Blended GPU compute costs fell 2.1% this week across major cloud providers, led by RTX PRO 6000's 87.7% gain.
Top Movers
| GPU | Blended Price | WoW Change | Class |
|---|---|---|---|
| RTX PRO 6000 | $0.58/hr | ▲ 87.7% | General |
| M60 | $0.28/hr | ▼ 46.1% | Inference |
| K80 | $0.23/hr | ▼ 24.3% | Training |
| MI25 | $0.23/hr | ▼ 21.5% | General |
| INFERENTIA2 | $0.12/hr | ▼ 16.0% | Inference |
| H200 | $10.44/hr | ▲ 15.0% | Training |
| MI300X | $3.40/hr | ▼ 13.8% | Training |
Blended pricing = average of spot, on-demand, and 1-year reserved rates across major cloud providers.
Training vs Inference
Training-class GPU pricing fell this week (avg $4.28/hr, -2.6% WoW), while inference-class pricing fell (avg $0.52/hr, -8.3% WoW).
The training-to-inference price ratio stands at 8.3x — widening compared to last week. The elevated spread suggests strong demand for training compute relative to inference, consistent with ongoing large model training activity.
Regional Spotlight: Asia Pacific
Asia Pacific trades at a 2% premium to global averages this week, with 25 GPU types available across the region. The most expensive GPUs in the region are GB200 ($14.72/hr), H200 ($10.10/hr), H100 ($6.27/hr).
For detailed pricing data across all Asia Pacific sub-regions, see the full regional profile.
Implications
For cloud buyers: Europe continues to offer the lowest average GPU pricing ($2.13/hr blended average). For workloads with regional flexibility, the gap between Europe and Middle East is $2.21/hr — a 104% premium. Compare regional pricing →
For semiconductor analysts: H100 pricing resilience signals sustained demand despite Blackwell availability — NVIDIA's pricing power remains intact. Blackwell (B200) blended pricing at $7.19/hr (-0.5% WoW) provides an early read on next-generation adoption curves. View all GPU profiles →
For GPU investors: Declining rental rates are compressing returns — monitor payback period shifts closely. Model scenarios with the GPU ROI Calculator →