Llama 3
Themes - DeepSeek, Kimi And AI Efficiency Paradigm Shift (Pt.2)
Summary
* DeepSeek has multifold lower opex and cluster costs than Western rivals, but its total capex is likely a lot closer.
* Despite US chip restrictions, DeepSeek has sustained its AI advancements through innovative training strategies.
* DeepSeek is scaling up with potential outside funding and deeper collaboration with Huawei.
Training
One
Themes - DeepSeek, Kimi And AI Efficiency Paradigm Shift (Pt.1)
Summary
* DeepSeek is a true innovator, not a copycat. Its model architecture, algos, training and inference frameworks are all firsts in the industry.
* DeepSeek hasn't used smuggled H100s, just 2048 H800s, as per its disclosure.
* For the near term, it should continue to use NVDA cards, but over
Updates: Meta - Underappreciated GenAI Potential (Pt.2)
Summary
* META's Llama project requires monetization strategies as funding scales to $100bn+, with potential approaches including SaaS and enterprise solutions.
* META can leverage Llama for consumer applications and enterprise solutions, fostering a developer ecosystem while generating revenue through governance and alignment services and enhanced features.
* META's
Notes: Meta - Controlling The Value Chain And Open-Core
Summary
* Our latest research on Meta has been divided into two Updates and one Notes report (this one). Updates (Pt.2) will be published in a few days.
* Controlling LLM technology is crucial for META's adtech dominance, enhancing ROAS, user engagement, and future AI-driven products like chatbots and
Updates: Meta - Underappreciated GenAI Potential (Pt.1)
Summary
* In Part 1 we review the battle for GenAI supremacy and how Meta's Llama has emerged as a foundation model leader, despite the odds.
* We discuss the challenges Mistral, the previous open-source LLM leader, is facing when competing against Meta and closed source LLM leaders.
* In Part