Ennostar Showcases Micro LED Optical Communication Technology with AUO and Tyntek

Why It Matters

The technology promises dramatically lower power and cooling needs for high‑density AI data centers, accelerating the shift to energy‑efficient, short‑reach optical interconnects. Its ecosystem partnership speeds commercialization of next‑generation optical modules.

Key Takeaways

• Ennostar demonstrated 1.25 Gbps per‑channel Micro LED link.
• Micro LED operates up to 125 °C, 30,000 h lifetime.
• Solution targets sub‑10 m interconnects for CPO and AOC.
• VCSEL and CW‑DFB lasers expand reach to 100 m and 2 km.
• Ecosystem partnership with AUO and Tyntek accelerates commercialization.

Pulse Analysis

AI‑driven workloads are pushing data‑center bandwidth requirements beyond the limits of traditional copper and legacy optical links. Ennostar’s Micro LED approach leverages GaN’s high‑speed modulation and exceptional thermal tolerance, delivering sub‑10‑meter interconnects that consume less than 1 pJ per bit. By integrating the transmitter with Tyntek’s micro photodetector inside AUO’s co‑packaged optics, the solution reduces module size, power draw, and cooling overhead—critical factors for densely packed AI racks.

Beyond Micro LED, Ennostar’s portfolio includes VCSELs for 100‑meter short‑reach links and InP‑based CW‑DFB lasers that can span up to 2 km. This tiered architecture lets data‑center operators select the optimal source for each tier of the network, from intra‑server connections to rack‑to‑rack and even campus‑scale links. The demonstrated 1.25 Gbps per channel performance, combined with a 30,000‑hour lifetime and operation at 125 °C, positions these sources as viable replacements for conventional edge lasers that demand higher power and more aggressive cooling.

The collaboration with AUO and Tyntek underscores a broader industry trend toward integrated optics ecosystems. By co‑designing the optical source, detector, and packaging, Ennostar shortens the path from prototype to volume production, addressing the urgent need for scalable, energy‑efficient interconnects in AI infrastructure. As AI models grow and data‑center densities increase, such innovations are likely to become foundational components of next‑generation high‑performance computing networks.