Can 800G ZR+ Replace the Optical Transport Shelf? A Data Center DCI Guide

Can 800G ZR+ Replace the Optical Transport Shelf? A Data Center DCI Guide

FiberQuotes • Data Center Interconnect Report • 2026

800G Between Two Data Centers Without a Traditional Optical Chassis: How ZR+ Pluggables Are Changing DCI

A conventional data-center interconnect can contain routers, short-reach optics, transponders, optical shelves, amplifiers and DWDM equipment before the traffic ever reaches the long-haul fiber. 800G ZR and ZR+ coherent pluggables are collapsing much of that stack into the router port itself. Coherent’s newest pluggable optical line system pushes the idea further — potentially allowing as much as 25.6 Tbps between sites without a traditional transport chassis.

Capacity per wavelength Up to 800G
Coherent POLS wavelengths Up to 32
Fiber-pair capacity 25.6 Tbps
POLS target distance 2–200 km

For decades, moving serious capacity between two data centers generally meant building two networks on top of one another. The IP network handled packets. A separate optical transport network converted those packets into long-distance wavelengths, amplified them, multiplexed them and carried them across the fiber. That architecture remains extraordinarily capable — and necessary in many networks — but coherent pluggable optics are steadily erasing the line between the router and the optical transport system.

Can You Really Skip the Big Optical Shelf?

For the right point-to-point DCI link, yes: the dedicated transponder or muxponder shelf can increasingly disappear.

An 800G ZR/ZR+ coherent transceiver can plug directly into a compatible QSFP-DD or OSFP router or switch port and generate a tunable DWDM wavelength itself.

What does not disappear is optical engineering. Multi-wavelength systems still need wavelength combining, fiber, amplification when required, monitoring and appropriate power budgets. Coherent’s new POLS matters because some of those remaining line-system functions can now also live in pluggable modules.

The Old DCI Architecture Had an Extra Conversion Layer

The most important architectural change is not simply that the new optic is faster. It is that the router can generate the long-distance coherent wavelength itself.

Traditional Optical Transport

Router / Switch
→
Gray Client Optic
→
Transponder / Muxponder
→
Optical Line System
→
DWDM Fiber

The router sends a short-reach client signal to a separate transport platform. The transport platform creates the coherent wavelength.

800G IP-over-DWDM

800G Router / Switch
→
800G ZR / ZR+ Pluggable
→
Compact / Passive Optical Layer
→
DWDM Fiber

The coherent wavelength originates directly at the router or switch, eliminating the external transponder conversion for suitable applications.

This is often called IP-over-DWDM or routed optical networking. The packet device is no longer handing traffic to a separate box whose primary job is converting an Ethernet client signal into a coherent DWDM wavelength.

How Did a Transponder Become Small Enough to Plug Into a Router?

Change 01

Smaller Coherent DSPs

New semiconductor processes allow sophisticated coherent digital signal processing to operate within QSFP-DD and OSFP thermal and power envelopes.

Change 02

Integrated Photonics

Lasers, modulators, receivers and optical integration have advanced enough to move functions once associated with large transport cards into pluggable modules.

Change 03

Industry Standards

OIF ZR, OpenZR+ and OpenROADM specifications have created increasingly interoperable operating modes instead of requiring every coherent interface to be proprietary.

Coherent Just Shrunk More of the Line System Too

Coherent’s September 2026 Pluggable Optical Line System expansion is especially interesting because it addresses the equipment that remains after the transponder disappears.

Optical capacity 25.6 Tbps
DWDM channels 32 × 800G
Target range 2–200 km
Amplifier output 21 dBm

The enhanced POLS uses a full-C-band, high-power variable-gain amplifier packaged in a compact QSFP form factor. Coherent says it is designed for the current generation of 800G ZR/ZR+ optics and provides autonomous plug-and-play operation.

The significance is architectural: the operator may no longer need a conventional rack-mounted optical chassis simply to provide amplification for a straightforward metro DCI span.

