Table of Contents
A new class of GPU providers have risen up in the era of AI
The rise of AI has created an unprecedented hunger for GPU compute — and traditional cloud giants are struggling to keep up with the demand of both compute, and the services around that compute. Into that gap has stepped a new class of infrastructure provider: the so-called NeoClouds. These specialized operators focus exclusively on delivering high-performance GPU capacity to AI developers, researchers, and enterprises, often at a fraction of the cost charged by hyperscalers.
The sector is currently growing at over 200% annually, with GPUaaS revenue forecast by some to hit $180 billion by 2030. That trajectory reflects not just rising demand, but a fundamental shift in how AI infrastructure is provisioned and consumed.
Here’s a look at these new players in the AI ecosystem and what they actually offer.
What are NeoClouds?
NeoClouds are specialized cloud providers focused exclusively on GPU-as-a-Service, built from the ground up to serve AI and high-performance computing workloads. Unlike traditional hyperscalers that offer broad, general-purpose cloud services spanning storage, networking, databases, and compute, NeoClouds concentrate delivering bare-metal access to cutting-edge GPUs with infrastructure optimized specifically for deep learning and AI training.
The operational model differs meaningfully from what enterprises have come to expect from AWS, Azure, or Google Cloud. NeoCloud providers typically offer transparent, flat-rate pricing on a per-GPU hourly basis, with no long-term contracts required. Clusters can be deployed in under fifteen minutes, and customers can scale elastically from a handful of GPUs to thousands without navigating complex procurement processes or waitlists.
This focus enables NeoClouds to move faster than their larger counterparts. While hyperscalers have faced multi-year timelines to build out high-density AI infrastructure, specialized providers have demonstrated the ability to deploy GPU clusters within months. The ecosystem now comprises approximately 190 distinct operators globally, collectively reshaping expectations around cost, availability, and time-to-market for AI compute.
Key players
CoreWeave has emerged as perhaps the most prominent name in the NeoCloud space. The company operates as a specialized NVIDIA partner and has grown very quickly since its March 2025 IPO, with its stock more than doubling in value. CoreWeave serves a broad customer base that includes AI labs, machine learning startups, and notably, hyperscalers themselves seeking to augment their own capacity.
Lambda Labs has carved out a developer-first identity, building on its heritage in AI workstations. The company maintains a broad GPU fleet spanning from A100s to the latest B200 accelerators, with transparent pricing and a strong focus on Kubernetes-based orchestration. Lambda has also pursued a leaseback model with NVIDIA, allowing it to maintain access to cutting-edge hardware.
Nebius, headquartered in Amsterdam, has built a global presence spanning North America, Europe, and the Middle East. The company secured $700 million in funding in 2024 followed by another $1 billion in 2025, positioning it to support large-scale deployments including H100, H200, and GB200/300 clusters. That capital base gives Nebius the firepower to compete for major enterprise contracts.
On the more accessible end of the market, RunPod has gained traction for providing immediate access to cutting-edge GPUs with minimal friction, while Voltage Park operates at considerable scale with over 24,000 NVIDIA H100s deployed across six global Tier 3+ data centers. Both emphasize bare-metal performance and predictable pricing as core differentiators.
NeoCloud advantages
The cost differential between NeoClouds and hyperscalers represents one of the most compelling reasons for adoption. Specialized providers price GPUs much lower than major cloud platforms, with some operators offering potential cost reductions of up to 66% compared to traditional approaches. Companies like CoreWeave and Lambda leverage 30 to 50% discounts tied to multi-year supply contracts, passing those savings through to customers. Without the overhead of enterprise toolset sprawl, these focused providers can consistently undercut the majors.
Performance architecture provides another advantage. NeoCloud infrastructure is built around GPU-centric nodes linked by high-bandwidth connectivity, including NVLink-4 providing up to 900 GB/s intra-node bandwidth and support for InfiniBand for ultra-low latency workloads. Bare-metal access ensures zero hypervisor drag, delivering the full performance envelope of the underlying hardware without virtualization overhead.
Speed to market has proven decisive for many customers. NeoClouds can stand up high-density GPU infrastructure in months rather than the multi-year timelines typical of hyperscale data center builds. That agility matters enormously in a field where model architectures and training requirements evolve rapidly.
Availability may be the most immediate draw. During periods when hyperscalers have faced GPU shortages and lengthy wait times for top-end accelerators, providers like CoreWeave, Lambda Labs, and RunPod have offered immediate access to cutting-edge hardware. Even OpenAI turned to CoreWeave to augment shortcomings in Azure’s capacity, a powerful validation of the NeoCloud value proposition.
Strategic relationships
The relationship between NeoClouds and hyperscalers has evolved from pure competition to something more symbiotic. Microsoft and Google are increasingly partnering with specialized GPU providers to fill compute supply gaps within their own platforms. When OpenAI needed additional capacity beyond what Azure could deliver, it turned to CoreWeave. This integration validates the NeoCloud model and demonstrates that even the largest technology companies recognize specialized GPU infrastructure fills a critical gap they cannot address alone.
Enterprise customers are leveraging NeoClouds as part of broader vendor diversification strategies. Hugging Face exemplifies this approach, spreading experiments across different third-party clouds rather than committing entirely to a single hyperscaler. The flexible, non-exclusive nature of NeoCloud offerings makes them natural components of multi-cloud strategies designed to avoid lock-in.
Infrastructure partnerships have proven essential to NeoCloud scaling. Colocation firms like Digital Realty and Aligned Data Centers enable these providers to host GPUaaS offerings within existing facilities, avoiding the capital intensity and time requirements of building data centers from scratch. This symbiotic relationship allows NeoClouds to expand capacity rapidly while colocation operators benefit from high-value tenants driving power and space utilization.
Big implications
The NeoCloud phenomenon carries significant implications for the broader AI ecosystem. Most immediately, these providers are democratizing access to enterprise-grade GPU compute. Startups, research labs, and smaller AI teams can now access the same caliber of infrastructure that was previously available only to well-capitalized technology giants. The elimination of long-term contracts and the availability of hourly pricing models lower barriers to experimentation and reduce the financial risk of AI development.
With approximately 190 operators now active globally, customers have genuine choices that create competitive pricing pressure and prevent any single provider from dictating terms. The ability to combine services from multiple providers without lock-in creates a more dynamic marketplace.
However, the business model carries risks too. Expanding AI capacity requires massive capital outlays for power guarantees, infrastructure investment, and chip purchasing. Billions in funding and debt financing cycle through the sector, enabling rapid growth while concentrating financial risk among a relatively small group of well-capitalized companies. The sector’s heavy reliance on venture and debt financing raises questions about long-term profitability and sustainability.
Supply chain dependency represents another vulnerability. NeoClouds’ competitive advantage relies significantly on their ability to secure cutting-edge GPUs from NVIDIA. Their business model depends on maintaining favorable allocation relationships and the ability to purchase inventory quickly. Any disruption to that supply chain, whether through allocation changes or broader shortages, would directly impact their ability to serve customers.
The rapid consolidation around well-funded players, evidenced by CoreWeave’s successful IPO and Nebius’s billion-dollar funding rounds, suggests that despite lower entry costs for customers, the provider side may see significant concentration. The capital requirements of operating at scale may ultimately limit how many NeoClouds can survive and thrive as the market matures.
