Prysmian Secures Molex Optical Cable Supply Deal
For practitioners, large AI clusters depend on data-center interconnect capacity as well as accelerators, power, and networking silicon. Long-term optical-cable procurement can therefore become a material infrastructure constraint for high-density deployments. The Next Web reports that Prysmian signed a 10-year, 5.5 billion euro agreement to supply Molex with optical cable for use inside AI data centers. Molex is providing 550 million euros upfront, according to The Next Web. The same report states that Prysmian is allocating 1.25 billion euros for capacity investment through 2031 and projects that the program will more than double its US optical-fiber production capacity.
For practitioners, AI data-center design increasingly depends on physical interconnect availability alongside GPU supply, electrical power, cooling, and switching equipment. Industry-pattern observations: as clusters add racks and raise bandwidth demands, the quantity, density, and qualification requirements of intra-data-center fiber can become procurement constraints rather than commodity details.
The Next Web reports that Prysmian signed a 10-year, 5.5 billion euro agreement with Molex to supply optical cable for AI data centers. The cable is intended for deployment within facilities, linking servers and switching equipment, rather than for long-haul networks between sites. According to The Next Web, Molex will make a 550 million euro upfront payment under the agreement.
Capacity and supply-chain implications
The Next Web reports that Prysmian is allocating 1.25 billion euros of capacity investment through 2031 and expects that spending to more than double its optical-fiber production capacity in the United States. The publication also reports that Prysmian projects more than 1,000 jobs globally from the program, including 600 in the US.
According to The Next Web, Prysmian has projected up to 10 billion euros in cumulative additional revenue from hyperscalers and data-center providers through 2035, and up to 1.1 billion euros annually from 2031 when new capacity is operating. Those wider revenue figures are company projections reported by the publication, not contracted revenue in the Molex agreement.
Why internal fiber matters
Editorial analysis
AI training and inference clusters exchange large volumes of data between compute nodes, storage, and network switches. Scaling these systems makes physical-layer decisions consequential: link reach, connector density, thermal behavior, installation complexity, failure rates, and supply lead times can all affect the pace at which a facility is brought online.
a decade-long supply agreement with an upfront payment is notable because it shifts part of the infrastructure discussion from spot procurement toward capacity reservation. Comparable arrangements can reduce uncertainty for a supplier and a systems integrator, while also making dependency mapping important for cloud operators and enterprise teams whose deployment schedules rely on new cluster capacity.
For practitioners
the relevant operational question is not simply whether fiber is available, but whether validated cabling assemblies, connectors, transceivers, switches, and installation labor can arrive on the same timeline. In comparable AI infrastructure expansions, a constraint in any one of those layers can delay usable compute even after accelerators have been procured.
Key Points
- 1Prysmian signed a 10-year, 5.5 billion euro optical-cable agreement with Molex, expanding attention beyond GPUs to AI cluster interconnect supply.
- 2Molex's 550 million euro upfront payment provides contracted cash support, while Prysmian's broader revenue figures remain reported company projections.
- 3Industry-pattern observations: fiber availability, validated assemblies, switching hardware, and installation capacity jointly influence when new AI compute becomes usable.
Scoring Rationale
The deal is a large, long-duration infrastructure commitment tied directly to AI data-center buildouts. It is highly relevant to practitioners tracking cluster deployment constraints, although it does not introduce a new model, platform, or broadly usable developer product.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
