Risks, Rewards & Building the Unicorn Chip Company Taking on Nvidia | Inside Groq with Jonathan Ross

[@StanHorrell-t5t]

What can one say? Excellent questions and even more impressive answers. I love the empathy/compassion piece on engineers.

[@fibersden638]

Why isn't this getting a million views? Easily one of the most iconic interviews of our time

[@LAinLA86]

Extremely intelligent guy

[@BTCKYLE]

You are just REALLY early... and honestly, smart as hell. Groq & Jonathan will go down in history making inference cost nearly nothing - which will impact every single aspect of our lives beyond what we can even imagine.

[@pandora244]

Completely agree. One of the best interviews I have heard this year!

[@sek_0000]

Thanks Eric and bigger thanks Jonathan for a very insightful and valuable talk.

[@mva6044]

Fascinating discussion. Just as a "know your audience" disclaimer: started to work on AI in college in early 2000s as part of a Howard Hughes fellowship in computational neuroscience. Really gratified to see where things stand right now. Now in academics, a run-of-the mill state medical school, ranked in the mid 20s. Advising 2 early biomedical diagnostic startups (series A and series B, respectively), and hold a very tiny private investment in GroQ (unfortunately I also held a slightly larger investment in MythicAI, another chipmaker that flamed out).

[@MeenakshiBabu]

Brilliant and insightful! thanks!

[@billvallier977]

Great interview!!!

[@defaultHandle1110]

Why are there so many Grok Groqs which are related to LLMs ?

[@chuckporritt6958]

I am a retiree just trying to piece AI all together in my own mind. So I watch videos like this to add puzzle pieces to my understanding. Somehow 'What is a TCU?' was not explained in this conversation (as opposed to the GPUs and CPUs that I know a little about). Is it what I have heard referred to as a 'tensor' unit? And isn't GROQ connected somehow with Elon Musk and AI-X? You have geniuses and you have 'bottom feeders' in your audience. The geniuses may have found this useful, but this bottom feeder is still hungry.

[@robertboroughs7824]

Transformer Compute Unit. A GPU-like processor , specifically created to accelerate transformer neural network processing The GPU was “repurposed” from its graphics origins. The TCU was designed for purpose. I am riding the learning wave as you are. Eventually when you hear smart people taking, you start connecting the dots. (Even though you remain amazed at the depth of their knowledge)

[@nickkowtko-h1j]

💜💜💜💜