Live Technology Panel
summary
Companies describe concrete pivots: a Tomahawk 6C 100‑Tb device to collapse tiers, a Tomahawk Ultra Ethernet switch claiming ~250 ns latency, Jericho silicon with deep packet buffers and congestion control, NVLink fusion and an open SDK for NIC programmability, and early support for UEC/UEC‑like standards. Teams report distributed inference, reasoning models, and surprising shrinkage of some models (50k–200k GPU clusters evolve), experiment with packet trimming and congestion signaling, and recommend simulating scale to validate changes.
AI-written summary · about
chapters (5)
- 0:00AI industry pivots
- 3:14Evolving AI model trends
- 10:47Ecosystem and optionality
- 14:41Standards and innovation
- 26:13Advice for engineers
related talks
15mLive Q&A with Speakers – Joseph Provine
26mEvolution of Application Visibility for Telcos: From Programmable Switches to Cloud Native
24mSilicon innovation and the modern network
31mLive Q+A 1
25mSilicon and Networking
17mMeta’s Network Journey to Enable AI
26mApplying New High-Throughput Network Solutions in IEC (Integrated Edge Cloud)
19mDisaggregating Stateful Network Functions