Microsoft developed a MicroLED-based datacenter interconnect using imaging fiber to provide thousands of parallel optical channels. Lab tests show ~50% lower energy use than laser fibers, longer lifespan, lower cost, and compatibility with existing transceiver equipment for deployment.
Microsoft announced a MicroLED-based datacenter cabling system developed at its Cambridge Research Lab. The system uses inexpensive MicroLEDs and imaging fiber to replace laser-fed optical cables inside datacenters.
Main feature and impact
The core change replaces laser-driven single-channel fibers with MicroLED transmitters and imaging fiber. Thousands of parallel optical channels travel in one imaging fiber core bundle. Lab tests show roughly 50% lower energy use than current laser optical cables. The system promises longer component lifespan and lower manufacturing cost versus existing transceiver and laser assemblies.Practical implications
Deployment targets server-to-server and GPU interconnects inside racks and aisles. MicroLED links can span tens of meters, exceeding copper reach for high-bandwidth connections. Compatible transceivers were prototyped with MediaTek to fit current datacenter gear. Operators can reduce power draw, heat load, and replacement cycles while keeping standard datacenter installation procedures.“This breakthrough has the potential to change nearly every aspect of computing infrastructure … starting with high-bandwidth optical cables.”Microsoft positions MicroLED alongside Hollow Core Fiber for complementary roles. MicroLED handles dense, short-range intra-datacenter links while Hollow Core serves longer, lower-latency spans. Both technologies are compatible with existing equipment and aim to simplify installation and scaling. Microsoft expects commercialization with partners by late 2027. Next steps include partner manufacturing scale-up and field trials across Azure regions. Engineers must validate reliability under datacenter thermal and dust conditions. Successful deployment will reduce energy and facility costs, and enable denser GPU clusters for AI workloads.
Key points from the article:
Related Coverage:
- Sign with a personal touch with eSignature for Microsoft 365
- Microsoft Agent Framework, Microsoft Foundry, MCP, Aspire를 활용한 실전 예제 만들기
- Microsoft Agent Framework, Microsoft Foundry, MCP, Aspire を使った実践的な AI アプリを構築するサンプルが登場
From the Source
