Macvoices Audio

The panel explores how M-series Macs—with huge unified memory and efficient silicon—are gaining traction for AI inference and on-device privacy, citing MacStadium use cases and enterprise angles like Copilot adoption. Chuck Joiner, David Ginsburg, Marty Jencius, Brian Flanigan-Arthurs, Eric Bolden, Guy Serle, Web Bixby, Jeff Gamet, Jim Rea, and Mark Fuccio contrast training vs. inference, discuss small language models, and corporate data policies. The session wraps up with the alleged FCC leak of iPhone 16e schematics, and Xiaomi’s unabashed Apple cloning—plus a quick note on viral AI fakes.  This edition of MacVoices is brought to you by the MacVoices Dispatch, our weekly newsletter that keeps you up-to-date on any and all MacVoices-related information. Subscribe today and don't miss a thing. Show Notes: Chapters: [0:30] AI workloads on Macs and unified memory advantages [1:36] Training vs. inference explained; why memory matters [3:49] M3/M4 bandwidth, neural accelerators, and privacy [5:35] Avoiding the “