The Changing Economics of Coherent at the Network Edge [2/6]
Paper 1 established that access networks face a structural capacity problem and identified three technology shifts as the reason a new solution class is emerging. This paper examines those shifts — tracing why coherent technology historically remained in the core and the engineering trends that are changing its deployment economics at the network edge.
Engineering Constraints That Limited Access Deployment
For two decades, coherent optics were a core and long-haul technology. The engineering rationale was straightforward: coherent modulation delivers superior spectral efficiency and reach, but the DSPs and optical components required to implement it carried a power and size overhead that access environments could not accommodate. That constraint shaped how the industry built access networks, and it was a reasonable constraint to design around. What is worth examining now is how that calculus is changing — and how quickly.
Paper 1 identified three technology trends as the reason a new class of access-deployable coherent modules is emerging: silicon photonics integration, access-optimized DSP co-design, and the maturation of coherent standards below 400G. This paper examines each in turn, drawing on what the market has demonstrated rather than on projected capabilities.
|
sub-6W/strong> Power range now reached by multiple GA 100ZR QSFP28 modules (industry, 2024) |
10× capacity increase vs. 10G DWDM in same QSFP28 slot |
2x Cignal AI 100ZR forecast revision after QSFP28 viability confirmed |
The Power Barrier Is Now Being Approached — and Recent Designs Show How
The single most important reason coherent technology was excluded from access networks was not performance — coherent has always been the superior transmission technology. It was power. Access equipment lives in thermally constrained environments: outdoor street cabinets, pole-mounted enclosures, aggregation shelves in un-air-conditioned facilities. The QSFP28 form factor, which is the dominant 100G slot across the access installed base, has a thermal budget of less than 6 watts. A coherent module that exceeds that budget cannot be deployed without cabinet redesign — and cabinet redesign is, effectively, a forklift upgrade by another name.
The trajectory is worth tracing. In the line card era, a coherent transponder drew 40 watts or more. CFP2 pluggables brought that to around 20 watts. The 400ZR generation in QSFP-DD form factor reached approximately 15 watts — still three times the QSFP28 budget. Closing the remaining gap required two parallel advances: silicon photonics integration, which consolidates laser, modulator, detector, and amplifier on a single chip; and purpose-built DSP design, examined in the next argument.
Figure 1: Coherent module power consumption by generation. Recent 100ZR QSFP28 designs are reaching access-compatible power ranges below 6W. Sources: Coherent Corp., Cignal AI, industry estimates. Values approximate.
Recent commercially available designs demonstrate this trajectory in practice. Multiple vendors have reached the QSFP28 power budget with generally available 100G ZR modules, including industrial temperature variants rated for -40°C to +85°C operation. That environmental rating matters: it was the last remaining barrier to outdoor plant deployment, and it is now being addressed by a growing number of suppliers.
| “Digital coherent technology is moving closer to the edge to support ever-increasing demand for more bandwidth. Coherent QSFP28 offers the industry the opportunity to seamlessly upgrade 10G DWDM networks with 100G transceivers without the need to re-engineer the network and allows reuse of existing 100G router and switch ports”
— Vladimir Kozlov, CEO, LightCounting LLC, September 2024 |
DSP Co-Design Is the Architectural Breakthrough — Not DSP Speed Alone
The power reduction story is often told as a chip fabrication story: smaller process nodes mean lower power. That is partially true, but it misses the more important engineering decision. The reason access-class coherent DSPs have achieved their power targets is not primarily that they use newer CMOS nodes. It is that they were designed from the start around the constraints of the access environment rather than adapted downward from data center or long-haul DSP architectures.
The distinction matters architecturally. A data center DSP is built for maximum baud rate, modulation order, and reach flexibility — all of which consume power headroom that access environments cannot spare. Adapting one downward by reducing clock speed saves energy at the margins but preserves the underlying overhead. A DSP co-designed for access starts from different constraints: fixed baud rate for the target reach, FEC calibrated for access link budgets, chromatic dispersion tolerance matched to metro distances, and an industrial temperature range that data center components are not rated for.
