All-in on AI
Elon Musk’s recent X post unveils a leap forward in artificial intelligence infrastructure. xAI (Grok), his ambitious AI venture, already operates Colossus 1, a supercluster built with 230,000 GPUs, including 30,000 of NVIDIA’s ultra-powerful GB200s. But that’s just the beginning. Currently they are building up Colossus 2 with a jaw-dropping 550,000 GB200s and next-gen GB300s—numbers that push the boundaries of what’s been attempted in AI compute to date.
To put it in perspective: this is one of the largest AI training operations ever assembled, dwarfing the compute scale of most commercial and academic projects. The sheer capital investment—likely in the tens of billions—demonstrates Musk’s unshakable belief that whoever leads in AI, leads the future.
Musk also echoes NVIDIA CEO Jensen Huang’s claim: when it comes to AI model training speed, “@xAI is unmatched. It’s not even close.” Given the scale of hardware behind Grok’s development, that’s more than plausible.
This signals Musk’s commitment to pushing AI to its limits—faster, bigger, bolder than anyone else. xAI is no longer a startup—it’s an AI superpower in the making.
Really impressive imo and it proves again that Elon Musk loves to go all-in once he starts a project. Both a strength and high risk.
Have you used Grok/xAI yet? What do you think of their visual Avatars?
Source: X
