Topic

World action models and infrastructure

Latest viewpoints

World action models are computationally expensive, a barrier to deployment

Bill and Guanming from General Instinct introduce infrastructure for optimizing world action models (WAMs), which predict future frames and kinematics simultaneously, highlighting their computational expense (requiring expensive GPUs) as a key scalability barrier for real-world deployment.

Bill Jiao World action models and infrastructure Source at 83:52

Distillation and separate transformers speed up world action model inference

Guanming explains optimizations for WAMs, including distillation and separating the model into video and action transformers to avoid decoding future frames, significantly speeding up inference by maintaining world representation in hidden states and reducing flow matching steps.

Guanming Wang World action models and infrastructure Source at 85:31

Leaders

Jensen Huang

CEO, Nvidia

Last interview Sep 2, 2026

Howard Lutnick

Public business viewpoint profile

Last interview Sep 2, 2026

Sam Altman

CEO, OpenAI

Last interview Sep 2, 2026

Elon Musk

CEO, SpaceX / Tesla

Last interview Sep 1, 2026

Julien Bek

Partner, Sequoia Capital

Last interview Aug 24, 2026

Thibault Sottiaux

Engineer, OpenAI

Last interview Aug 24, 2026

Matthew Berman

Influencer, Forward Future

Last interview Aug 24, 2026

David Senra

Podcaster, Senra Show

Last interview Aug 23, 2026

LeAnn Foster

Founder, EUTHENICS

Last interview Aug 22, 2026

Eric Ho

CEO, Goodfire

Last interview Aug 21, 2026