberts, SVP Strategic Network Investments and Product Management at EXA Infrastructure, said: “This is a real milestone for robust connectivity options in Europe and includes the first new subsea cable on this complex corridor – the North Sea – in 25 years. This new route complements our investment in the Channel Tunnel, delivering scalable, modern and optimized fibre paths between key FLAP hubs.”
Portfolio, the solution enhances operational efficiency through
automation for enterprise network test labs, AI/ML laboratories, data centres, and equipment manufacturer validation.
6
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
www.opticalconnectionsnews.com
7
ISSUE 42 | Q3 2025
FADY MASOUD 800G COHERENT PLUGGABLES
OF 800G COHERENT PLUGGABLES THINK THIN: UNLOCKING THE POWER
MARKET DRIVERS Major improvements in increasing fiber capacity have come from the evolution of optical engines – from direct-detect to coherent – and from increased wavelength bit rate enabled by the higher order modulations schemes. However, optical engines are approaching the Shannon limit. As a result, fiber capacity has begun to flatten as spectral efficiency is becoming constrained. This has triggered service providers and network operators to begin looking at new strategies for increasing the capacity of their networks including the use of multi-fiber strategies to connect locations. As network operators shift to these multi-fiber strategies it can cause a significant shift in their priorities when selecting network solutions. When networks are designed using a single fiber pair between location, capacity per fiber is typically the most important factor in selecting an optical engine. However, when multi-fiber strategies are used, capacity per fiber becomes significantly less important enabling network operators to focus on other factors – such as power and space efficiency. This shift in priorities is aligning perfectly with the evolution of coherent pluggables. Advances in DSP and CMOS technology – from 28nm to 7nm to 3nm – have led to significant enhancements in optical performance, increasing capacity-reach in compact form factors such as QSFP-DD and OSFP. And while these coherent pluggable optical engines provide less fiber capacity compared to their embedded optical engine counterparts, they provide a significant reduction in power consumption and footprint per bit. While every application will have its own set of drivers, with embedded engines continuing to be the technology
The second model is the deployment of coherent pluggables directly into routers , commonly referred to as IP over DWDM (IPoDWDM). This model has the advantages of low CAPEX, low power consumption, and reduced footprint by eliminating the need for an optical transport platform. Moreover, coherent pluggables benefit from a complete eco- system of Multi-source Agreements (MSA) and interoperability forums for a seamless line interworking and service provisioning. However, IPoDWDM presents operational challenges related to management and compatibility across different host devices and operating systems. Furthermore, this model does not come at par with fully-fledged transponder-based platforms due to the lack of complete traffic aggregation capabilities and operational functionality. Additionally, the IPoDWDM model has a direct one-to-one mapping between the router port speed and the pluggable speed – for example, where 400G pluggables are deployed in 400G router ports and 800G pluggables in 800G router ports. Nonetheless, sometimes the coherent pluggables must be dialed down to operate at a lower bit rate, such as 600 Gb/s or 400 Gb/s to close a specific link which results in a wasted router port capacity. Moreover, pluggables cannot fully replicate the optical functionality of embedded transponders.
of choice when fibers are scarce and adding incremental fibers is high, the compelling economics of coherent pluggables coupled with the shift in priorities away from maximizing fiber capacity and toward cost, space, and power efficiency is significantly expanding the applications scope. CURRENT DEPLOYMENT MODELS There have historically been two main deployment models of coherent optical engines: 1. High-performance optical engines embedded in a transponder card in an optical transport platform 2. Coherent pluggables hosted in IP/ Ethernet platforms such switches and routers. Embedded optical engine based transponders are designed and built from the ground-up to maximize fiber capacity with best-in-class optical performance (capacity-reach). They aggregate traffic from various types of client interfaces onto one or two high bit-rate wavelengths such as 1.2 Tb/s. Designed for deployment in fully-fledged optical transport platforms, they typically support multiple client ports for aggregation in addition to a full suite of integrated optical functions designed for traffic grooming, aggregation, hairpinning, protection, alarm correlation, and so on. In addition to the performance and host of traffic management features, another advantage of embedded transponders is the operational domain separation , acting as a demarcation point between the IP and the optical domain. This is useful in lawful interceptions or enforcing SLAs through setting clear boundaries and business demarcation. However, transponders come with trade-offs such as high power consumption and large footprint.
