Majestic Labs has unveiled its Prometheus AI server, designed to tackle the memory constraints that have long hindered the performance of large-scale artificial intelligence systems. The server is equipped with up to 128 terabytes of LPDDR6 memory, a significant leap from the typical memory capacities found in current AI infrastructure. According to the company, this high memory capacity is intended to address the 'memory wall' issue, which refers to the growing gap between processing speed and memory access speed in modern computing systems. The Prometheus server is built to support data-intensive AI workloads, such as training large language models and processing vast datasets, by providing a more efficient memory architecture. Majestic Labs claims that the server's design allows for faster data access and reduced latency, which can enhance the performance of AI applications. The company emphasized that the Prometheus server is a key step toward enabling more efficient and scalable AI solutions. *Source: [ieee](https://spectrum.ieee.org/huge-memory-ai-server)*