Earlier this year, we highlighted several trends shaping the future of IP-over-DWDM (IPoDWDM), including AI-driven data center interconnect (DCI), the growing use of coherent pluggables and simpler, more automated optical architectures.

Six months on, the market data is putting hard numbers behind those trends. In June, Dell’Oro forecasted the IPoDWDM market will grow at a 27% compounded annual growth rate over the next five years, surpassing $7 billion by 2030. That growth is driven by the demand for 800ZR+ optics and the growing need to connect distributed AI infrastructure.

August’s latest quarterly figures reinforce that momentum. Dell’Oro reports that DCI revenue from IPoDWDM ZR/ZR+ and WDM systems grew 45% year over year in Q2 2026, while disaggregated DWDM optical line systems grew 80%. Cloud providers also accounted for 34% of optical transport market revenue, with their purchases of DWDM systems and IPoDWDM ZR/ZR+ optics both growing at double-digit rates.

The significance of this goes beyond traffic growth. AI is changing where networks need to provide connectivity – and that creates a new role for optical transport.

AI is driving scale-across DCI, not just more traffic

The massive power, cooling and water demands of AI infrastructure cause data centers to spill across multiple sites, driving new requirements for connectivity.

This is different from the traditional data center networking model. Instead of simply connecting users and applications to a data center, scale-across DCI connects distributed compute resources, allowing multiple facilities to operate as a single AI environment. This requires much higher networking reliability and capacity.

Dell’Oro identifies this shift toward scale-across DCI as a major driver of IPoDWDM growth, citing stronger demand for ZR+ optics as cloud and AI infrastructure providers connect data centers over longer distances.

This has important implications for optical transport. Bandwidth remains fundamental, but power consumption, resiliency, deployment speed and operational efficiency have become increasingly important. As optical connectivity becomes more closely integrated with distributed compute, the network must scale without adding unnecessary layers of complexity.

Why IPoDWDM is gaining momentum

IPoDWDM places coherent transceivers directly into routing platforms, reducing the need for a separate transponder layer.

For high-capacity DCI and AI scale-across, that can make it easier to add capacity as demand grows, while reducing the amount of dedicated optical infrastructure that needs to be deployed and managed.

The next step is already taking shape. While the industry transitions from 400Gbit/s to 800Gbit/s deployments, the foundations for the next generation of coherent pluggables are already being put in place. OIF recently published its 1600ZR implementation agreement, defining an interoperable 1.6Tbit/s coherent interface for DCI applications and providing a common foundation for the industry’s next jump in capacity. This evolution provides a clear path to greater capacity while preserving much of the existing operational model.

That doesn’t mean every optical network will become pure IPoDWDM. Transponders and muxponders continue to have a role in mixed environments, legacy networks and applications where their capabilities provide specific operational or commercial benefits, for example, in managed optical fiber network (MOFN) services where demarcation and comprehensive management are key. The more realistic shift is toward architectures that use each approach where it makes the most sense.

Simplicity matters at AI scale

Higher capacity alone won’t solve the operational challenge. AI infrastructure can require networks to be deployed and expanded at a pace that traditional optical operating models were not designed to support.

IPoDWDM can help organizations keep pace with this rapid growth by simplifying network scaling through router-integrated coherent optics. However, that simplicity must extend beyond the routers and pluggable optics. Removing the transponder layer does not eliminate the need for an optical layer that delivers data over distance.

An optical layer that requires manual planning and intervention whenever capacity needs increase can undermine the benefits of a simplified IPoDWDM architecture. The open line system (OLS) must be easy to deploy and manage, while preserving the visibility and control needed to operate high-capacity networks with confidence.

IPoDWDM is emerging as the architecture of choice for AI scale-across.

The OLS is an important part of the equation

The OLS has always been a crucial component in DCI for delivering the required capacity over distance. More recently, optical line protection has become increasingly relevant for hyperscale architectures. And as network growth accelerates, operational simplicity is becoming just as important as capacity itself.

Traditional OLS platforms often require significant optical expertise for deployment and day-to-day operations. Purpose-built open line systems optimized for IPoDWDM reduce much of this complexity and can help operators deploy and manage high-capacity links without requiring specialists at every stage. This is exactly the approach behind a new generation of IPoDWDM-optimized open line systems. 

The goal is straightforward: enable operators to deploy, scale and troubleshoot their optical links end-to-end at the pace required by AI infrastructure growth without relying on specialized optical experts or separate IP/optical operational teams.

The resulting solution is a differentiated architecture matched to the capacity and operational requirements of IPoDWDM-based AI scale-across as well as traditional DCI.

What network operators should look for next

As IPoDWDM adoption accelerates, the key question operators must address is how the complete architecture will perform as the network grows.

Three considerations should be high on the list:

• Capacity headroom: Networks need a path from today’s 400Gbit/s deployments toward higher-capacity optics without requiring wholesale architectural changes.
• Operational simplicity: Eliminating a transponder layer only delivers its full benefit if the remaining optical infrastructure can be deployed, monitored and maintained efficiently.
• Architectural flexibility: IPoDWDM should complement rather than constrain the wider network. Operators need the freedom to combine router-integrated optics, open line systems and traditional optical technologies where each provides the best fit.

The next generation of optical networking

As AI compute scales across campuses and multiple data centers, efficient, high-capacity connectivity becomes increasingly important, while keeping power consumption, deployment complexity and operational overhead under control.

The rapid growth of IPoDWDM, strong demand for 800ZR+ optics and accelerating adoption of disaggregated OLS highlight the industry’s focus on achieving both scale and operational simplicity.

For operators and cloud providers, the opportunity is to aim for that trajectory now, building an optical infrastructure that can be expanded as quickly and efficiently as the AI environments it connects.

Learn more about Adtran’s OLS for IPoDWDM or talk to us for more information.