01BGPU cloud · hyperscaler provider· quotes observed 2026-10-14
Google Cloud GPU prices
Every current Google Cloud price in the quote table, one table per billing model, each with its listing, observation date and source. On-demand list prices show their place among every comparable on-demand price for the same chip.
On-demand list prices
On-demand5 prices · place among comparable on-demand list prices for the chip| Chip | Listing | US$ per hour | Place | Observed | Source |
|---|---|---|---|---|---|
| H100 | Google Cloud (A3)a3-highgpu-8g; 8 × H100, 208 vCPUs, 1,871GB memory, 6,000GiB local SSD | $11.06per GPU-hour | 11th of 12 | Google Cloud (A3) published pricing | |
| A100 80GB | Google Cloud (A2 Ultra)a2-ultragpu-8g; 8 × A100 80GB, 96 vCPUs, 1,360GB memory, 3,000GiB local SSD | $5.07per GPU-hour | 8th of 8 | Google Cloud (A2 Ultra) published pricing | |
| TPU v6e | Google CloudTrillium / TPU v6e; per chip-hour | $2.70per chip-hour | only quote | Google Cloud published pricing | |
| TPU v5e | Google CloudTPU v5e; per chip-hour | $1.20per chip-hour | only quote | Google Cloud published pricing | |
| TPU v5p | Google CloudTPU v5p; per chip-hour | $4.20per chip-hour | only quote | Google Cloud published pricing |
Prices are US$ per accelerator-hour (per chip-hour for TPU and Trainium). Observation dates vary by row: an unavailable source keeps its last valid quote, never a new observation date. Attribution for these prices: “AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Google Cloud”.
Background reading: Google Cloud on AI Wiki. AI Wiki is run by the same team as AI Compute Tracker.