Those billions in AI datacenter infrastructure will eventually be repurposed to run smart models like Mythos, not ChatGPT or even Opus/Sonnet. That future "rev

Those billions in AI datacenter infrastructure will eventually be repurposed to run smart models like Mythos, not ChatGPT or even Opus/Sonnet. That future "revenue model" is quite robust to any foreseeable competition from on-prem FLOSS inference. It's a natural fit to the actual capabilities of large datacenter-scale compute.