0:32
Eric Siu breaks down xAI's Grok-3 launch and Elon Musk's 200,000-GPU cluster built for it. He compares the model to OpenAI's O-1 and explains how massive funding, talent, and data center infrastructure enabled rapid scaling from 100,000 GPUs.
Eric Siu breaks down xAI's Grok-3 launch and Elon Musk's 200,000-GPU cluster built for it. He compares the model to OpenAI's O-1 and explains how massive funding, talent, and data center infrastructure enabled rapid scaling from 100,000 GPUs.
Did you see xAI launch Grok-3? Elon built this 200,000 mega cluster of these GPUs, which is insane. It's never been done before. I think it was 100,000 first and then he doubled it, right? And they didn't think he would do it. He did it in record time, right? And Grok-3 is still kind of behind OpenAI's like O-1 model, for example. But it's not even about that. It's how quickly they've caught up. Because a year ago, nobody talked about Grok, right? You look at— he sourced top talent. He used his pile of money as fuel, built a data center, ran it on natural gas, and got the model out right in time. And he also added deep research to it as well. And so focus and speed and you will get there.