EXPANDING DEPLOYMENT MODULES WITH THIN TRANSPONDERS
Thin Transponders introduce a third deployment model , that combines some of the strengths from the other two models without their associated disadvantages. Thin transponders are a set of small modules or “sleds” that are optimized for coherent pluggables and
Figure 1: Expanding deployment models with Thin Transponders
8
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
FADY MASOUD 800G COHERENT PLUGGABLES
Figure 2: Thin Transponder Applications
normally equipped in compact modular platform or an optical platform, where two or four of them can be equipped in a single rack unit. Thin transponders offer multiple client ports (100G, 200G, 400G or 800G) for grey optics to carry traffic from other platforms such as routers, and multiple line ports for high capacity coherent pluggables such as 400G ZR, 400G ZR+, 800G ZR or ZR+. Similar to IPoDWDM solutions, thin transponders offer lower CAPEX, lower power consumption and smaller footprint but without the operational challenges mentioned earlier. They also combine some of the advantages of embedded transponders such as multiple client- side aggregation, operational domain separation and some of the optical capabilities of fully fledged embedded transponders. Figure 1 summarizes the pros and cons of each deployment model. Thin Transponders also enable a technology lifecycle separation between the long-lasting refresh cycle of the photonic layer and the shorter cycle of the IP layer. This allows network CASE STUDY To quantify the benefits of thin transponders, a network analysis was performed on a full-filled (C-Band) 1,000 km long optical link where all three deployment options were compared side-by-side. Compared to embedded
operators to benefit from the latest generation of coherent pluggables, such as ICE-X 800G ZR/ZR+, in existing 400G routers. This leads to a maximized ROI and operational flexibility by avoiding the network-wide upgrade of all routers to the latest 800G-capable generation.
Furthermore, some thin transponders offer traffic aggregating functions to maximize throughput and avoid bandwidth waste. By leveraging a feature called “virtual bandwidth”, two 800GE client interfaces can be carried over four 400 Gb/s coherent wavelengths, or three 800GE client interfaces over four 600 Gb/s wavelengths. Figure 2 depicts two applications where thin transponders are used in an optical transport application and as a regeneration site for IPoDWDM applications.
USE CASES As noted, thin transponders
combine key attributes of embedded transponders and IPoDWDM, providing cost-effective, flexible, and highly reliable optical transport leveraging the latest generation of coherent pluggables. Thin transponders are the solution of choice when space and power are limited, and a low variety of client services is required. The thin transponder solution delivers a significant reduction in CAPEX and OPEX thanks to their compact footprint and low power consumption of coherent pluggables. Thin transponders can also enhance IP traffic reliability by leveraging the optical control plane and protection and restoration schemes of optical networks. transponders, thin transponders offer compelling reduction in CAPEX ($/G), power consumption (W/G) and footprint (RU). However, they offer slightly less (20%) less capacity per fiber as depicted in Figure 3. *Reflects different volumes and scale
CONCLUSION Thin transponders combine key
advantages and benefits of embedded transponders and IPoDWDM to provide deployment flexibility, while offering compelling economics through reduced CAPEX, power consumption, and footprint. 1: Fiber Broadband Association
Fady Masoud is a Senior Director for Solutions Marketing at Nokia focusing on next-generation Intelligent Coherent Pluggable optics (ICE-X) and cloud / data center interconnect solutions. His area of expertise is the architecture and requirements of next-generation optical networking infrastructure.
Figure 3: Quantifying the benefits of Thin Transponders
www.opticalconnectionsnews.com
9
ISSUE 42 | Q3 2025
KAZUICHI ICHIKAWA FLEXIBLE ROADM TECHNOLOGY
LOWERING COSTS WITH AN OPEN SPECIFICATION FLEXIBLE ROADM TECHNOLOGY:
EVOLUTION OF OPTICAL NETWORKS The rapid increase in data traffic is driving the global construction of additional data centres, as well as the expansion of existing networks. As a result, the demand for high-performance, scalable, and flexible network solutions has risen, leading to growth in the reconfigurable optical add/drop multiplexer (ROADM) market. ROADMs can switch the destinations of optical signals with different optical wavelengths, with many ROADMs being deployed in optical communication networks, where they serve as key devices. ROADM TECHNOLOGIES AND ADVANTAGES ROADM simplifies network design, supports bandwidth change requests, and improves operational efficiency through remote reconfiguration. A typical ROADM consists of multiple Wavelength Selective Switches (WSS), which in turn consist of a combination of MUX/DEMUX, optical switches, variable optical attenuators (VOAs), and control modules. The functionality of ROADMs, which switch the destination of optical signals without electrical conversion, provides operators with the capability to streamline route design.
