BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.denog.de//denog18//talk//73Q8NL
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-denog18-73Q8NL@cfp.denog.de
DTSTART;TZID=CET:20261116T114000
DTEND;TZID=CET:20261116T121000
DESCRIPTION:Inference is quietly overtaking training as the dominant AI wor
 kload\, and unlike training — concentrated in a handful of hyperscale ha
 lls — inference wants to sit close to users\, pulling AI capacity into d
 istributed sites across Europe. \nThe industry has spent the last two year
 s re-engineering Ethernet (via the Ultra Ethernet Consortium and platforms
  like NVIDIA Spectrum-X) to match InfiniBand's performance for exactly thi
 s kind of workload\, and now even ships Ethernet fabrics purpose-built to 
 stitch multiple physical data centers into one logical AI factory. \nThat'
 s a solved problem when one company owns every site. This talk asks what h
 appens when it isn't — when distributed inference has to span operators\
 , jurisdictions\, and ownership boundaries — and what a federated\, poli
 cy-aware Ethernet fabric for that world would need to look like.
DTSTAMP:20261111T000024Z
LOCATION:Saal B
SUMMARY:Ethernet Won Inside the AI Data Center. Can It Cross an Operator Bo
 undary? - Markus Hendrich
URL:https://cfp.denog.de/denog18/talk/73Q8NL/
END:VEVENT
END:VCALENDAR