Over the past few years, neoclouds have emerged as a new class of cloud providers mainly focused on renting high-end GPU infrastructure for AI workloads. Driven by surging demand for large language models, real-time inference, and research applications, they offer rapid deployment, competitive pricing, and greater flexibility than many traditional cloud services. Rooted in cryptocurrency infrastructure, neoclouds aren’t replacing hyperscalers; they’re complementing them, filling specialised niches in the rapidly evolving tech landscape.
Overview of neoclouds
Neoclouds are cloud providers devoted almost entirely to renting high-end GPUs for AI workloads. They power hyperscalers such as Microsoft, Meta, Oracle, etc., that build large language models (LLMs), delivering specialised computing capabilities quickly and at scale. Also, companies with pressing AI needs (startups, research labs, and enterprises) are increasingly choosing this new type of cloud provider for their workload. Nvidia is central in this ecosystem, both supplying hardware and investing in neoclouds. Key neoclouds players include CoreWeave, Lambda, Nebius, Nscale, and Crusoe.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Neoclouds’ appeal comes from competitive pricing, speed, and flexibility. Neoclouds offers GPU as a service (GPUaaS), their core business strategy, at rates between two and seven times cheaper than hyperscalers. They offer a quick setup and place a few constraints on vendors. Neoclouds also enable digital sovereignty through their localised and specialised operations.
Interestingly, many neocloud players come from the cryptocurrency world. For example, founded in 2017 under the name Atlantic Crypto, CoreWeave began as a cryptocurrency-mining company. They were able to transition from cryptocurrency mining because they already had the high-density GPU infrastructure and expertise needed to meet the growing demand for AI computing. As the crypto market cooled, these companies pivoted their GPU fleets toward generative AI training and inference.
In addition, GPUaaS is the core strategy of neoclouds. GPUaaS is a cloud-based model allowing organisations to rent GPUs in the cloud instead of buying and managing costly hardware. This service also allows users to scale their GPU resources based on their needs. GPUaaS is gaining traction rapidly. According to Research and Markets, the GPUaaS market will be worth $7.36bn in 2026, growing from the 2025 value of $5.70bn, to $26.43bn in 2031, growing at 29.12% compound annual growth rate (CAGR) over 2026-2031.
Neoclouds do not compete with traditional cloud providers
In an environment still defining how AI will scale, neoclouds are not necessarily a competitor to hyperscale providers. Instead, they can serve as valuable partners, acquisition targets, or sources of GPU capacity when the big cloud providers face demand spikes.
GlobalData sees neoclouds as complementary to hyperscalers for multiple reasons. For example, AI is advancing rapidly. New model architectures, real-time inference, data sovereignty, and more, create untapped opportunities for specialised players. Neoclouds can also fill gaps in clients’ multi-cloud strategies, securing positions in niches such as sovereign computing and specialised workloads. Furthermore, neoclouds are deeply tied to data centre operators; they are dependent on shared infrastructure for space, power, and networking. This makes it difficult for them to compete with hyperscalers who own and control their massive, vertically integrated facilities.
However, neocloud providers still need to reduce their risk exposure. For example, they have a high concentration of revenue, with a few large customers accounting for much of their sales. Neoclouds are highly vulnerable if one major contract is lost. These companies also rely heavily on a limited number of suppliers, such as Nvidia or AMD, for hardware stock. CoreWeave, for example, reported that three suppliers accounted for 76% of all purchases in 2024, with Nvidia being the key supplier for their GPU technology. Microsoft contributed 65% and 67% of its revenue in Q3 2024 and Q3 2025, respectively.
