Majestic Labs, a Tel Aviv-based startup founded in 2023 by former Google and Meta engineers, has unveiled the Prometheus server, designed to rival Nvidia's GPU and HBM combination by replacing GPUs with custom Arm-powered Ignite AI Processing Units that combine Arm cores with RISC-V vector and tensor engines. Each Prometheus server can house up to 12 AIUs, sharing between 8 TB and 128 TB of LPDDR6 memory across one contiguous, coherent pool, accessed through custom memory aggregation chiplets linked by copper cables up to one metre long. A standard 40U rack can hold four such servers, drawing 120 kW total and cooled through cold-plate liquid systems. By comparison, an Nvidia DGX B300 system with eight Blackwell GPUs offers 2.3 TB of HBM3e plus up to 4 TB of DDR5 system memory; Majestic claims its architecture delivers over 50 times more fast memory and 1.7 times the interconnect bandwidth. The company said one Majestic rack holds the fast memory capacity of 25 Nvidia NVL72 Vera Rubin racks at a fraction of the power, with up to "1000× more memory per processor." Majestic Labs says the server could cost between 10 and 50 times less than a GPU system of equivalent performance once it ships next year, and it will support PyTorch, vLLM and OpenAI's Triton frameworks. Founded by CEO Ofer