H200/H100 Market Disruption:

• Nebius sets new baseline: H200 SXM at $1.89/GPU/hour

• Hyperstack leads H100 pricing: $1.90/hour

• RunPod follows with H100 PCIe at $1.99/hour

 

A100 Price War Intensifies:

• Hyperstack: A100 PCIe now $0.95/hour (-29.6%)

• RunPod: A100 PCIe reduced to $1.19/hour (-27.4%)

• Gcore: 8xA100 SXM clusters at €1.67/GPU/hour (-11.6%)

 

New Enterprise Options:

• DataCrunch: 4xB200 clusters at $4.90/GPU/hour

• Seeweb: 8xMI300X clusters at $2.35/GPU/hour

• Contabo: 4xH200 clusters at $0.79/GPU/hour

 

Market Analysis:

• Virtualized instances showing steeper discounts than bare metal

• Multi-GPU clusters becoming increasingly cost-effective

• European providers driving market competition

• Consistent downward pricing trend across all GPU tiers

 

What does this mean for AI teams?

• More options for scaling AI workloads

• Better price-performance ratio for large deployments

• Increased accessibility to premium GPU resources

• Opportunity to optimize cloud spending

 

Find out more about GPU Cloud opportunities on gpucompare.com