configurations, ensuring efficient communication conditions. This approach minimises the risk of network disruptions caused by human error and significantly reduces the operational workload incurred by network management. ROADMs enable network operators to add, drop or pass-through the required wavelengths at any site to ensure optimal network efficiency, as traffic routing and services can be altered with no physical changes. They are easily scaled, enabling the quick addition of new wavelengths, services or network nodes as bandwidth demand increases. Multi-degree ROADMs, which will allow connections between multiple fibre paths, can also be used to reroute traffic in the case of fibre cuts or hardware failures, providing enhanced network resilience and redundancy. The ROADM has become a key technology in developing high-capacity, flexible, agile optical networks such as 400G Wavelength Division Multiplexing (WDM) for demanding cloud, 5G and data centre use cases. LOW LATENCY EDGE COMPUTING There are various architectures that network designers can deploy. One example is CDC (Colourless, Directionless and Contentionless), which enables non- blocking mesh configurations. Colourless adds or drops any wavelength or colour at any port, directionless adds or drops wavelengths in any direction, while contentionless adds or drops wavelengths without interference. Multi-degree and CDC ROADMs are essential to enabling detours and fast switching that ensure low latency. This type of remote dynamic rerouting enables edge services such as AI workloads, real-time analytics, and 5G use cases to maintain optimal direct low- latency routes even under high congestion or when a link fails.
ROADMs not only provide automation and maximise spectral efficiency, but can also keep data in the optical domain for longer by passing wavelengths through optically without optical-electrical-optical (OEO) regeneration. Such ROADMs cut equipment costs and lower latency across edge nodes and data centre interconnections. PROMOTING OPEN NETWORKING Since the 2010s, the adoption of ROADM technology has gained significant momentum, leading to the emergence of new technical challenges. Historically, the design and production of ROADM equipment were undertaken only by a small number of specialised hardware vendors. This lack of diversity resulted in minimal interoperability across devices from different manufacturers. Consequently, ROADM-related systems were constrained in their ability to be flexibly integrated based on specific functional requirements or cost considerations. The increasing demand for optical networks to be flexible, scalable, and autonomously operable has necessitated significant efforts to ’open up optical networks.’ These efforts encompass the establishment of standardised technical specifications and the enhancement of interoperability among devices from multiple vendors. Such initiatives are being actively pursued under the ‘OpenROADM’ project, which was initiated in 2015, with AT&T and other industry leaders as founding members. The project currently has over 30 member organisations. OpenROADM focuses on standardising optical interfaces and their specifications to ensure seamless interoperability among ROADM systems supplied by diverse manufacturers. This initiative further emphasises the development of integrated
Figure 1: Typical WSS configuration with four ports.
ROADMs enable the dynamic adjustment of wavelength bandwidth based on data transmission requirements. They facilitate the remote management of optical signal routing for individual wavelengths, effectively eliminating the need for manual on-site interventions by network technicians. In networks utilising ROADM technology, software- driven automation optimises connection
Figure 2: Optical network systems become open.
