How partners tamed the token bill: putting AMX on Intel Xeon 6 into production
Date
Wednesday, December 2, 2026
Time
4:00 - 5:00 PM PST
Location
Fogo de Chão
Inside the Grand Canal Shoppes at The Venetian Resort, 2nd Floor
Room: Venezia 2
About the Roundtable
Inference tokens are now the fastest-growing NRE cost, often outstripping engineering headcount. During this interactive roundtable session, AWS and Intel Partners running in production today (Anyscale, Blunom, Riskfuel and OpenInfer) will talk about how they leverage the AMX matrix engine in every Intel Xeon 6 core to accelerate token throughput. The AWS-exclusive high-memory-bandwidth Xeon 6 processors in the 8i instance family complement GPU instances by carrying the many workloads that do not need a full GPU, or where a GPU running 24/7 does not make financial sense. Hear what our partners discovered is best routed to CPU, what stays on GPU, and how this changes the numbers.
Speakers

Registration is subject to approval.
Please complete the form below to request attendance.
Our team will review your submission and be in touch soon.