32 Wavelengths Changes the Capacity Math Quickly

1 800G wavelength 0.8 Tbps
2 800G wavelengths 1.6 Tbps
4 800G wavelengths 3.2 Tbps
8 800G wavelengths 6.4 Tbps
16 800G wavelengths 12.8 Tbps
32 800G wavelengths 25.6 Tbps

That capacity travels over a single fiber pair because dense wavelength-division multiplexing places each coherent carrier at a different optical frequency.

This distinction is important. Thirty-two 800G interfaces do not require 32 separate long-distance fiber pairs. With a properly designed DWDM system, they can share the same pair.

800ZR and 800G ZR+ Are Related — But They Are Not the Same Tool

Characteristic 800ZR 800G ZR+
Primary goal Standardized high-density point-to-point DCI Higher-performance metro / regional / extended DCI
Full 800G wavelength Yes Yes
Small pluggable form factor QSFP-DD / OSFP QSFP-DD / OSFP
Direct router insertion Yes Yes
Standard DCI target Short / metro point-to-point Metro through regional and beyond
FEC / modulation flexibility Optimized standardized profile Greater performance flexibility
Lower-rate modes for longer reach More limited Common ZR+ advantage
ROADM network use Primarily point-to-point Better suited

Do Not Confuse the POLS 200 km Limit With the Limit of ZR+

This distinction matters.

Coherent’s new POLS product targets interconnects from approximately 2 to 200 km. That does not mean an 800G ZR+ coherent module stops working at 200 km.

Current 800G ZR+ implementations from major vendors support substantially longer distances when paired with an appropriate amplified DWDM system. Cisco, for example, specifies more than 1,000 km for its 800G ZR+ module under suitable line-system conditions. Nokia likewise offers 800G ZR+ implementations targeting regional and long-haul networks.

Think of 2–200 km as the sweet spot of this compact Coherent line-system architecture — not the absolute transmission limit of coherent pluggable technology.

A Practical Distance Guide

2–10 km Campus or nearby data centers. Coherent may be useful primarily for fiber efficiency and DWDM scaling rather than reach.
10–50 km Excellent metro DCI territory. Direct coherent pluggables can greatly simplify the architecture.
50–120 km Core ZR territory. Fiber loss determines whether amplification is required.
120–200 km ZR+ and amplified DCI become increasingly relevant. This is within Coherent’s new POLS target range.
200 km+ Pluggable ZR+ may still work extremely well, but expect a more engineered optical line system and possibly ROADMs or regeneration.
Distance alone does not determine reach. Fiber loss, connector loss, splices, dispersion, OSNR, amplifier noise, channel count and ROADM traversal all contribute to whether a particular 800G mode will close.

What Actually Disappears?

Component Traditional DCI ZR+ / Pluggable DCI
Router / switch Required Required
Short-reach gray client optic Often required between router and transport Can disappear
Transponder Required Can disappear
Muxponder Often used Can disappear for native Ethernet wavelengths
Transport shelf Usually present Can disappear in simple architectures
DWDM mux / demux Required for multiple wavelengths Still required in some form
Optical amplification As required by span Still required when optical budget demands it
Dark fiber pair Required or carrier optical network used Still required for private DCI

What You Do Not Get to Eliminate

Still Required

Optical Power Engineering

Launch power, receiver power, amplifier gain and span loss still have to fall inside the supported operating envelope.

Still Required

Wavelength Management

Multiple coherent channels require a channel plan and optical multiplexing. Tunable optics simplify wavelength assignment but do not abolish DWDM.

Still Required

Fiber Diversity

One 25.6 Tbps fiber pair can create an enormous single failure domain. A second physical route matters more, not less, as capacity concentrates.

The Economics Are Bigger Than the Price of the Optic

Fewer Boxes Removing the transponder layer can eliminate chassis, line cards and client optics.
Less Rack Space Valuable data-center rack units no longer need to be consumed by a separate transport platform.
Less Power One coherent pluggable replaces several electrical and optical conversion stages.
Less Management Optical parameters can increasingly be monitored directly from the router instead of through a completely separate transport-management layer.