10
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
KAZUICHI ICHIKAWA FLEXIBLE ROADM TECHNOLOGY
management functionalities facilitated through multi-vendor software-defined networking (SDN) controllers. The current challenges facing ROADM systems include limited interoperability between equipment from different suppliers. Variations in network configurations across vendors frequently result in increased costs and operational complexity during the integration of new components into existing systems. Furthermore, reliance on a single vendor for network equipment constrains competition, reduces innovation, and lowers system flexibility. This vendor dependency, commonly referred to as ’vendor lock-in,’ restricts overall interoperability within the network ecosystem. OpenROADM aims to address these limitations by providing a standardised framework that enhances vendor-neutral interoperability. OPENROADM SPECIFICATIONS OpenROADM addresses five types of hardware with optical interfaces, including pluggable optics, transponders, in-line optical amplifiers, transponders/switches, and the ROADMs themselves. Combined with software-based controllers, these devices can be managed through SDN controllers that utilise a common data model and Application Interface (API). Open APIs empower operators to develop custom network applications, enabling features like low latency and high reliability. By adopting the YANG language for data modelling and control methods, OpenROADM ensures compatibility and seamless integration across different vendors. Furthermore, contentionless and directionless switching capabilities in ROADM architectures, when orchestrated by SDN, effectively resolve wavelength conflicts and automatically reallocate routes throughout the network. This enhances redundancy and supports uninterrupted data continuity, even in the event of multiple failures. Integrating Optical Performance Monitoring (OPM) with SDN control allows the controller to optimise resource allocation and avoid deploying infeasible lightpaths. In addition, SDN provides a unified control layer, orchestrating protection and restoration schemes across both optical and higher network layers, thereby maximising network resilience and operational efficiency. EMERGENCE OF THE OPEN ZR+ STANDARD While huge data centres are distributed across multiple locations, the communication distance between them (data centre interconnects, DCI) is kept relatively short (about 80 to 120 km) to minimise latency. In addition, point-to- point connections are used between data centres. The Optical Internetworking Forum (OIF) has developed 400ZR as an interface specification specifically for this
DCI application. 400ZR also defines a transmission method with WDM, enabling the realisation of Internet Protocol (IP over DWDM) using high-density multiplex modulation. On the other hand, the optical interface of OpenROADM uses a communication protocol called Optical Transport Network (OTN), which has been standardised by ITU-T and can support data rates from 100 to 400 Gbps. However, the cost is also high.
Figure 3: Demo system and the role of Anritsu products at OFC 2024.
Although the two standards have different starting points, both use a digital coherent scheme suitable for WDM. For this reason, the hardware specifications of the corresponding optical transceivers have many parts in common, and a movement to integrate the standards has emerged. This led to the creation of the “OpenZR+” standard. This standard inherits the basic specifications of 400ZR while enabling DCI networking and longer-distance (around 480 km) communications. In December 2023, 600 Gbit/s and 800 Gbit/s specifications were added to the OpenROADM optical interface, while the OIF published the 800ZR standard in October 2024. Shortly, 800 Gbit/s specifications are expected to be added to OpenZR+. NEED FOR TESTING AGAINST ROADM To ensure interoperability between devices from different vendors, maintaining communication quality across individual devices and the network as a whole is essential. Rigorous testing is required to guarantee communication stability, prevent errors, and enhance reliability. Furthermore, real-time configuration adjustments and rapid recovery mechanisms are critical for maintaining network flexibility in the event of failures. Keeping pace with rapid technological advancements necessitates the adoption of testing solutions that can seamlessly adapt to emerging technologies and specifications.
were connected through the Add/Drop line of an Open ROADM system. This setup enabled the orchestration system to configure channel settings while monitoring critical network performance metrics, such as bit error rate, throughput, and latency, via the MT1040A. The unified system facilitated by the real-time evaluation of ROADM path changes based on performance quality data provides a framework for monitoring and adaptation.
ACCELERATING ROADM INNOVATION
The pursuit of flexibility, automation, and cost-efficiency for future ROADM technology will continue to drive the evolution of ROADM technology. Photonic integrated circuits (PICs), co- packaged optics, and advanced materials are instrumental in miniaturising and optimising ROADMs for future network demands. A notable recent trend is the push toward higher transmission speeds. In January 2024, OIF launched a project focused on the 1600ZR+ specification in response to market demand for enhanced ZR+ mode performance. Naturally, this development is expected to impact both OpenZR+ standards and OpenROADM specifications. Amid these advancements, it is important that the vendor ecosystem continues to actively participate in OpenROADM and offers products that contribute to the control and quality monitoring of optical communication networks.
