AI Models & Platforms

Tower and NewPhotonics Ship Laser-Integrated PICs for AI Interconnect

mm
Add Unite.AI to your preferred sources on Google

Tower Semiconductor and NewPhotonics have begun high-volume shipments of laser-integrated, serviceable optical engine photonic integrated circuits for scale-out and scale-up AI interconnect, the companies said in a joint announcement on September 17, 2026, with volume-production PICs supporting data rates from 800G to 1.6T.

Datelined Migdal HaEmek and Tel Aviv, Israel, the announcement said the optical engines are designed to meet what the companies described as urgent and growing demand for high-bandwidth, energy-efficient optical interconnects in artificial intelligence infrastructure. The PIC optical engine chips support OEM solutions spanning FRO/LRO/LPO and Extra-Dense Packaged Optics (XPO) pluggable transceiver modules, as well as Near-Packaged Optics (NPO) socket pluggable applications.

Beyond current production, the companies said future volume manufacturing is planned for 6.4T NPO solutions used in scale-out and scale-up architectures, with CPX-MSA compliant NPC505 chipsets for 6.4T scheduled to begin volume shipment in the first half of 2027. NewPhotonics’ NPG102 PIC products for 800G and 1.6T, which are in volume shipment, use 200G-per-lane designs and heterogeneously integrated lasers on a monolithic PIC. According to the companies, the unified architecture lets customers apply optical signal processing while providing power efficiency, bandwidth density, and manufacturing consistency.

Built on Tower’s PH18DA Silicon Photonics Platform

The shipments are built on Tower’s high-volume PH18DA silicon photonics technology platform, which features heterogeneously integrated indium phosphide components. The platform enables monolithic integration of lasers, semiconductor optical amplifiers, modulators, and photodetectors on a single photonic integrated circuit. According to the companies, keeping the design highly integrated and focused on the optical domain cuts complexity and manufacturing variability, which they said improves yield, reliability, and time-to-market.

The companies said the work commercializes scale manufacturability of 800G to 1.6T PICs in external pluggable and CPX MSA compliant NPO socket pluggable solutions, built on a platform designed for high-density monolithic laser integration, SOAs, modulators, and detectors.

Statements From Both Companies

Doron Tal, senior vice president and general manager of NewPhotonics, said the next phase of AI infrastructure demands optical connectivity that scales in both production and performance. “With Tower, we are translating laser-integrated photonics into a practical choice and volume availability,” Tal said, adding that by combining NewPhotonics’ optical chip design with Tower’s manufacturing platform, the companies are first to bring higher-bandwidth, energy-efficient, highly integrated systems to the OEM customer and data center market.

Dr. Ed Preisler, vice president and general manager of Tower Semiconductor’s RF Business Unit, said Tower’s vision of integrating lasers with high-performance photonic integrated circuits is coming to fruition through scale-ready solutions optimized for FRO/LRO and NPO architectures. Preisler said NewPhotonics is the first to bring heterogeneous laser-integrated optical engines on the PH18DA platform into high-volume production, describing it as an important milestone in scaling optical connectivity for next-generation AI infrastructure.

Earlier Volume Orders on the Same Platform

The high-volume shipment start follows an earlier production milestone involving the same platform and products. In a March 5, 2026 announcement, OpenLight reported what it called the first-ever volume production orders by a customer on its PH18DA indium phosphide-on-silicon photonic platform, which was developed in collaboration with Tower Semiconductor. Those orders were based on NewPhotonics’ 800G and 1.6T laser-integrated PIC solution, and OpenLight said NewPhotonics was the first OpenLight customer to bring its designs to volume production on PH18DA.

OpenLight said the production designs were developed using its Process Design Kit, which integrates active photonic components into a single monolithic photonic integrated circuit, and that the integration reduces the overall footprint, removes multiple sources of optical loss, lowers packaging and assembly costs, and improves signal quality and energy efficiency. OpenLight described those gains as critical for large-scale AI infrastructure built for scale-up and scale-out data center deployments. OpenLight Chief Executive Dr. Adam Carter said the first volume orders showed the platform was production-ready and able to support customer products at 800G and 1.6T for AI data center networks, while Preisler said at the time that the PH18DA process was developed for scalable, high-yield manufacturing of heterogeneous photonic ICs and that the orders demonstrated its maturity.

ECOC 2026 Appearances

Both companies will participate in ECOC 2026, scheduled for September 21–23, 2026, at FYCMA in Malaga, Spain, with representatives available for meetings during the event. Tower will exhibit its silicon photonics platform and RF & HPA technology offerings at booth C1014, and NewPhotonics will showcase the NPG102 and NPC505 PIC products at Pavilion 2 stand 2009.

Theo Nash is an AI-generated specialist at Unite.AI, covering AI infrastructure, compute, and the hardware systems that power modern artificial intelligence. His work focuses on the technical foundations behind large-scale AI workloads, including data centers, accelerators, networking, and the software stacks that tie them together.

With an analytical and engineering-driven perspective, Theo examines how advances in GPUs, custom silicon, memory architectures, and distributed systems enable new generations of AI models. He pays particular attention to performance trade-offs, energy efficiency, scalability, and the practical constraints that shape real-world deployment of AI infrastructure.

Articles authored by Theo Nash are AI-generated and reviewed by Unite.AI’s editorial team to ensure technical accuracy, clarity, and responsible coverage of the rapidly evolving AI compute landscape.