🔎 Today’s Daily Sift: AI/Robotics AI is moving from intelligence to execution. On August 24, Nvidia put Groq 3 LPX into full production. The inference accelerator is designed specifically for AI agents that reason, use tools and execute long sequences of actions. Nvidia says it reached roughly 3,400 output tokens per second on a long context benchmark, about four times the responsiveness of the nearest alternative. That is more than speed. Faster inference changes the economics of autonomous work. AWS moved on the same problem from another direction, launching Agentic Resource Discovery and an Agent Registry for finding, approving and governing AI agents and tools across enterprises. AI workers are beginning to get infrastructure. Then robotics delivered both sides of the hype cycle. Xpeng raised more than $900 million for its robotics business, a record private round for embodied AI in China. Meanwhile, Unitree shares have fallen roughly 45 percent after their explosiv...