Neocloud messaging confuses enterprise buyers despite rapid AI infrastructure growth
Specialised AI cloud providers are scaling fast, but unclear marketing is hindering adoption among enterprises redesigning their infrastructure.
Key points
- Neocloud revenues hit $9 billion in Q4 2025, up 223 per cent year over year.
- ABI Research forecasts more than 2,200 neocloud data centres globally by 2035.
- Confusing provider messaging is preventing enterprises from including neoclouds in their plans.
Enterprises actively redesigning their artificial intelligence infrastructure are rarely including neoclouds in their strategic plans, according to an analysis by InfoWorld. This oversight persists despite the sector experiencing the fastest-scaling infrastructure growth in industry history. The disconnect stems not from a lack of technical merit, but from marketing communication that fails to clearly articulate value to potential buyers.
What defines a neocloud platform?
A neocloud is a specialised cloud environment focused entirely on serving AI-based systems, rather than offering the thousands of general services provided by traditional hyperscalers. These platforms deliver specific hardware components, including GPU compute power, high-bandwidth memory, and fast interconnects. This architecture is purpose-built to handle the rigorous demands of AI training and inference workloads efficiently.
How fast is the market expanding?
Synergy Research Group reports that neocloud revenues reached $9 billion in the fourth quarter of 2025 alone, representing a 223 per cent year-on-year increase. Full-year revenues for the sector exceeded $25 billion. The market is forecast to approach $400 billion by 2031, growing at a sustained compound annual rate of 58 per cent. CoreWeave alone has surpassed $5 billion in annual revenue faster than any previous cloud platform in history.
What infrastructure changes are expected?
ABI Research projects that GPU-as-a-service revenue from neocloud providers will surpass $250 billion by 2030, rising from $42 billion in 2025. The physical footprint is also expanding rapidly, with forecasts indicating more than 2,200 neocloud-operated data centres globally by 2035. This is a significant increase from the 558 facilities recorded in 2025, reflecting the dense deployment of liquid-cooled GPU clusters.
Why are enterprises hesitating to adopt?
Neoclouds offer an architectural solution to AI workload constraints around parallelism and compute concentration that generalised cloud elasticity cannot serve. They sit between public hyperscalers and managed service providers. However, InfoWorld notes that the messaging from providers is confusing the very buyers they need. This lack of clarity means enterprises often overlook these purpose-built layers during infrastructure redesigns.
What to do and how to stay safe: CoreWeave
- Review current AI infrastructure plans to identify gaps in GPU compute and high-bandwidth memory capacity.
- Evaluate whether generalised cloud services can efficiently handle specific training and inference workload constraints.
- Request clear technical comparisons between hyperscaler offerings and specialised neocloud architectures from vendors.
- Monitor provider messaging for clarity on how their solutions address parallelism and locality requirements.
Step-by-step guide: Shared Responsibility Model Cloud: Who Owns What Security
General security guidance from the DarkWeb Dispatch newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What is the projected revenue for neoclouds by 2031?
The market is forecast to approach $400 billion by 2031, according to Synergy Research Group.
How many neocloud data centres are expected by 2035?
ABI Research forecasts more than 2,200 neocloud-operated data centres in operation globally by 2035.
What specific hardware do neoclouds provide?
Neoclouds deliver GPU compute, high-bandwidth memory, and fast interconnects purpose-built for AI workloads.
