Neo Cloud Providers Comparison
Compare GPU-first and AI-native cloud providers - GPU availability, pricing, performance, and specialized AI infrastructure.
Last updated: 2026-02-11
| Feature |  CoreWeaveCoreWeave Inc. |  LambdaLambda Inc. |  Together AITogether AI |  Crusoe CloudCrusoe Energy Systems |  Voltage ParkVoltage Park |  RunPodRunPod Inc. |
|---|---|---|---|---|---|---|
| General | ||||||
| Headquarters | Livingston, NJ | San Francisco, CA | San Francisco, CA | San Francisco, CA | San Francisco, CA | Cherry Hill, NJ |
| Founded | 2017 | 2012 | 2022 | 2018 | 2023 | 2022 |
| Company Type | Private (~$35B valuation) | Private (~$1.5B valuation) | Private (~$3.3B valuation) | Private (~$10B valuation) | Private | Private (~$500M+ valuation) |
| Total Funding | ~$12B+ (debt + equity) | ~$800M+ | ~$400M+ | ~$1.4B+ (Series E) | ~$100M+ | ~$20M+ |
| Business Model | GPU cloud IaaS (bare metal & VMs) | GPU cloud + on-prem GPU servers | AI inference & training API platform | Clean-energy GPU cloud | GPU-as-a-service marketplace | Serverless GPU cloud for AI |
| Target Customers | AI labs, enterprises, hyperscalers | ML researchers, startups, enterprises | AI developers, startups, researchers | AI companies, enterprises | Startups, researchers, AI companies | Indie developers, startups, researchers |
| GPU Availability | ||||||
| Nvidia H100 | ||||||
| Nvidia H200 | Limited | |||||
| Nvidia B200 (Blackwell) | Coming 2025-2026 | Coming 2026 | Coming 2026 | Coming 2026 (GB200) | TBD | TBD |
| Nvidia A100 | ||||||
| Nvidia L40S / L4 | ||||||
| Consumer GPUs (RTX 4090, etc.) | ||||||
| Total GPU Fleet Size(?) | 100,000+ H100s (growing to 250K+) | 10,000+ H100s | Thousands (not disclosed exactly) | Planned 400,000 GB200s (Abilene) | Thousands of H100s | Thousands (community + owned) |
| Pricing | ||||||
| H100 On-Demand (per GPU/hr) | ~$2.23/hr | ~$1.99/hr | ~$2.50/hr (dedicated) | ~$2.35/hr | ~$1.89/hr | ~$2.49/hr |
| A100 80GB On-Demand (per GPU/hr) | ~$1.28/hr | ~$1.10/hr | ~$1.25/hr | ~$1.20/hr | ~$0.99/hr | ~$1.64/hr |
| Reserved / Contract Pricing | 1-3 year contracts (significant discount) | Reserved instances available | Custom enterprise pricing | Long-term contracts available | Flexible commitments | Savings Plans available |
| Spot / Interruptible | ||||||
| Serverless (Pay-per-token/sec) | ||||||
| Price Competitiveness(?) | ~50-70% cheaper than hyperscalers | ~60-75% cheaper than hyperscalers | Competitive for inference APIs | ~50-65% cheaper than hyperscalers | ~60-80% cheaper than hyperscalers | ~50-70% cheaper than hyperscalers |
| AI Services & Features | ||||||
| Managed Inference API | ||||||
| Model Fine-tuning Service | ||||||
| Pre-built Model Catalog | 100+ open source models (Llama, Mistral, etc.) | Community templates | ||||
| Bare Metal Servers | ||||||
| Virtual Machines | Dedicated endpoints | GPU Pods | ||||
| Kubernetes Native | ||||||
| On-Prem GPU Servers (Purchase)(?) | ||||||
| Networking & Storage | ||||||
| InfiniBand(?) | Limited | |||||
| RDMA Support(?) | Limited | |||||
| NVLink / NVSwitch | ||||||
| High-Performance Storage | NVMe SSD, shared filesystems | NVMe SSD, persistent storage | Managed (transparent to user) | NVMe SSD, object storage | NVMe SSD | Network volumes, NVMe pods |
| Object Storage | ||||||
| Infrastructure & Regions | ||||||
| Data Center Locations | US (NJ, TX, IL, WA), Europe (London, Norway) | US (TX, UT, AZ) | US-based | US (TX, WY), Iceland | US-based | US, EU (distributed community + owned) |
| Number of Regions | 6+ | 3+ | 1-2 | 3+ | 1-2 | 10+ (incl. community) |
| Clean / Renewable Energy | Partial (varies by location) | Not emphasized | Not emphasized | Not emphasized | Not emphasized | |
| Uptime SLA | 99.9%+ | 99.9% | 99.9% (API) | 99.9%+ | Best effort | 99.9% (varies by tier) |
| Notable Customers & Partnerships | ||||||
| Key Customers | Microsoft, Nvidia, Cohere, Stability AI | ML researchers, universities, startups | Scale AI, Stanford HAI, Hugging Face | AI labs, enterprise GPU tenants | AI startups, researchers | Indie AI developers, Stable Diffusion community |
| Strategic Investors | Nvidia, Microsoft, Magnetar, Coatue, Cisco | G Squared, Gradient Ventures (Google) | Nvidia, Salesforce, Kleiner Perkins | Nvidia, Fidelity, Mubadala | Not publicly disclosed | Dell Technologies Capital, Intel Capital |
| Nvidia Partnership | Preferred cloud partner, DGX Cloud | Cloud partner, hardware reseller | Investor + GPU partner | Strategic investor | GPU customer | GPU customer |
| Key Differentiators | ||||||
| Primary Strength | Largest independent GPU cloud; Nvidia-preferred; enterprise-grade Kubernetes | Simplicity + best GPU cloud pricing; also sells on-prem servers | Best platform for open-source model inference + fine-tuning APIs | Clean energy powered; vertically integrated (energy + compute) | Lowest-cost H100 access; flexible commitments | Most accessible GPU cloud; serverless + community marketplace |
| Best For | Large-scale AI training, enterprise GPU infrastructure | ML researchers and teams wanting simple, affordable GPU access | Developers building AI apps with open-source models | AI companies wanting sustainable, clean-energy compute | Budget-conscious teams needing bulk H100 access | Individual developers, hobbyists, small teams, inference workloads |