Neocloud
A neocloud is a GPU-focused cloud provider that emerged outside the hyperscaler ecosystem to serve AI and high-performance computing workloads. Unlike traditional cloud providers that offer broad infrastructure services, neoclouds specialise in accelerated compute — typically offering bare-metal GPU access, high-bandwidth interconnect fabrics, and pricing models optimised for training and inference workloads. CoreWeave, Lambda, and Nebius are among the most capitalised neoclouds. The term distinguishes these operators from hyperscalers (AWS, Azure, GCP) and legacy managed hosting providers.
Neoclouds typically deploy dense GPU clusters with InfiniBand or RoCE interconnects, offer bare-metal access rather than virtualised instances, and optimise their stack for ML frameworks like PyTorch and JAX. Their data centres run at higher power densities (30-80kW per rack) compared to traditional cloud facilities (8-15kW). Most operate on NVIDIA hardware, though some are diversifying to AMD MI300X and custom silicon.
Neoclouds are the primary subject of our due diligence work. Understanding the competitive positioning, unit economics, and technical differentiation between neoclouds is essential for investors evaluating this space. We track 122 GPU cloud providers and 13 inference-focused neoclouds across our intelligence platform.
This glossary is maintained by Disintermediate as a reference for GPU infrastructure professionals, investors, and operators. Each entry reflects terminology as used in active advisory engagements and market intelligence work.