Interconnects

Interconnects is a long-form newsletter written by Nathan Lambert, a researcher with experience inside major AI labs, covering how frontier models are trained (including RLHF and reasoning methods) and the strategic and political decisions of AI labs. It's aimed at readers who already grasp AI fundamentals — engineers, researchers, and investors — and want technical depth rather than surface-level news. Its main differentiator is the author's insider vantage point, delivering grounded analysis from someone whose day job is actually training AI models, read by over 82,000 subscribers.