OpenAI sidesteps Nvidia with unusually fast coding model on plate-sized chips

the era of running ‘your own’ model on x86 chips will be very short

the future is all about premium subscribers accessing proprietary hardware from vendor-locked endpoints