Elon Musk’s Secret AI Empire: The End of Everything We Know

Elon Musk's companies—Tesla, SpaceX, xAI, and Optimus—form a single vertically integrated physical AI system rather than separate ventures, the hosts argue. This setup bypasses competitors' fights over scarce chips and grid power by creating its own infrastructure, from data collection to compute and manufacturing.

Vision-Based AI Agents

The core innovation is Digital Optimus (under the "Macrohard" project), an AI agent that interacts with software visually, like a human. Unlike API-dependent systems from Microsoft or Google, which require custom bridges that break on updates, it processes recent screen video via cameras, using a digital mouse and keyboard. This achieves universality: it works on any interface, past or future, without integrations.

Running on Tesla's $650 AI4 chips, it handles reflexes locally (e.g., clicking buttons) and offloads reasoning to xAI's Grok cloud. Parked Teslas become distributed compute nodes, tapping idle hardware and 7 gigawatts at superchargers—consumer-funded, unlike centralized data centers.

Tesla's edge stems from training on real-world chaos: its fleet logs 8.4 billion miles of video data (1 billion in early 2026 alone), capturing edge cases like unicycles or snowstorms. Pure vision infers depth like human eyes, avoiding lidar's costs and geofencing limits used by Waymo.