Optics bottleneck
InfrastructureJune 17, 20264 min read

Coherent’s $50 Million CHIPS Award Turns AI Optics Into a Manufacturing Bottleneck Story

Coherent’s June 16 CHIPS letter of intent for its Sherman, Texas indium-phosphide facility is more than a subsidy headline. It shows that AI infrastructure bottlenecks are moving deeper into optical interconnects, wafer throughput, and domestic manufacturing capacity that sits between the GPU and the data center.

By Nawaz LalaniPublished June 17, 2026
More in Infrastructure
At a glance
  • Coherent’s June 16 announcement is one of the strongest infrastructure stories worth publishing this run because it highlights a bottleneck most casual AI coverage still misses.
  • The press release makes the capacity point explicit.
  • The useful angle is where indium phosphide sits in the stack.
Article details
Section
Infrastructure
Read time
4 min read
Data included
Why Sherman matters
Illustrated hero showing optical interconnects, wafers, and data-center links as AI infrastructure bottlenecks
Image note
Coherent's Sherman expansion is a reminder that the next AI bottlenecks are not just GPUs and megawatts. They also include the photonics hardware that moves data inside AI factories.
Data snapshot

Why Sherman matters

The Sherman expansion is a narrow manufacturing story with broad infrastructure consequences because it sits inside the AI interconnect layer.

Visual brief

Coherent’s June 16 manufacturing signals

Direct CHIPS support
$50M
Coherent said the letter of intent provides up to $50 million in direct funding.
Wafer-capacity increase
4x
Coherent said the Sherman project would quadruple wafer production capacity.
Expected jobs
1,000+
The company expects more than 1,000 jobs, including over 550 direct technical roles.
Manufacturing signalWhat was announcedWhy it matters
Federal supportUp to $50 million in CHIPS direct fundingShows photonics is now being treated as strategic AI infrastructure.
Facility scaleProduction space doubles and wafer capacity quadruplesSuggests demand for optical components is scaling faster than current domestic throughput.
Prior incentivesAbout $20 million of state and local support already committedReinforces that multiple layers of policy are backing this capacity buildout.
Commercial anchorExpansion deepens Coherent’s recently expanded NVIDIA relationshipConnects optical manufacturing directly to hyperscale AI cluster demand.

Source: Coherent’s June 16, 2026 press release and NVIDIA’s March 2, 2026 partnership announcement.

Coherent’s June 16 announcement is one of the strongest infrastructure stories worth publishing this run because it highlights a bottleneck most casual AI coverage still misses. The company said it signed a letter of intent to receive up to $50 million in direct CHIPS funding to expand its 6-inch indium-phosphide semiconductor facility in Sherman, Texas. Coherent said the project will double manufacturing production space and quadruple wafer production capacity. That matters because AI factories do not scale on GPUs alone. They also need dense, fast, energy-efficient optical links that can move data across racks, systems, and clusters without turning the interconnect into the next constraint.

The press release makes the capacity point explicit. Coherent said the Sherman expansion supports growing demand for optical networking technologies that power AI datacenters and strengthens its recently expanded partnership with NVIDIA. At completion, the site is expected to create more than 1,000 jobs, including more than 550 direct manufacturing, engineering, and technical roles. The project also builds on roughly $20 million of prior support from Texas state and local development programs. In other words, this is not just a grant. It is a layered industrial policy effort around a specific component of the AI supply chain.

The next AI choke points are not only chips and power. They are the optical components and wafer lines that keep giant compute clusters connected.

The useful angle is where indium phosphide sits in the stack. In March, NVIDIA and Coherent said they had signed a multiyear strategic agreement under which NVIDIA would invest $2 billion in Coherent, deepen access to advanced laser and optical networking products, and support U.S.-based manufacturing. NVIDIA described optical interconnects and advanced package integration as foundational to the next phase of AI infrastructure because they enable ultra-high-bandwidth, energy-efficient connectivity across AI factories. Coherent’s June 16 update is the manufacturing follow-through on that thesis.

That is why this story clears the duplicate bar even in a site already heavy on power and compute. The next AI infrastructure bottlenecks are not only megawatts, transformers, and land. They also include the hardware that keeps increasingly dense compute clusters talking to each other. If optical components lag, GPU supply alone does not solve the problem. Clusters become harder to scale efficiently, latency and throughput constraints get worse, and the economics of bigger AI systems can degrade in places that are less visible than the power queue.

Operators and investors should read this as a clue about where strategic value is concentrating. Power-ready campuses still matter, but so do the manufacturing nodes that determine whether optical networking hardware can scale fast enough to keep up with model and rack density. The companies that own those choke points may capture more durable pricing power than generic “AI infrastructure” baskets imply, especially when domestic capacity, supply-chain resilience, and government support all start reinforcing each other.

The final signal is political as well as industrial. Federal CHIPS support, Texas incentives, and NVIDIA purchase commitments are converging around a narrow but critical layer of the stack. That suggests policymakers and hyperscale buyers both understand that AI competitiveness is now about securing entire production systems, not merely buying more accelerators after the fact.

The publishable conclusion is simple: AI infrastructure bottlenecks are moving downstream into optics and wafer throughput. Coherent’s Sherman expansion is evidence that the industry has started treating that layer as strategic.

Sources

Coherent, “Coherent Announces a CHIPS Letter of Intent for $50 Million to Expand World-Leading Manufacturing Facility for AI Infrastructure,” published June 16, 2026: https://www.coherent.com/news/press-releases/a-chip-letter-of-intent-for-50m-to-expand-world-leading-manufacturing-facility-for-ai-infrastructure

NVIDIA Investor Relations, “NVIDIA and Coherent Announce Strategic Partnership to Develop Optics Technology to Scale Next-Generation Data Center Architecture,” published March 2, 2026: https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-and-Coherent-Announce-Strategic-Partnership-to-Develop-Optics-Technology-to-Scale-Next-Generation-Data-Center-Architecture/default.aspx

Author and standards

By Nawaz Lalani

The Grid Report is written by Nawaz Lalani and focuses on source-backed coverage of AI infrastructure, grid power demand, automation systems, and market signals.

Related reporting
Get the brief

Follow the signal, not just the headline.

Get the daily Grid brief for source-backed coverage on AI power demand, infrastructure timing, automation, and market signals.