Majestic Labs

Majestic Labs builds power-efficient AI servers designed to overcome the "memory wall" that limits GPU-based AI systems on the largest and most advanced workloads. Its flagship server architecture supports both training and inference for large language models and delivers up to 128 terabytes of RAM per server - the memory capacity of 10 racks of traditional equipment packed into a single machine - enabling more users per server and shorter training times.

The company develops core components in-house, including custom accelerator silicon built by its own silicon engineering team and internally designed memory interface modules that underpin the memory capacity of its systems. Its technical work spans silicon engineering, custom accelerator chip design, memory interface modules, AI server architecture, large language model training and inference, AI infrastructure and high-memory computing.

Majestic Labs was founded by former Google and Meta executives with deep silicon engineering expertise. The company has raised $100 million in initial funding and plans first system deliveries starting in 2027. Teams are located in Israel and California. Target verticals include hyperscale data centres, financial services, pharmaceuticals and AI/machine learning infrastructure.