Anthropic's Claude Opus 5, previously pulled, is back. This episode covers its return alongside Sonnet 5 and Claude Science, plus Anthropic's updated safety framework that preceded the relaunch. One host, Josh Kale, works with Anthropic as a contractor, which makes his commentary on the safety updates and model capabilities worth scrutinizing directly rather than taking at face value.
The episode spans nine distinct topics in under 40 minutes: Meta's brain-to-text research, Meta's compute infrastructure shift, OpenAI's latest strategic moves, Etched's inference chip announcement, a memory market disruption, and two separate robotics segments distinguishing home robots from industrial applications. Etched and the memory market coverage alone justify a listen if you track AI hardware.
The robotics split is the structural choice worth noting. The hosts break humanoid robots into two separate conversations: consumer home use versus industrial deployment. That framing reflects where the real investment and near-term revenue actually sit. This is episode 200 of the Limitless podcast, available on Spotify and Apple Podcasts.
[WATCH ON YOUTUBE →]