Enter for the full results page · Esc to close
Module price is the number everyone negotiates and the smallest part of what a port actually costs. A 1.6T module burning 28 W draws roughly its own purchase price in electricity over three years once cooling is counted. This puts all four cost lines side by side, and compares an OEM-priced build against a compliant third-party one.
The four lines that make up a port. Hardware is the obvious one. Energy is the module wattage running continuously for the ownership period. Cooling and power distribution add a further multiple of the energy line, which is what PUE measures. And a share of the fleet fails each year and has to be replaced, which costs the module plus the hands to swap it. Leave any of the four out and the comparison is meaningless.
Why power matters more every generation. A 100G SR4 draws about 3.5 W. An 800G module draws about 16 W, and 1.6T is landing around 28 W. Per bit that is a large improvement, but per port it is an eight-fold increase, and it lands on a rack power budget that has not grown eight-fold. At 64 ports of 800G in one switch, you are dissipating roughly a kilowatt in transceivers alone before the switch ASIC is counted.
Where third-party optics actually save you money. The saving is nearly all in the hardware line, because a compliant third-party module draws the same power as the OEM part it replaces \u2014 same optics, same DSP, same MSA. What varies between suppliers is failure rate, and that is worth asking about: a 0.5%/year fleet and a 3%/year fleet look identical on the quotation and very different three years in.
The optics and electronics are built to the same MSA and IEEE specifications, so the link behaves the same. What differs is the EEPROM coding that tells the switch what it is looking at, and the depth of testing before shipment. Ask for the traffic-test report and the measured failure rate \u2014 a supplier who cannot produce either is selling you risk, not savings.
In practice, no \u2014 major vendors will support the switch; what they may decline is to debug a fault they attribute to the optic. The practical mitigation is to keep a couple of OEM modules on the shelf so you can swap and demonstrate the fault is not the transceiver. That is cheap insurance next to the difference this calculator shows.
Ask the supplier for their measured figure rather than guessing. Well-screened optics run well under 1% per year in a temperature-controlled data centre; the same part in a hot, dusty, uncooled cabinet will do worse. If nobody will give you a number, assume 2% and see whether the business case still stands.
No \u2014 this is transceiver TCO only, so you can compare like with like across suppliers. For the cabling side of the same build, use the MPO cabling planner, and for the rack heat and power side use the power and thermal calculator.
Per-port wattage rolled up to switch, rack and cooling load — and whether the rack can take it
Open tool →Turn a switch port count and a breakout ratio into the exact optics and cables to order
Open tool →Total channel loss, power budget and pass/fail margin for single-mode and multimode links
Open tool →Transceivers, MPO assemblies, WDM and patch cords — built and tested to the figures you just calculated.