AI data centers drive optical transport market to $19B by 2029

After a challenging 2024 that saw the optical transport market contract by 9%, the telecommunications infrastructure sector is poised for a significant rebound. The catalyst driving this recovery isn’t traditional telecom growth, but rather the explosive expansion of artificial intelligence computing infrastructure.

The optical transport market—which encompasses the fiber-optic cables, switches, and networking equipment that carry data across long distances—is projected to grow at a steady 5% annually through 2029, reaching $19 billion by the end of the forecast period. This turnaround represents a dramatic shift from the broader telecom industry slowdown that characterized much of 2024.

AI data centers reshape connectivity demands

The primary driver behind this market recovery is data center interconnect (DCI), the high-speed networking infrastructure that links AI computing facilities across different locations. As companies rapidly deploy AI capabilities, they’re discovering that individual data centers cannot handle the massive computational workloads alone—they need to work together as interconnected networks.

“We are anticipating that the time has come to interconnect all those new AI data centers being built,” said Jimmy Yu, vice president at Dell’Oro Group, a telecommunications market research firm. “We are forecasting data center interconnect to grow at twice the rate of the overall market, driven by increased spending from cloud providers.”

This growth stems from the unique networking requirements of AI workloads. Unlike traditional applications that can operate within a single data center, AI training and inference often require distributing computational tasks across multiple facilities. This creates enormous demand for the high-capacity optical connections that can move data between locations at the speeds AI applications require.

Technical infrastructure driving growth

The equipment purchases fueling this market expansion include several specialized components that enable long-distance, high-capacity data transmission. Cloud providers and AI companies are investing heavily in IP over DWDM systems (Internet Protocol over Dense Wavelength Division Multiplexing), which allow multiple data streams to travel simultaneously over single fiber-optic cables.

Additionally, companies are purchasing optical line systems for transport and DWDM systems specifically designed for high-performance, long-distance terrestrial and subsea transmission. These technologies enable AI data centers separated by hundreds or thousands of miles to communicate as if they were located in the same building.

Front-end networks experience explosive growth

The AI boom is creating ripple effects throughout data center networking infrastructure. Dell’Oro Group recently identified what it calls “explosive scale” in AI back-end networks—the internal connections between AI processors within data centers. This growth is driving corresponding demand for front-end networks, which handle data flowing in and out of the facility.

Data center front-end networks are projected to experience a compound annual growth rate exceeding 40% through 2029, according to Dell’Oro’s analysis. This represents one of the fastest-growing segments in the telecommunications infrastructure market.

Sameh Boujelbene, vice president at Dell’Oro Group, noted that “AI back-end deployments are breathing new life into this market.” The firm projects that nearly 90 million high-speed switch ports operating at 800 gigabits per second (Gbps) and 1.6 terabits per second (Tbps) will be integrated into front-end networks over the next five years.

To put these speeds in perspective, a single 800 Gbps connection can transfer the equivalent of approximately 200 high-definition movies per second, while 1.6 Tbps connections can handle twice that volume.

Emerging connectivity requirements

Perhaps more significantly, AI infrastructure is creating entirely new categories of networking demand. Boujelbene highlighted an often-overlooked requirement: “Within front-end networks, there’s now a growing need for a new segment of connectivity—linking accelerated servers not to each other, but to the front-end network for data ingest.”

This refers to the specialized connections needed between AI processing units (such as graphics processing units designed for machine learning) and the broader network infrastructure. These connections require high-speed connectivity and command premium pricing due to their specialized performance requirements.

Market outlook and implications

While some deployment delays have occurred as companies navigate the technical complexities of AI infrastructure buildout, the underlying demand remains robust. Dell’Oro projects that shipments for back-end networks will be more than triple those for front-end networks, indicating the massive scale of internal AI processing infrastructure being deployed.

For businesses and investors, this optical transport recovery signals a fundamental shift in how computing infrastructure is being architected. The days of isolated data centers are giving way to interconnected networks of facilities that can dynamically share computational workloads based on capacity, geographic requirements, and cost optimization.

The 5% annual growth rate for optical transport may seem modest compared to the explosive growth in AI applications themselves, but it represents a significant turnaround for a mature infrastructure market. More importantly, it reflects the essential role that high-speed, long-distance connectivity will play in the AI-driven economy of the coming decade.