Elon Musk made a bold statement on the social network X, claiming that his company xAI will have more computational power within less than five years than all other data centers in the world combined. The billionaire made this statement in response to discussions about xAI’s strategy. The seriousness of xAI’s intentions is underscored by the Colossus 2 data center project in Tennessee, whose roof features the ironic inscription “Macrohard” – a challenge to Microsoft’s dominance.
To realize these plans, xAI is raising up to $20 billion for the acquisition of additional Nvidia graphics processors. The company has already demonstrated its seriousness: it deployed the first supercluster of 100,000 H200 chips in just 19 days – a process that, according to Nvidia chief Jensen Huang, usually takes four years. Musk’s goal is to achieve 2 gigawatts of power at a single site, for which he has even ordered the construction of a power plant.
In the long term, xAI plans to build up computational resources equivalent to 50 million H100-class GPUs. Currently, 230,000 accelerators are involved in training the Grok model. Experts from SemiAnalysis confirm the real progress of the project but doubt that Musk will be able to surpass the combined power of data centers from Google, Amazon, Microsoft, and OpenAI, as well as China’s state investments, which are putting tens of billions of dollars into developing AI infrastructure.
Recently, major tech companies have accelerated investments in AI. According to industry analysts, these companies are exploring new frontiers in AI-driven technologies and increasing cloud infrastructure. Google’s parent company Alphabet has reportedly doubled its AI infrastructure investments in the last year alone, and China’s government has earmarked substantial funds to boost its national AI capabilities. Public announcements indicate that X’s ambitious plans, if successful, could position the company as a formidable competitor within a rapidly advancing AI landscape.
