VAST's Simon Song: Why Large Language Models Hit a Wall, 3D AI is Just Starting

Overview

In this episode, we speak with Simon Song, the founder and CEO of VAST, a leading 3D large model company. Simon, a young entrepreneur, shares insights from VAST's rapid growth, including three funding rounds totaling tens of millions of dollars, and the story behind their latest 3D generative model, Tripo 3.0. This marks his first podcast appearance, where he unpacks his strategic vision for AI + 3D.

Guest

Simon Song (宋亚宸) is the founder and CEO of VAST, a company specializing in 3D large models.

What You'll Learn

• Why Simon believes large language models have 'hit a wall,' creating space for application and agent development, while 3D large models are just getting started.
• VAST's strategy of simultaneously developing foundational 3D models and building applications like Tripo Studio.
• Simon's 'decompression' theory of technology, viewing human media history (text -> image -> video -> 3D) as a progressive unlocking of the original 3D world.
• How human value might be redefined in a future where robots handle most physical tasks, shifting focus to creativity and experience.

Topics Covered

• The nascent stage and vast potential of 3D large models compared to mature language models.
• The strategic imperative for startups in the 3D AI space to develop both models and applications.
• A philosophical perspective on technology's evolution as 'decompression' from a 3D source world.
• The future of human creativity and experience economy in an AI-driven world.
• Simon's entrepreneurial journey, from childhood RPG creations to building a 'limitless world' with AI.

Why This Matters

This episode offers a critical perspective on the next frontier of AI beyond current LLM trends, providing valuable insights for investors, researchers, and entrepreneurs interested in the transformative potential of 3D AI and its market implications in China and globally.