According to reports, SpaceX is adjusting its AI data center construction method. The new management team plans to install more power and cooling backup systems and conduct more thorough tests before the facility is put into operation, while reducing reliance on temporary power supply and cooling equipment, which may slow down the expansion of the new data center park. The report said that this adjustment stemmed from reliability issues revealed by xAI's previous rapid construction model. Its Tennessee and Mississippi data centers had frequent breakdowns in the early days, causing AI research projects to be suspended and part of the training progress lost. After SpaceX acquired XAI this year, it reorganized its data center team at the end of June to replace some traditional data center managers with rocket and Starlink engineers. More than 300 SpaceX engineers have joined related projects in the past month. The report said that some of the existing facilities are also currently being redesigned. SpaceX is not expected to completely abandon XAI's rapid construction model, but will place more emphasis on redundancy and standardization. Previously, xAI launched the first data center equipped with 100,000 GPUs within 122 days, but its initial backup system was insufficient, and it relied on temporary facilities such as gas turbines for a long time. Last week, a large data center in Memphis also experienced a power outage due to construction affecting power lines, which affected some Grok models and external computing power customers. Musk later said the company is taking corrective measures to prevent similar incidents from happening again.

According to reports, SpaceX is adjusting its AI data center construction method. The new management team plans to install more power and cooling backup systems and conduct more thorough tests before the facility is put into operation, while reducing reliance on temporary power supply and cooling equipment, which may slow down the expansion of the new data center park. The report said that this adjustment stemmed from reliability issues revealed by xAI's previous rapid construction model. Its Tennessee and Mississippi data centers had frequent breakdowns in the early days, causing AI research projects to be suspended and part of the training progress lost. After SpaceX acquired XAI this year, it reorganized its data center team at the end of June to replace some traditional data center managers with rocket and Starlink engineers. More than 300 SpaceX engineers have joined related projects in the past month. The report said that some of the existing facilities are also currently being redesigned. SpaceX is not expected to completely abandon XAI's rapid construction model, but will place more emphasis on redundancy and standardization. Previously, xAI launched the first data center equipped with 100,000 GPUs within 122 days, but its initial backup system was insufficient, and it relied on temporary facilities such as gas turbines for a long time. Last week, a large data center in Memphis also experienced a power outage due to construction affecting power lines, which affected some Grok models and external computing power customers. Musk later said the company is taking corrective measures to prevent similar incidents from happening again.