Peak current in an ESD protection context flows from
the "threat model" (model backing voltage, series RL)
and your goal is to return all of the energy to the source,
losslessly as possible (because losses are the face-punch
you end up having to take).
All ESD sources are transient, all have a fixed charge
(expressed by voltage, the source C is implicit by the
threat-model callout) and so the deliverable energy is
known.
I do not know what Ethernet standards demand, and
your tasking may be different anyhow - maybe there's
"Big Marketing B@lls" in offering a more robust than
minimum ruggedness, provided you don't ruin the
gigabit bandwidth story.
I recommend finding an ESD threat-model document,
setting up some SPICE simulations, and look at pulse
energy deposited to the devices / networks you plan
to investigate. Take device pwer and terminal currents
to Joule energy and temperature rise, interconnect
current density and terminal voltages on single devices
around the protection current loop and see which of
those may be exceeding device level abs max.
You can design ESD protection in detail. Most people
prefer to just hang on whatever somebody else has
blessed, if that road's been plowed ahead of time. If
you're cutting trail, best learn how to do it all yourself.