Intel Begins Volume Shipments of Silicon Photonics Products
By The data center has become the hub of a ton of innovation lately. One example of this innovation is Intel’s work around silicon photonics.
Silicon photonics leverages integrated circuit and semiconductor laser technology for more efficient transfer of traffic between servers and switches in the data center. This laser-based technology, Intel said, can eliminate the networking bottlenecks created today by the use of copper wires.
Silicon photonics can carry more 100Gper over cables up to 2 kilometers long; meanwhile, copper cable tops out at around 25Gbps over a 10-
foot stretch, according to a Data Center Knowledge story.
Intel’s first two silicon photonics products – 10G optical transceivers for interconnecting data center switches – are shipping in volume now. That includes the 100G PSM4 QSFP28 Optical Transceiver, which has a reach of 2 kilometers on parallel single-mode fiber and is compliant with 100G PSM4 MSA, QSFP28, and CAUI-4 specifications. It also includes the 100G CWDM4 QSFP28 Optical Transceiver, which has the same reach on the same kind of fiber and the same standard compliance, but also features low power consumption.
The company claims it offers “the only fully integrated silicon photonics solution with Intel’s hybrid laser technology, featuring greater than 90 percent coupling efficiency.” The company reportedly expects to introduce 400G interconnects in a couple of years.
451 Research recently published a white paper titled “Intel’s Silicon Photonics Products Could Change the World of IT.” In it, Peter Christy of 451 Research writes that “the application of silicon wafer planar manufacturing technology to the volume manufacture of electro-optical transceivers – enables a significant cost reduction compared to current manufacturing techniques and promises to accelerate the rate of performance and price/performance improvements as well.”
He adds that the silicon photonics also hold the potential to decrease costs within large data centers by enabling rack-scale system and application architectures.
Edited by Alicia Young
