SambaNova and the SN50 reminds me a lot about Cerebras. When we first went to Cerebras last year to deploy their systems we thought their wafer scale chip was the most overlooked asset in compute. Apparently so did @sama who went and brought up 750MW of it. 10 months later + one massive IPO it's becoming clearer to everyone. GPU's mean no fast inference. Not a little, none. Decode silicon is the only path right now that's commercially viable. As for the @SambaNovaAI we feel the same way about the SN50 as we did with Cerebras last year. Many people are underestimating it massively, leaving a huge opportunity for @general_compute to front run this amazing hardware platform and bring to market arguably the most profitable fast inference in the world right now. @RodrigoLiang their CEO is sitting on a gold mine. Both teams worked almost 10 years for this moment. I'd say they hit massive product market fit in Q127 and speed run to massive growth.