Nvidia has agreed to buy Hugging Face for $12.9 billion, according to The Information.

That's the platform where most of the world's open-source AI models live — the "GitHub of AI," where developers go to find, share, and deploy models and datasets.

Nvidia already backs Hugging Face at a $4.5 billion valuation from 2023.

Earlier this year, it offered $500 million for a stake that would've valued the company at $7 billion — and Hugging Face turned it down.

Now the chip giant gets the whole thing.

Here's why this matters beyond the price tag:

1. The front door — Hugging Face is the default starting point for open AI development. Owning it gives Nvidia a grip on the mindshare of every AI developer on the planet.

2. A hedge on chips — Google, OpenAI, and Anthropic are all building their own silicon. If chip revenue slows, Hugging Face becomes a non-chip source of revenue.

3. The robotics angle — Hugging Face just unveiled Microduck, a $399 open-source robot. Nvidia's already deep in physical AI with Jetson and Isaac.

The catch: Hugging Face only generates about $150 million in annualized revenue. That makes this deal roughly 80 times forward revenue.

The bigger question is whether Hugging Face's neutral, open-source reputation survives being owned by the company with the most to gain from picking winners.

Microsoft made GitHub feel like an extension of its own strategy. Developers noticed.

The same test now applies to Nvidia — and the entire open-source community is watching.

Mads Lorentzen lost his job as a geophysicist.

So he built an AI framework to run his own job search — and it worked.

69 tailored applications. 20 first interviews. One signed contract.

Then he open-sourced the whole thing.

The clever part is the architecture: a drafter-reviewer pattern where two AI agents treat job applications as adversarial documents, not form-fill exercises.

One agent writes. The other tries to tear the application apart. The draft survives only if it holds up.

Think of it like a writer and an editor who hate each other — and that tension is exactly what makes the output better.

Lorentzen's framework also auto-writes cover letters for every role, tailored to each job description.

The tool is free, open-source, and built on Claude.

You can adapt it to your own search in an afternoon — or just steal the drafter-reviewer pattern for any high-stakes document you need to write.

The job market may be brutal. At least one person found a way to fight back with software.

OpenAI is cutting prices on GPT-5.6 Sol by more than 20% for the next three months — now $4 per million input tokens and $20 per million output.

That puts Sol cheaper than Anthropic's Opus 5, and nearly competitive with DeepSeek on intelligence-to-dollar.

The context: enterprises are questioning AI spending hard, with Gartner projecting inference costs to grow fivefold by 2028 as agents drive up token usage.

Data from payments platform Ramp shows most business spending on Anthropic models isn't on its flagship Fable 5 — only 6% of token spend. Opus 5 has overtaken it.

The signal is clear: the era of "frontier model at any cost" is over. Companies are getting disciplined about what intelligence is actually worth.

DeepSeek also shipped V4-Flash-Vision-Exp this week, matching Opus-level multimodal performance at Flash prices — another squeeze on the high end.

LINK: The Deep View's pricing analysis: https://www.thedeepview.com — "Why OpenAI is resetting frontier AI prices"