1.6T Silicon Photonics Transceivers Enter Trials at Major Cloud AI Data Centers, Accelerating Towards 2025 Mass Production

July 5, 2026

EE Times USA

Overview

1.6T optical transceivers based on silicon photonics technology have reportedly commenced trial deployments at AI data centers operated by leading cloud providers. These transceivers deliver the ultra-high bandwidth and low power consumption essential for next-generation AI workloads, achieving approximately 30% better power efficiency than previous 800G modules. This marks a critical step towards full-scale production in 2025, promising dramatic improvements in AI cluster performance and efficiency, while significantly reducing data center operational costs.

In Depth

Key Findings

State-of-the-art 1.6T optical transceivers, leveraging silicon photonics technology, have reportedly initiated trial deployments within AI data centers of several prominent cloud providers. These cutting-edge transceivers successfully combine the immense bandwidth demanded by next-generation AI workloads with the imperative of low power consumption for environmental sustainability, achieving an approximately 30% improvement in power efficiency compared to previous 800G generation optical modules.

Technical / Clinical Details

These 1.6T transceivers are built upon a silicon photonics platform manufactured using standard CMOS processes. This enables a high degree of integration between optical and electronic circuits, resulting in reduced footprint and optimized manufacturing costs.
The modulation scheme employed is PAM4 (Pulse Amplitude Modulation 4-level) technology, facilitating a total data transmission speed of 1.6 terabits per second (Tbps) through parallel processing across multiple lanes. This effectively mitigates bandwidth bottlenecks in inter-rack and GPU-to-GPU connections within data centers.
The reduction in power consumption is achieved through optimization of the digital signal processing (DSP) chip and enhanced efficiency of optical components within the silicon photonics platform, directly translating to lower operational costs for large-scale AI clusters.

Background & Context

The escalating complexity of AI models and the explosive growth in data volumes demand ultra-high-speed and high-capacity data transfer capabilities in data centers, particularly within AI/HPC clusters, far beyond what traditional network infrastructure can provide. While 800G optical modules are only just beginning to proliferate, demand for 1.6T and even 3.2T is already anticipated, with power efficiency and cost remaining primary concerns. Silicon photonics, due to its inherent scalability and cost advantages, has emerged as the most promising technology to address these challenges.

Strategic Significance & Outlook

These trial deployments represent a pivotal milestone for 1.6T silicon photonics transceivers as they move towards full-scale mass production in 2025. Real-world evaluation by major cloud providers will provide final validation of compatibility, reliability, and performance. The widespread adoption of this technology is expected to dramatically enhance the computational power of AI clusters and serve as a foundational technology supporting data center sustainability. This will, in turn, accelerate the further evolution of generative AI and large language models (LLMs), contributing significantly to the advancement of digital society.

Source: https://www.eetimes.com/silicon-photonics-1-6t-transceiver-trials-data-centers/

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.