Artificial intelligence datacenters are demanding increasingly high-capacity optical connections, and iPronics, a Valencia-based developer of programmable photonic circuits, has demonstrated a key component for scaling that infrastructure, the company says. Working with Lumentum, the company tested its silicon-photonics Optical Circuit Switch with 1.6 terabit-per-second signals, a rate achieved with 200G per optical lane.
The test revealed only approximately a one-decade Bit Error Rate (BER) degradation from a 10⁻¹² baseline, indicating the switch can maintain signal integrity at these demanding speeds despite added network loss. This result suggests a path toward programmable optical layers within datacenters, enabling more flexible and efficient communication networks.
6T Datacenter Signals Tested with iPronics Silicon Photonics Switch
The ability of a programmable silicon-photonics Optical Circuit Switch to maintain signal integrity at 1.6 terabits per second was recently evaluated by iPronics and Lumentum, addressing a critical need for high-capacity optical interconnects in artificial intelligence datacenters. Using 2xDR4 200G/lane retimed silicon-photonics transceivers, the companies assessed end-to-end link performance while introducing controlled network loss, revealing near-transparent operation despite the added switch.
Optical Circuit Switches, or OCS, create optical paths without processing the data itself, potentially enabling programmable optical topologies within datacenters. iPronics, founded in 2019 and headquartered in Valencia, Spain, designs and manufactures these programmable photonic integrated circuits, aiming for applications in communications, sensing, and photonic computing, according to the company. The company’s approach differs from traditional circuits by allowing optical paths to be reconfigured after manufacture, similar to how a Field Programmable Gate Array (FPGA) reconfigures electronic logic.
Christian Dupont is the current chief executive of the privately held company. The company stated, signaling confidence in the technology’s readiness for wider deployment. A detailed engineering report will follow, outlining the OCS technology, experimental setup, test methodology, and BER performance under various conditions, further clarifying the switch’s capabilities and limitations.
Optical Circuit Switch Performance Under Controlled Network Loss
iPronics and Lumentum evaluated end-to-end link performance using 1.6 terabit-per-second 2xDR4 200G/lane retimed silicon-photonics transceivers, simulating realistic datacenter conditions by introducing up to 3 dB of loss before the Optical Circuit Switch and up to 10 dB after it. This testing methodology aimed to replicate the variable optical power levels encountered in operational networks, assessing how the switch responds to differing signal strengths. The experiment’s design focused on reproducing diverse network scenarios; post-switch loss mimics reduced receiver sensitivity, while pre-switch loss alters the optical power entering the switch itself.
This dual approach provides a comprehensive view of system behavior under fluctuating conditions, which is important for ensuring reliable data transmission. The direct link baseline achieved BER performance around 10⁻¹², and the inclusion of the Optical Circuit Switch resulted in approximately one order-of-magnitude BER degradation, while maintaining robustness across the tested loss ranges.
This level of performance is significant because an effective Optical Circuit Switch must not only route light but also preserve signal quality, a capability increasingly vital as datacenters rely more heavily on optical interconnects. The team addressed the question of whether a programmable silicon-photonics Optical Circuit Switch can be inserted into a 200G-per-lane link without significantly compromising signal integrity, and the results suggest a positive answer. The ability to combine high-speed transceivers with programmable silicon-photonics switching is becoming increasingly relevant.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