ANRITSU AND OPENROADM Anritsu, in collaboration with
the University of Texas at Dallas, demonstrated advancements in OpenROADM/IP over DWDM (IPoDWDM) orchestration systems at OFC 2024 and SC24. Using the YANG model—a vendor-independent network control methodology developed by OpenROADM and IETF—their demonstration showcased networks integrated with an orchestration system managed by the University of Texas at Dallas. Two 400G ports on Anritsu’s Network Master Pro (400G Tester) MT1040A, equipped with embedded 400G OpenZR+ transceivers,
Kazuichi Ichikawa, Assistant Manager, Anritsu Corporation
www.opticalconnectionsnews.com
11
ISSUE 42 | Q3 2025
SPONSORED FEATURE INTEL
I/O CHIPLETS FOR AI INTERCONNECT HIGH-BANDWIDTH CO-PACKAGED OPTICAL
W ith the rapidly increasing deployments of larger and larger AI clusters built with more and more GPUs in the high-bandwidth scale-up domain, interconnect based on copper cables is becoming a bottleneck due to their limited bandwidth and reach. For scale-out networks, pluggable optics has served us well and offers a scalable path with future linear optics either in pluggable form or co-packaged. However, for the scale-up fabric, we are looking for lower power and lower cost- per-bandwidth than can be supported with current optical interface solutions. In this article, we share some of Intel’s work around high-density power- efficient silicon photonics-based optical I/O chiplets that are purpose-built for AI infrastructure. The I/O chiplet is intended for co- packaging with networking SOCs or compute resources to enable the reach and capability of optical interconnects, at the power consumption of copper. The chiplet is a fully integrated optical subsystem on a single silicon photonics chip built in Intel’s fab and taking advantage of the unique capability of integrating the laser light source and optical gain on the silicon die. The PIC was fabricated on the same silicon photonics line that is volume-proven with ~10 million Tx PICs deployed in pluggable modules in datacenter networks. The I/O chiplet has 8 Tx/ Rx fiber-pairs, each carrying data on
8 wavelengths spaced at 200GHz and generated from the on-chip DFB laser- array. Designed for operation at 64Gbps modulation speed, the 64-channel optical subsystem on a chip supports 4Tbps bandwidth per direction and >100m reach. The PIC is combined in a compact die-stack with a CMOS IC containing the analog front-end, control loops and other electronics, to form a complete transceiver array. The image above shows the complete die-stack. In the initial implementation, the chiplet is a linear E/O converter that attaches to a standard high-speed SERDES interface on the host IC, but in future implementations, currently under development, the host-side interface will be based on a high-density power efficient die-to-die interface such as UCIe, for the highest density and lowest power. With this approach where the link is supported with a purpose-built SERDES and DSP optimized for the optical channel, the end-to-end link can be realized with <5pJ/b power consumption. On the path to product and volume deployment, Intel demonstrated the technology capability with an optical I/O chiplet co-packaged with an Intel CPU. The industry-first co-packaging of an optical I/O chiplet with an CPU or GPU showcased CPU-to-CPU optical PCIe Gen5 interconnect using in-package optics and pointed to the future of scaling AI infrastructure based on co- packaged optical interconnect. Scaling in bandwidth and density comes through
cost and power efficient increase in wavelength count enabled by the on- chip laser technology, and from higher modulation bandwidth to increase the bandwidth per wavelength. Transmitter PICs operating at up to 200Gbps (PAM4) have been demonstrated for Ethernet pluggable chipsets. Besides the scalability and manufacturability that comes with the on-chip lasers manufactured at wafer-scale, the technology also offers excellent control and uniformity, and highly reliable lasers with demonstrated failure rate <0.1 FIT. In summary, the bandwidth scaling challenges presented by future AI deployments call for a new class of optical network interfaces that can support high bandwidth in a more power- and cost-efficient manner than current optical interfaces. Intel has demonstrated the first implementation of a co-packaged scalable high-bandwidth I/O chiplet based on a mature, volume-proven silicon photonics technology platform. The work highlighted the capability of the silicon photonics platform, using a 64-channel PCIe gen5 PIC as demonstration vehicle, with a clear bandwidth scalability path to support the requirements of future compute systems.
12
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
CIOE 2025 THE 26TH CHINA INTERNATIONAL OPTOELECTRONIC EXPO September 10-12, 2025 Shenzhen World Exhibition & Convention Center 240,000M² Exhibition Area 130,000+ Visitors 3,800+ Exhibitors
8 SUB-EXPOS
Colocated with
Semiconductor Expo
SIGN UP TO VISIT
CIOE 2025 AT A GLANCE
#CIOE2025
GEORGE ASHWIN THE AI-LED OVERHAUL
THE AI-LED OVERHAUL OF DATA CENTRE FACILITIES HOW ALTERNATIVE OPTICAL SOLUTIONS SUPPORT Across Europe, the current data centre landscape is enabling growth not only for the major players, but also for smaller enterprise entities. The capacity of facilities and IT installations continues to expand exponentially, primarily because of new legislation and sustained investments in both established businesses and Artificial Intelligence (AI) start-ups alike.
