Neoclouds Emerge as AI Computing Powerhouses Amid Energy and Compute Shortages

The emergence of 'neoclouds', a new class of hyperscalers specializing in accelerated compute for AI workloads and HPC, has sparked debate about their ability to dominate the AI era.

Initially, these neoclouds dropped in valuation when Meta announced plans to sell excess compute, but experts argue that this validated their business models given the shortage of both compute and energy.

These companies have been formed by a mix of legacy players like bitcoin miners and new venture-backed entities capitalizing on GPU shortages. They focus on provisioning GPU-centric compute for AI workloads via massive data centers they construct or colocate, often energized with on-site power generation rather than relying on the grid.

While some neoclouds function as bare-metal GPU wholesalers, others are building out software and orchestration layers to address enterprise workloads. However, their business models remain fragile due to factors like depreciation and circular financing models that have drawn criticism.