Elon Musk has claimed that his artificial intelligence venture, xAI, will have more AI computing power than everyone else in the world combined within five years. Musk made the statement on X while responding to a post praising xAI’s efficiency and scale, arguing that its ability to convert energy and hardware into intelligence gives it a decisive advantage.
The discussion referenced analysis from SemiAnalysis, which highlighted the “Macrohard” branding painted on the roof of xAI’s Colossus 2 data center. The name is a tongue in cheek jab that signals Musk’s intent to directly challenge Microsoft and its dominance in AI software and cloud infrastructure.
According to SemiAnalysis, Colossus 2 is progressing toward more than 400 megawatts of computing capacity at its Tennessee site. Musk has reportedly ordered a complete power plant from overseas as part of a longer-term plan to reach 2 gigawatts of compute at a single location. At the same time, xAI is said to be raising up to $20 billion to purchase additional GPUs from Nvidia, further accelerating its expansion.
xAI’s pace has already drawn industry attention. Its first supercluster, built around 100,000 H200 Blackwell GPUs, was reportedly deployed in just 19 days, a process that Jensen Huang has said normally takes years. Musk has stated that xAI ultimately targets the equivalent of 50 million H100 class GPUs within five years, with roughly 230,000 GPUs already active and training Grok. Reaching that scale would require spending well over a billion dollars per month, something Musk appears prepared to do.
Even so, the claim remains controversial. Major players such as Amazon, Google, OpenAI, Oracle, and Microsoft are all investing heavily in AI infrastructure. OpenAI alone is developing a Texas data center complex expected to reach gigawatt scale by mid 2026, while China is reportedly considering allocating around $70 billion to domestic chip fabrication.
It is plausible that xAI could become the single largest AI compute operator in the world if this trajectory continues. Surpassing the combined computing power of the entire global AI industry, however, would require an unprecedented concentration of capital, energy, and industrial capacity, even by Musk’s standards.
