AI data centers face a bottleneck in optical networking, according to a new market argument

AI is moving more data than ever, and a Yahoo Finance commentary argues that the infrastructure bottleneck may be less about AI chips and more about how data gets moved through the network. The piece frames optical networking as an area that is often overlooked by investors focused on accelerators and servers.

The commentary centers on the idea that training and inference workloads stress the entire data movement chain. As AI clusters scale, the need to move large volumes of traffic quickly between racks, switches, and across sites becomes a central operational constraint, the article contends.

Optical networking, broadly, refers to using light to transmit data through fiber or optical components, rather than relying only on electrical indicating. In AI data centers, that matters because fiber-linked systems can carry high bandwidth with lower announcement degradation than purely electrical interconnects over distance, while also helping operators manage power and cooling tradeoffs. The Yahoo Finance post argues that this network layer will be increasingly important as AI deployments expand.

The article’s investment thesis, as described in its title, ties the optical networking theme to Coherent, a company commonly associated with technologies used in high-performance communications. However, the commentary provided here does not include details on what specific Coherent products the author expects to benefit, what customer programs are in play, or how the company’s revenue or guidance maps to AI networking demand.

NVIDIA is referenced in the piece’s framing, reflecting how the company has become a central supplier to AI computing stacks. NVIDIA’s role in AI infrastructure matters because the performance of the training and inference hardware can only translate into system-level speed if the surrounding compute and communications layers keep up. Even when GPU supply and throughput are strong, the network can become the limiting factor for end-to-end application performance, the commentary implies.

Still, the case has an important caveat: the underlying Yahoo Finance post is not included in the material provided, so key elements of the argument, including any named contracts, product roadmaps, or quantified estimates of demand, cannot be verified here. As a result, readers should treat the “stock to own” framing as an opinion rather than a disclosed, company-backed forecast.

What can be said with confidence from the available information is that optical networking is being positioned by market commentators as a potentially constraining part of AI systems, and that Coherent is being singled out as a beneficiary in that narrative. The broader investment question becomes whether buyers of AI infrastructure are willing to fund next-generation optics and associated equipment at a pace that matches compute growth.

Investors and industry watchers will likely look for confirmation through primary disclosures such as customer orders, supply agreements, and component-level performance milestones, as well as for NVIDIA ecosystem indicates indicating tighter integration of compute and networking. Until more detailed, source-backed evidence is available, the thesis should be viewed as a thematic bet on data movement bottlenecks rather than a specific, validated forecast of earnings.