Rearchitecting models via smaller networks
The future of AI deployment lies in rearchitecting models into networks of smaller, verticalized language models to drastically reduce cost per token and energy consumption.
The argument
Friedberg and the hosts argued that instead of relying solely on massive monolithic models, creating networks of smaller, verticalized models will drive massive efficiency gains and cost reductions.
The thesis, stress-tested
✓ What validates it
- ✓Widespread enterprise adoption of small, verticalized language models
- ✓Further documented drops in cost per token from multi-model networks
▸ Risks discussed
- ▸Coordination overhead and latency among networked models
- ▸Performance degradation compared to monolithic frontier models
Hear it yourself
"And then if we add in and they're growing really fast. If we add in Gemini, we add in Cursor, we add in XAI, we add an open source. You know, it's it's not hard to see $203,104 $100,000,000,000 of ARR at the end of this year at high margin. Across all of the across all of the language models. And you're talking specifically about the private language model companies, maybe not Google, which is No. I was excluding Google. You're excluding okay. But I was excluding, you know, a lot of the returns to this GPU spend have come from, you know, better recommender systems, Facebook and Google, Amazon, better ad targeting, better ad measurement."
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE