NVIDIA, Asana, and New Models

NVIDIA agreeing to acquire Hugging Face changes more than who owns the biggest model repository in open-source AI. It may also make the open ecosystem harder to push around.

In Episode 24 of System Prompt, Peter and Val break down NVIDIA’s $12.9 billion Hugging Face deal, Asana’s claim that Codex compressed years of engineering work into weeks, and a wave of open models getting harder to dismiss.

The conversation moves past the headlines into the systems underneath them: why NVIDIA benefits from open models thriving, why AI productivity claims need methodology and labor context, and why better local models are changing how developers think about cost, privacy, and infrastructure.

Peter also walks through a one-shot AI Signal dashboard built with GLM 5.3 Flash and explains why improving open models are forcing him to rethink how much scaffolding smaller models need.

WHAT WE DISCUSS

•