scripod.com

Every Muse User Gets 2 CPUs and 8GB in Meta's Cloud. Your Phone Already Has That

Semi Doped

2 DAYS AGO
Semi Doped

Semi Doped

2 DAYS AGO

Shownote

The push to run AI agents on edge devices will not be driven by model labs or hardware vendors, but by service providers like Meta whose ad-based business models create a powerful financial incentive to offload massive cloud compute costs onto the user's o...

Highlights

Semi Doped examines why service providers, rather than chipmakers or model labs, may drive AI inference onto phones and other edge devices. The hosts connect Meta’s economics, Qualcomm’s hardware and software strategy, and the rise of personal agents that route work between local devices and the cloud.
00:05
App companies decide where inference runs
05:21
Agents running anywhere
07:37
The AI PC push became a race to nowhere
10:24
The phone is already an AI portal
16:58
Service providers have the incentive to move AI to the edge
20:38
Local context should not always go to the cloud
32:22
AI should provide the format you want
34:23
The edge becomes a context translation layer
36:36
Fast-thinking models belong at the edge
42:28
Mojo could help AI run anywhere
44:48
The user—not the device or cloud—is at the center of the AI ecosystem
48:59
Meta is in the lead.

Chapters

Who Makes Local AI Happen?
00:00
Making the Intro with AI
01:25
Qualcomm's Agent Vision
04:43
Why AIPCs Failed
06:54
The Phone as a Portal
08:45
Meta's Incentive for Edge AI
13:52
The End of Apps & UIs?
24:05
The Ideal Personal Agent
32:31
The Edge as a Context Layer
34:25
Qualcomm's Modular Acquisition
38:17
The Ecosystem of You
43:16
Google's Missed Opportunity
48:01

Transcript

Austin Lyons: Last week, Qualcomm said agents will run on your phone, your glasses, and your earbuds. But chip companies don't decide where inference runs, app companies do. And the biggest app companies of our time right now, OpenAI and Anthropic, well, t...