T his momentum is placing increasing pressure on operators, who are already trying to overcome a number of other pain points: for example, data centres remain significant energy consumers, and rising energy prices and limited grid capacities can hinder operators in the race to implement optimal network infrastructure. As economies become increasingly digitalised, and we continue to see a desire for greater data sovereignty across the continent, these challenges will only grow harder to deal with. CHANGES IN LEGISLATION A number of initiatives have been launched in recent years to establish Europe as a word-class hub where AI is concerned. In February 2025, the European Union (EU) launched the InvestAI project, which will mobilise approximately €200 billion into AI, including a specific €20 billion for AI gigafactory sites across the continent. Recognising that optimal infrastructure is essential for open, collaborative development of AI models – and to make Europe an AI powerhouse – the strategy’s end goal is to implement AI successfully across key industries such as healthcare, telecoms and more. Meanwhile, the European Commission’s AI Continent Action Plan, launched in April 2025, includes the proposal of a Cloud and AI Development Act to triple the EU’s data centre capacity by 2032. This
followed the announcement of seven new AI factories at the end of 2024, located in Spain, Italy, Finland, Germany, Luxembourg, Greece and Sweden. These developments will help house close to 100,000 AI chips – four times more than the current capacity of AI factories today – with the additional aim of bolstering the amount of supercomputers found within the region. There are also a number of country- specific AI initiatives too. Announced for the first time in January 2025, the UK Government has launched its own AI Opportunities Action Plan to enhance the lives of its citizens and public services through AI. It details plans to encourage greater investment from AI firms to create growth zones within the country, which will be used to deliver new infrastructure needed for tomorrow’s demands. One such zone is already under development in Culham, Oxfordshire, with more on the horizon. The plans are also set to enhance public compute capacity by 2000% by developing a brand- new supercomputer, as well as the development of an AI Energy Council to fully grasp the demands and challenges expected to be placed on resources as a result of these developments. In April 2025, the German Coalition Agreement was also drafted to maintain the country’s position as Europe’s leading data centre hub. Several cities in Germany – including Frankfurt – are considered key markets for data centre
developments and are seeing major investment and expansion in this area. This agreement aims to bolster these growing markets, while encouraging greater AI-specific investment across the country. THE CHALLENGES TO OVERCOME However, there remains several hurdles when it comes to building the necessary infrastructure for AI powered data centres. One key issue remains the power requirements and overheads of these facilities. AI requires significant computational power to work effectively, which will drive up energy demands across the board. At the same time, such powerful servers and components will need state-of-the-art cooling systems that can manage the heat generated within the data centre, which also contributes to greater consumption, and more demand on current energy sources. In Germany, natural gas accounts for near 33% of its final energy consumption, which is a vulnerability considering the country faces some of the highest electricity prices in Europe in what remains a volatile period in the sector. While energy sources are becoming increasingly diversified, this is still a point of concern for operators within the region, especially when factoring the sheer amount of power required for data centre infrastructure. The electricity consumption of German data centres alone is expected to rise to 80 TWh by 2045, an almost 400%
14
www.opticalconnectionsnews.com
| ISSUE 42 | Q3 2025
GEORGE ASHWIN THE AI-LED OVERHAUL
increase from the rates seen in 2024. This is not a problem limited to Germany. In fact, the International Energy Agency (IEA) believe that both AI-driven and non AI-driven data centres will likely use 80% more energy in 2026 than they were in 2022. While China and the United States will likely experience the most growth in energy consumption, Europe is not lagging too far behind. As such, initiatives aimed at meeting the increasing demand with renewable energy sources are well underway in the majority of EU Member States. There also remains some concerns over the safety of these data centres. As compute power grows, so too does the amount of sensitive data being stored and processed. If these data centres are breached by cyber criminals, it can lead to significant data loss and extravagant ransom demands.
