Key Takeaways
- Today, we are announcing a multi-year agreement with Corning worth up to $6B. Corning will supply fiber optic cables for our data centers.
- By choosing Corning’s fiber-optic cables, we are supporting advanced manufacturing in the United States and helping the country remain competitive in the global AI industry.
- The agreement will also help increase jobs at Corning’s North Carolina facilities by 15-20%.
Meta has declared a multi-year, $6B fiber supply agreement with Corning. This move shows that AI infrastructure issues now go beyond computing power and include the physical network.
Corning will provide Meta with optical fiber cable and connectivity solutions. According to both companies, this will help speed up the construction of cutting-edge data centers in the United States to support Meta’s apps, technologies, and AI goals.
As large tech companies build larger AI clusters, warnings from companies like Microsoft about a potential networking bottleneck are drawing attention to physical network capacity. Fiber is now seen as a key factor limiting the growth of AI data centers.
Until now, most discussions about AI infrastructure have focused on access to GPUs and power. However, the massive data traffic generated by large AI models is pushing data center networks to their limits, prompting cloud providers to revise their strategies.
GPU power and cooling have always been seen as key limits to data center growth, but now fiber, once considered just a basic resource, is becoming important in its own right, said Shriya Mehrotra, Director Analyst at Gartner.
Mehrotra also said that as large tech companies secure long-term fiber supplies and invest in dedicated connections, competition for capacity is growing. This makes it harder for other businesses to get access and is causing longer deployment times.
The Role of Fiber in AI Scaling
As AI systems grow, network limits are starting to slow down performance. This means expensive GPUs are not used as much as they should be, which lowers the return on big infrastructure investments.
Manish Rawat, a semiconductor analyst at Technical Insights, said that optical fiber is now becoming the next big limit on AI growth, and this could have long-term effects.
Fiber is the silent dependency that scales disproportionately with AI growth, Rawat said. AI workloads generate massive east-west traffic requiring tight harmonization across thousands of GPUs, which sharply increases intra-data-center and inter-campus optical demand.
However, Sanchit Vir Gogia, chief analyst at Greyhound Research, said the so-called networking wall was not just one bottleneck.
It’s an overlapping set of constraints that surface when AI workloads hit scale:
- Fiber Availability
- Switching Density
- Optical Trans-receiver Limits
- Architectural Inefficiencies
Gogia said.
The combined stress of AI scale and concurrent government broadband rollouts has snapped the historical assumption that fiber is abundant and cheap, Gogia added.
However, simply deploying more fiber will not be sufficient to address the challenges. Analysts said the underlying network architecture should improve as well.
Beyond raw fiber and switching, the entire network architecture needs to evolve efficiently route, process, and manage AI-generated traffic, Mehrotra said. Existing designs may not be suitable for the aggregated, bursty traffic patterns generated by massive AI deployments, requiring new architectures such as AI network fabrics and optimized data center interconnect solutions.
A New Approach to Data Center Strategy
Rawat explained that the Meta-Corning deal is not meant to fix short-term fiber shortages. Instead, it gives Meta more certainty and control as it expands its AI infrastructure.
Meta is locking in guaranteed optical capacity manufacturing priority during AI build cycles and fiber designs customized for its architectures, while insulating supply from geopolitical risks, Rawat said. It follows the same vertical integration of playbook hyperscalers used for custom AI chips, power contracts, and grid planning. Fiber is simply the next layer.
Rawat said this strategy is speeding up the creation of a two-tier networking system. Hyper-scalers now manage their own supply chains, while other businesses rely on shared resources that take longer to access and offer less flexibility.
The model shifts from buying fiber to securing fiber via forward agreements, standardization, and multi-year planning, Rawat said. Cloud Interconnect pricing will stay firm as hyperscalers absorb surplus capacity. Ultimately, network architecture choices will matter more than vendor selection for entities seeking security and scalability. Hyperscalers are effectively moving from being tenants in the fiber ecosystem to strategic owners. According to Gogia, what used to be fought out on the spot market is now being locked in years ahead to secure cost certainty, deployment speed, and operational control as AI infrastructure becomes increasingly capital-intensive.
Source: Meta-Corning fiber deal signals a new bottleneck in AI infrastructure
