The growing appetite for computing power to support artificial intelligence (AI) workloads is driving the server market to new heights. Industry giants like Dell, Supermicro, and Hewlett Packard Enterprise (HPE) are capitalizing on the surge in demand for high-performance servers, with recent years marking significant growth in this sector.
A report from Bloomberg reveals that HPE has secured a contract exceeding $1 billion to provide Elon Musk’s X social network with servers specifically designed for AI-related tasks.
Though the exact purpose of these servers remains unspecified, their integration into X strongly hints at supporting Grok, the network’s AI chatbot. In late 2024, X announced the free rollout of Grok-2 for all users, suggesting that this expanded access requires additional server infrastructure.
The agreement between HPE and X was finalized toward the end of 2024, according to individuals with knowledge of the matter.
Table of Contents
Expanding AI Infrastructure
Elon Musk’s ventures, including xAI and Tesla, are significant purchasers of AI hardware. Notably, the Colossus supercomputer, developed by xAI in Memphis, has been described by Musk as the “most powerful AI training system in the world.” This system, assembled in just 122 days, incorporates 100,000 Nvidia H100 GPUs, with plans for even larger expansions in the pipeline.
Supermicro and Dell originally supplied servers for Colossus. However, despite bidding for this new deal, Bloomberg notes that HPE ultimately emerged as the supplier of choice. The company’s advanced liquid-cooling technology likely played a key role in its selection, according to Bloomberg Intelligence analyst Woo Jin Ho.
Supermicro, despite challenges like auditor resignations, remains a prominent player in AI hardware. Its recent establishment of operations in Memphis is aligned with xAI’s goal of building a facility equipped with 1 million GPUs.