Cisco, Juniper and other routed-optical vendors report substantial reductions in equipment, facilities and operating cost when coherent pluggables replace external transponders. Exact savings vary too much by topology to treat any vendor percentage as a universal enterprise benchmark.

The Router Port Is Now Doing Much More Work

In the old architecture, a router mostly had to provide an Ethernet client signal. The optical transport equipment worried about wavelength tuning, dispersion, modulation and coherent transmission.

In an IP-over-DWDM architecture, much of that functionality is visible directly through the coherent optic hosted in the router.

Modern platforms can expose optical parameters such as receive power, transmit power, chromatic dispersion, FEC performance and wavelength settings through the router’s own management interfaces.

That operational convergence may ultimately be as important as eliminating a chassis: the IP and optical layers become part of one troubleshooting and automation workflow.

Where the Traditional Optical Chassis Still Wins

Requirement Pluggable-First DCI Traditional Transport Platform
Simple point-to-point DCI Excellent fit May be more equipment than necessary
2–200 km DCI Strong target for compact architecture Still valid
Complex multi-node ROADM mesh Possible with appropriate ZR+ Strong fit
Heavy OTN grooming Not the primary strength Strong fit
Mixed legacy client services May require additional handling Purpose-built
Very long-haul network ZR+ increasingly capable Full photonic engineering toolkit
Complex optical protection Can be engineered at IP / optical layers Mature transport options
Maximum spectral efficiency Very strong, but constrained by pluggable envelope High-end coherent engines may offer more flexibility

The 200 km Enterprise DCI Sweet Spot

For many enterprises, the most interesting application is not a continental backbone. It is two or three major facilities in the same metro or neighboring region.

Examples include a primary data center connected to a disaster-recovery site, two colocation facilities, an enterprise campus connected to a cloud on-ramp facility, or two AI-compute locations separated by tens or hundreds of kilometers.

Strong Pluggable Candidate

  • Two fixed endpoints.
  • Private dark fiber available.
  • Distance under roughly 200 km.
  • Routers support 800G coherent optics.
  • Native Ethernet traffic dominates.
  • Need ranges from 800G to multiple terabits.
  • No complex intermediate optical routing.

Consider a Full Transport System

  • Many intermediate sites.
  • Multiple ROADMs.
  • Complex wavelength add/drop requirements.
  • OTN grooming is important.
  • Numerous legacy client rates.
  • Specialized protection schemes.
  • Regional or long-haul mesh topology.

There Is Also a Huge Resilience Question

Packing 25.6 Tbps onto one fiber pair is an extraordinary utilization of physical infrastructure.

It is also an extraordinary amount of capacity to lose to one backhoe.

At this scale, a resilient DCI design should normally be discussed in terms of physically diverse routes rather than simply two logical links.

Two 800G wavelengths on the same fiber are not redundant. Thirty-two wavelengths on the same cable are not 32 independent paths either. The physical duct, cable, building entrance and carrier route are the failure domain that matters.

A Second Fiber Pair May Be More Valuable Than Another Transport Shelf

Simplifying the electronics can free budget for something much harder to replicate: physically diverse fiber.

For example, an enterprise planning 6.4 Tbps between two facilities could use eight 800G wavelengths on one route. A more resilient design could place another eight wavelengths on a geographically separate route.

That doubles optical equipment and dark-fiber requirements, but it protects the business against the failure that coherent pluggables cannot solve: a physical cable cut.

Do You Need 800G Immediately?

Not necessarily.

One of the benefits of ZR+ is that the newest modules can support multiple operating modes. Depending on vendor and platform, an 800G-capable coherent optic can operate at lower line rates to gain additional reach or improve optical margin.

