#490 – State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI

Nathan Lambert and Sebastian Raschka join Lex to survey the state of AI heading into 2026. They compare leading models for chat and coding, look at open versus closed source LLMs, and explain how pre-training, mid-training, and post-training work. The talk covers scaling laws, the AI race between the US and China, and new research like text diffusion, tool use, and continual learning. They also weigh timelines to AGI, whether AI will replace programmers, and the future of NVIDIA and major AI labs.