How Coherent and NVIDIA are Scaling Data Centre Connectivity

Coherent is expanding the manufacturing capacity behind one of the most critical technologies in modern AI data centres.

The company has broken ground on an enlarged facility in Sherman, Texas, where it will produce indium phosphide semiconductors used in the optical networking systems that connect AI infrastructure at scale.

While the announcement focuses on manufacturing expansion, its effect actually reaches far beyond the factory floor.

With hyperscale operators working to build larger AI clusters, demand is surging for the photonic components that enable high-speed data transmission between chips, servers and racks.

Coherent manufactures lasers, optical components and compound semiconductors and operates what it describes as the world's first 6-inch indium phosphide fabrication facility.

The Sherman site produces the InP wafers used in optical interconnects and high-speed networking systems that support AI workloads.

Building the Optical Backbone of AI

The expansion follows a major commitment from NVIDIA, which announced in March 2026 that it would invest US$2bn in Coherent to expand supply, deepen research and development and strengthen US manufacturing capabilities.

Speaking at the groundbreaking ceremony, Jensen Huang, CEO of NVIDIA, said: “Coherent is a world-class company, and the work you do is vital to our future, vital to the future of artificial intelligence and vital to reindustrialising the United States.”

The project is expected to create more than 1,000 jobs when complete, including more than 550 advanced manufacturing, engineering and technical positions.

Expanding Capacity for AI Infrastructure

At the event, Coherent also announced a US$50m CHIPS Act grant to support the expansion.

The funding builds on approximately US$17m in previous support from the Texas CHIPS programme and the Sherman Economic Development Corporation.

The project will add advanced wafer fabrication equipment and increase cleanroom capacity, enabling greater production of InP-based photonic devices.

According to Coherent, the expansion will double manufacturing space and quadruple wafer production capacity. The company says this will significantly increase domestic production of technologies essential to AI infrastructure.

“Today marks an important milestone, not just for Coherent, but for American manufacturing and for the future of AI infrastructure,” Coherent CEO Jim Anderson said at the event.

Why AI Data Centres Need Photonics

The investment comes as data centre architectures evolve to support increasingly large AI deployments.

Jensen explained that connecting hundreds of thousands of processors across large facilities requires optical networking technologies rather than traditional copper-based connections.

In systems such as NVIDIA's Vera Rubin Ultra NVL576 platform, which links eight racks containing 72 Rubin Ultra GPUs each into a single 576-GPU domain, optical interconnects are necessary because copper cannot efficiently transmit signals across those distances.

As signalling rates increase, copper connections require additional power-intensive equipment for signal conditioning and retiming.

Optical technologies provide a more efficient alternative, allowing operators to dedicate more power to computing resources rather than networking overhead.

Jensen said: “AI factories are the infrastructure of the new industrial revolution. Connecting millions of GPUs into one thinking machine requires optical technology built for scale, speed and energy efficiency.

"Coherent has been an important NVIDIA partner for more than two decades, and its expanded InP manufacturing in Texas will help strengthen the US supply chain for the AI infrastructure the world is racing to build.”