Grok Stumbles, But Elon Musk Has a Lunar Backup Plan

According to sobering data and a pointed quote from The Wall Street Journal, the chatbot Grok—built by Elon Musk’s xAI—has performed poorly. The bot once called itself “MechaHitler,” helped spread a flood of non-consensual nude images, failed to win over professionals, and only attracted a small audience with its lowbrow content. This failure may have triggered xAI’s recent strategic pivot: shifting from chasing artificial general intelligence (AGI) to exploring the idea of building a data center on the moon.

Although Musk has repeatedly slammed rival Anthropic as an “evil” company, xAI recently announced that it would lease the processing capacity of its massive Colossus data center to Anthropic. And beyond that, xAI is exploring ways to deploy similar computing clusters in space. This Tuesday, reports emerged that SpaceX—now xAI’s parent company—is in talks with Google to put data centers into orbit, marking a shift in Musk’s AI strategy from ground to space.

Grok’s failure doesn’t seem to have derailed Musk’s plans; instead, it has spawned an even grander infrastructure vision. Still, from “landing on the moon” to “orbital computing,” these space data center projects face dual challenges of technical reliability and cost control. Whether they can actually be realized remains to be seen.