Good examples of this co-design approach is the Steelerton DSP or Arycs Technologies in house DSP, purpose-built for 100ZR applications in collaboration with access equipment OEMs. The design decisions that determine whether a module reaches the sub-6W target are primarily architectural — and that architecture starts with the access operating envelope, not with a scaled-down version of a higher-performance platform.
| “10G DWDM is widely deployed at the edge of the network, but carriers are now searching for higher bandwidth 100G solutions. Low-power QSFP28 is required at the edge, and the 100ZR solution is well-positioned to be the format of choice for such upgrades.”
— Scott Wilkinson, Lead Analyst, Cignal AI, December 2024 |
The management interface dimension of co-design is equally important and less frequently discussed. Access equipment deployed over the past decade uses SFF-8636 interfaces — the legacy standard that predates CMIS. A module supporting only CMIS cannot function in that equipment without a management system upgrade, reintroducing the complexity the hardware advance was meant to eliminate. Modules designed for access deployment support both standards: deployable today on legacy equipment, with a clear path to CMIS telemetry as management systems are modernized.
Standards Maturation Is What Converts Technology Into a Market
The 400ZR precedent is instructive. Before the OIF Implementation Agreement was published in 2020, the 400G coherent market was fragmented: proprietary implementations, vendor lock-in, no cross-vendor interoperability. The IA changed that — creating conditions for a multi-vendor ecosystem that drove volume, price reduction, and a market far larger than any proprietary approach could have reached alone.
100ZR is following the same trajectory, with one notable difference: standardization and commercial hardware availability are arriving closer together than they did for 400ZR. The OIF has included 100ZR in multi-span interoperability demonstrations at both ECOC 2024 and OFC 2025, with 35 member companies at ECOC 2025 alone — spanning module vendors, system vendors, test equipment, and network operators. That breadth of participation signals an ecosystem forming, not merely a few modules in existence.
Figure 2: The 400ZR standardization template and the 100ZR trajectory. The two stories follow the same flywheel dynamic with a compressed timeline. Sources: OIF, Cignal AI, Coherent Corp., Arycs Technologies analysis.
The analyst data confirms the market read. When Cignal AI first modelled 100ZR demand in 2023, forecasts were sized around a limited set of high-priority deployments. After consulting with access hardware manufacturers and confirming that dedicated QSFP28-based modules could address their requirements, Cignal AI doubled its 100ZR shipment forecast for 2025 — driven by operator and equipment vendor signals, not new product announcements.
| “We learned that dedicated QSFP28-based 100ZR modules will fill a central role in the next generation of fiber access networks. As a result of our discussions with access hardware manufacturers, Cignal AI’s forecast for 100ZR module shipments was doubled for 2025.”
— Scott Wilkinson, Lead Analyst, Cignal AI, February 2023 |
Interoperability is not a market convenience for Tier-1 carriers — it is an architectural requirement. Operators with tens of thousands of access nodes cannot accept single-vendor dependency in their pluggable ecosystem. The 400ZR market made this lesson concrete: operators who committed to proprietary 400G solutions before the OIF IA was finalized found themselves managing a second migration to standards-compliant modules at additional cost. Operators evaluating 100ZR today are applying that experience directly, requiring OIF compliance and demonstrated multi-vendor interoperability before committing to deployment programs.
Where the Industry Stands — and What Comes Next
The engineering case for coherent at the access layer is no longer purely prospective. Power consumption is reaching access-compatible ranges in commercially available hardware. Access-optimized DSP architectures have demonstrated that the thermal and management constraints of the access domain can be addressed by design. The standards ecosystem is in active formation, with multi-vendor interoperability demonstrated at the industry’s leading technical forums across 2024 and 2025. What will determine adoption speed is less the technology and more the operational models’ operators choose — deployment strategies will vary by access architecture, installed base density, and how operators weigh migration cost against continued capacity constraints. That migration question is the subject of Paper 3.
#NetworkArchitecture #AccessNetworks #CoherentOptics #LegacyMigration #AIScaling
About Arycs Technologies
Arycs delivers power-efficient, coherent-class optical connectivity based on silicon photonics, coherent DSP, and advanced optical architectures. Our solutions provide industry-leading bandwidth per watt, deterministic performance, and flexible network evolution for AI, cloud, telecom, and edge infrastructure. Designed for real-world deployment, Arycs Technologies enables networks to scale with growing AI demand without disruptive redesign or hardware replacement.
arycs-tech.com | LinkedIn: Arycs Technologies
The Changing Economics of Coherent at the Network Edge [2/6]
