NVIDIA and Coherent have broken ground on a $2 billion manufacturing facility in Texas aimed at solving one of the most pressing infrastructure challenges in AI computing: optical interconnects for data centers. The joint venture, supported by CHIPS Act funding, represents a significant bet that optical networking hardware will become as critical to AI infrastructure as GPUs themselves. This move signals growing recognition that semiconductor supply chains must expand beyond processing power to include the high-speed optical components that connect AI chips within and across data centers, enabling the massive bandwidth requirements of modern large language models and training clusters.
The optical interconnect bottleneck has become increasingly apparent as AI workloads scale. While NVIDIA's Blackwell architecture and competing AI chips continue advancing rapidly, the infrastructure linking these processors together has lagged, creating efficiency constraints in hyperscale deployments. Coherent's specialization in optical components positions the partnership to directly address this gap. By securing CHIPS Act support, the facility also reflects U.S. policy priorities around domestic semiconductor and advanced manufacturing capacity, reducing reliance on overseas supply chains for critical AI infrastructure components.
This development underscores how AI infrastructure buildout now encompasses far more than GPU supply. As data centers scale to support AI applications across advertising, autonomous systems, and enterprise workloads, the complete stack—processors, interconnects, power delivery, and cooling—demands coordinated investment. NVIDIA's involvement signals the company's expanded view of its role beyond chip design into the broader infrastructure ecosystem, positioning it as an enabler of the entire computational stack necessary for the next generation of AI deployment.