AI compute could become more expensive even as GPUs improve

Majestic Labs, a startup founded in 2023 by former Google and Meta engineers, has unveiled a server called Prometheus that replaces Nvidia GPUs with custom Arm-powered AI processing hardware. The server packs up to 128TB of unified LPDDR6 memory onboard, promising a 1,000× increase in memory capacity per processor compared to traditional GPU systems.

Majestic Labs claims that the Prometheus server could cost between 10 and 50 times less than a GPU system of equivalent performance while consuming less electricity per rack. The server is designed to be OCP-compliant and will support PyTorch, vLLM, and OpenAI's Triton frameworks, allowing existing AI models to run without modification.

However, several details remain unclear, including the exact number of memory aggregation chiplets required for a 128TB configuration.

Brief written by urgent.news from TechRadar's own syndicated text. Machine-written — may contain errors; check the original before relying on it.