GRT’s optical module range extends through the 40G–800G high-speed classes, and this page covers the top two: 400G and 800G transceivers. These are spine and cluster optics — they interconnect top-of-rack and spine switches, and increasingly they wire GPU clusters whose east-west traffic dwarfs everything else in the building. At these rates the module decision is inseparable from the fabric decision: switch radix, breakout strategy and fiber discipline are all chosen together.
For most enterprise buyers the practical question is timing. A fabric bought today at 100G with QSFP28 will serve for years; a fabric designed around AI training or a multi-year spine refresh should be evaluated at 400G now, with 800G as the documented next step. GRT’s range covering both lets a design standardise on one optics supplier across the migration — 100G at the leaf today, 400G at the spine, 800G when the cluster demands it.
Planning questions we will ask
- Switch platforms and their cage types at each fabric layer
- Breakout plan: native rate or fan-out to lower-rate ports
- Fiber discipline: single-mode duplex or parallel multimode
- Power and thermal budget per rack
Is 800G real, or roadmap?
800G modules are part of the documented 40G–800G module range. For project commitments, confirm the exact variant and schedule with our team — high-speed optics are specified per fabric design.
Can 400G ports break out to 100G?
Breakout is a standard fabric technique at this tier. The viable fan-out depends on the switch and module pairing; describe your design and we will spec the optics.