That creates a useful lifecycle strategy: install an 800G-capable router and coherent optical ecosystem, then choose the capacity/reach tradeoff appropriate for the actual fiber path.

An “800G optic” should increasingly be viewed as a programmable coherent engine, not simply a fixed 800G laser.

Eight Questions to Ask Before Removing the Transport Shelf

Question Why It Matters
Does the router actually support this 800G ZR/ZR+ module? Mechanical fit is not enough. Check software, thermal budget, power class and vendor qualification.
What is the measured end-to-end fiber loss? Route distance alone does not prove the optical link will close.
How many wavelengths will we need? One wavelength can be simple. Sixteen or 32 require a deliberate DWDM design.
Is amplification required? Longer spans and multi-channel systems change the line-system requirement.
Are there intermediate ROADMs? Each optical node adds filtering, loss and OSNR penalties.
Do we require OTN grooming? A router-native wavelength architecture may not replace every service function traditionally provided by transport equipment.
Is the second path physically diverse? Logical redundancy is meaningless if both links share the same conduit.
Who will own optical troubleshooting? Removing the optical shelf does not remove responsibility for wavelength, power, OSNR and fiber-performance troubleshooting.

When the Business Case Gets Especially Strong

Best Case 01

Existing Dark Fiber

If the enterprise already controls a good fiber pair between two facilities, coherent pluggables can turn that passive asset into multi-terabit transport without buying managed wavelength services individually.

Best Case 02

High Router Port Density

A platform already equipped with 800G-capable QSFP-DD or OSFP ports reduces the need to introduce another active transport layer.

Best Case 03

Rapid Growth

DWDM allows additional wavelengths to be lit progressively instead of installing new long-distance fiber pairs every time inter-site demand increases.

The Bigger Trend Is Not Really 800G

800G is the capacity headline. The architectural story is convergence.

At 400G, the industry demonstrated that coherent DWDM technology could move from specialized transport shelves into familiar switch and router pluggables. The OpenZR+ ecosystem was created specifically to extend that architecture beyond the original 400ZR point-to-point DCI envelope.

800G extends the same model to twice the wavelength capacity. Coherent’s pluggable line-system work now asks the next question: if the transponder can be turned into a module, how much of the line system can become modular too?

The likely future is not “optical transport disappears.” It is that operators get much more choice about which optical functions deserve a chassis and which can live directly inside the packet platform.
Research basis: Coherent’s September 17, 2026 Pluggable Optical Line System announcement and POLS product documentation; Coherent 800G ZR/ZR+ transceiver documentation; Cisco’s July 2026 800G ZR/ZR+ coherent-optics specifications and Routed Optical Networking materials; OpenZR+ MSA documentation; Nokia ICE-X 800G ZR/ZR+ documentation; Juniper JCO800 and Converged Optical Routing Architecture documentation; and current OIF/OpenROADM/OpenZR+ interoperability specifications.

Published reach figures are product and architecture targets, not guarantees for every fiber span. Actual transmission performance depends on fiber type and age, attenuation, connector and splice loss, OSNR, amplifier design, channel count, ROADMs, modulation, FEC mode and operating margin.

Can You Skip the Transport Shelf?

Enter a proposed data-center interconnect. The tool calculates the number of 800G wavelengths required, total pluggable count, fiber-pair capacity utilization and whether the project resembles the 2–200 km point-to-point architecture targeted by compact pluggable line systems.

Wavelengths per route —
Total coherent pluggables —
Fiber-pair utilization —
Architecture fit —
Architecture Endpoint Optics / Transponders Fixed Optical Equipment Router-Port Cost Modeled Hardware Total
—

    Planning model only. “Transport shelf fit” evaluates topology at a high level and does not establish optical feasibility. A real implementation requires an engineered link budget and validation of transceiver compatibility, launch power, receive power, OSNR, dispersion, amplification, filtering, channel plan, fiber loss and protection. Product pricing varies widely; all dollar fields default to zero so the tool never invents equipment prices.