themselves. This is especially true when it comes to AI-led data centres. Instead, they are working with system integrators who acquire and develop full turn-key solutions on their behalf. This means these integrators have carved out their own niche in an increasingly prosperous market, with a wide array of NEMs to choose from when it comes to developing these solutions. While this is a lucrative position to be in, they must also be careful to avoid agreements that tie their projects to specific NEMs; in fact, the most effective option in terms of cost and performance would be to utilise solutions from the whole ecosystem, and avoid unnecessary vendor lock-in. As a result, system integrators should be considering compatible optical transceivers if they are to unlock the full potential of their offerings. Some business - often smaller than their NEM counterparts - now offer optical components that deliver the same performance levels as typical market offerings, but with faster delivery times and greater support. The best of these alternative suppliers carry out 100% testing to guarantee their solutions will mirror a customer’s data flow before it is used in their data centre. This means operators can use transceivers tailored to their specific requirement, and implement it on a plug-and-play basis. Integrators evaluating the use of these solutions need not be concerned about potential warranty implications. The use of compatible components does not void service contracts, due to the Treaty of functioning of the European Union (TfEU), which protects businesses and their customers when alternative optical solutions are used. Any explicit or implicit warranty tie with a manufacturer would be considered an illegal action under Articles 101 and 102 of the TfEU. A similar legal standing is also ensured within the United Kingdom through the Competition Act of 1998, which forbids any agreements deemed ‘anti- competitive’ so that businesses within dominant market positions do not abuse their power. Both give operators and system integrators the opportunity to select the most applicable solution for their data centre infrastructure without fear of repercussion. NEXT-GENERATION OPTICAL SOLUTIONS This includes the 400G and 800G NEM alternative components developed to overcome the demands of hyperscale networks, and the new 1.6T components which deliver the ultra- fast speeds vital to managing intense AI workloads. The unique testing process offered by NEM alternative vendors can ease operators’ concerns over performance, since the optical products are tested in
customer environments to ensure both compliance and suitability. Their service also extends to supporting operators post-delivery, with many alternative optic suppliers offering customers round-the-clock access to support teams and engineers. By ensuring customer products are in working order and continue to meet requirements after the initial sale, the consistency and reliability of their transceivers can be maintained effectively. As data centres become increasingly localised, this service may prove invaluable to keep the infrastructure up and running in the wake of increased pressure. Using an alternative optic supplier can also empower operators to avoid unnecessary complications when building enhanced data centre infrastructure, and benefit from a compressed timeline from order to delivery. OVERCOMING PRESENT AND FUTURE CHALLENGES At this point, it is important to note than these new vendors do not see themselves as competitors to the established NEMs, and both certainly have a key role to play when it comes to supporting the AI-led overhauling of data centres. The key word is ‘compatible’: using a combination of NEM and alternative optical components is the best option for system integrators. As Europe accelerates towards becoming a global AI powerhouse, the scale and speed data centre infrastructure is already evolving is unprecedented. Rising energy demands, alongside regulatory complexity and the need for enhanced performance has placed enormous pressure on operators to build smarter, more efficient facilities. In this high-pressure environment, NEM alternative vendors are offering flexible, high-performance solutions that bypass vendor lock-ins and speed up deployment timelines.
OVERCOMING GROWING CONCERNS In expectation of these demands,
operators are increasingly migrating their workloads to the cloud at a rapid rate. The hope is that as more businesses make this transition, greater automation of basic tasks can take place, ensuring that existing resources can be maximised through data-driven insights. In turn, this should give operators greater capacity to handle the increasing workload. To reduce potential data loss and bottlenecks, operators are also transitioning from the large-scale facilities common today in favour of smaller, decentralised sites. By placing data centres closer to urban areas, essential resources can instead be hosted across multiple regions. Frankfurt is one such area, with many facilities situated along the Hanauer Landstrasse and Weismüllerstrasse due to the city’s close proximity to the Deutsche Commercial Internet Exchange (DE- CIX) – one of the world’s largest internet exchange points. Across Europe and beyond, the concerns related to handling data securely has also seen the rise of economic protectionism and data sovereignty. If data centres are decentralised and hosted in urban regions, then operators can prioritise the local storage and processing of data and better align with any relevant regulations linked to where these facilities are established. SETTING UP OPTIMAL FACILITIES When it comes to establishing these data centres however, we have also seen a new approach emerge. The rate these centres are being built means businesses no longer have the time to procure individual components from Network Equipment Manufacturers (NEMs) and assemble the infrastructure
George Ashwin, a UK Channel Director at AddOn Networks
www.opticalconnectionsnews.com
15
ISSUE 42 | Q3 2025
Page 1 Page 2 Page 3 Page 4 Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18 Page 19 Page 20 Page 21 Page 22 Page 23 Page 24 Page 25 Page 26 Page 27 Page 28 Page 29 Page 30 Page 31 Page 32 Page 33 Page 34 Page 35 Page 36 Page 37 Page 38 Page 39 Page 40 Page 41 Page 42 Page 43 Page 44Made with FlippingBook Ebook Creator