LLMs may have kicked off this AI boom, but the ceiling is closer than the hype suggests. As models run out of text data to train on, the companies and investors paying attention are already moving on. The next wave isn't better chatbots; it's machines that can understand the physical world. Luma AI, the Bay Area lab that raised over $1.4 billion from a16z, Nvidia, and Amazon, is betting on exactly that.

On episode of TechCrunch's Equity podcast, we’re bringing you a conversation Rebecca Bellan sat down with Amit Jain, co-founder and CEO of Luma AI, at Web Summit Qatar. Together, the pair dug into where the next trillion-dollar AI opportunity actually gets built, and whether the companies chasing it even know what they're building yet.

Listen to the full episode to hear about:

Why video, audio, and images are the real frontier for AI training data, not text

What an "intelligent world model" actually is, and why Jain thinks most companies building them are getting it completely wrong

The case for why AI won't kill creative jobs, and why Jain thinks studio heads are the real problem

How the path from video generation to robotics to AGI is simpler than anyone's making it sound

Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.

Chapters:

00:00 Intro

01:13 Why LLMs are hitting a ceiling

02:43 The data problem & what comes after LLMs

04:30 What actually makes a world model a world model

06:05 Why 3D data is a dead end

07:39 What Luma is building next

09:08 How much humans stay in the loop

10:00 Near-term use cases for agentic video

11:22 Will AI kill jobs in film & production?

13:30 Why the entertainment industry is already dying

15:27 Why we actually need more content, not less

17:46 Luma's roadmap: generation, understanding, and robotics

19:54 Outro

Learn more about your ad choices. Visit megaphone.fm/adchoices

Luma AI's Amit Jain on why most world model companies are getting it completely wrong