AI data centers require 36 times more fiber than designs with standard servers — severe glass shortages push cable lead times out to a full year

AI data centers now require 36 times more fiber optic cable than traditional server setups, according to industry sources, creating significant supply chain challenges for fiber manufacturers and data center builders.

Major Chinese fiber manufacturers, Hengtong and FiberHome, have reported full capacity and extended delivery cycles, with some orders stretching into early 2027. Industry data indicates that fiber demand for AI infrastructure grew approximately 76% in 2025 and is projected to constitute 30% of global fiber demand by 2027, up from below 5% in 2024.

Leading tech companies such as Meta and Nvidia have responded to the demand surge by securing multiyear supply agreements, including Meta’s $6 billion deal with Corning and Nvidia’s $300 million investment in new manufacturing plants. However, these new facilities are not expected to be operational until 2027 or later, meaning supply shortages are likely to persist in the near term.

Why It Matters

This development highlights a critical bottleneck in the expansion of AI infrastructure, as the large fiber requirement—36 times higher than standard servers—places increased pressure on supply chains. The shortage could influence the pace of AI deployment, data center growth, and industry costs.

Background

As AI applications grow rapidly, data center infrastructure must scale accordingly. In 2025, fiber demand surged by 76%, driven by AI and high-performance computing needs. Major manufacturers like Corning, Hengtong, and FiberHome are operating at full capacity, with delivery delays extending up to a year for some clients. Industry forecasts indicate that fiber demand from AI will dominate the market by 2027, requiring substantial capacity expansion.

“The demand for fiber in AI data centers is significant, requiring 36 times more fiber per setup than traditional servers, which is impacting supply availability.”

— Industry analyst

“Our production lines are operating at full capacity, and delivery times have extended from weeks to months, with some orders pushed into early 2027.”

— A representative from FiberHome

“We have secured multiyear commitments to ensure supply, but capacity expansion takes time, and current demand exceeds existing supply levels.”

— Meta spokesperson

What Remains Unclear

It remains uncertain how quickly fiber manufacturers will be able to increase capacity to meet the rising demand, or whether technological innovations could reduce fiber requirements in future data center designs.

What’s Next

Future steps include capacity expansions planned by major manufacturers, expected to be operational by 2027. Industry analysts will observe whether these efforts will sufficiently address the supply shortfall and how supply chain dynamics evolve in response to ongoing demand.

Key Questions

Why do AI data centers require so much more fiber than traditional servers?

AI data centers demand higher bandwidth and faster data transfer rates, which necessitate increased fiber optic cabling to support the data flow and connectivity requirements.

When will new fiber manufacturing capacity come online?

Major expansions are scheduled for 2027 or later, with existing projects by companies like Corning expected to take several years before full capacity is restored.

How might this fiber shortage impact AI development and deployment?

The shortage could slow the expansion of AI infrastructure, lead to increased costs, and delay some data center projects, potentially affecting AI development timelines.

Are there alternative solutions to reduce fiber requirements in data centers?

Advancements in data transmission technologies and architectural designs may eventually reduce fiber needs, but currently, no widely adopted alternatives are available to meet the increasing demand.