Side-by-side GPU cloud pricing comparison with live data and real-time availability.
Google Cloud is cheaper for 6 of 8 shared GPU models.
AWS brings the broadest GPU lineup, the most regions, and the most battle-tested multi-node networking (EFA); Google Cloud counters with TPUs, deep committed-use discounts, and tight Vertex AI / GKE integration. Both are premium-priced — neoclouds like Lambda and Runpod undercut either by 2-5x on raw $/hr.
AWS
$3.06/GPU/hr
on-demand
—Google Cloud
$3.05/GPU/hr
on-demand
—Difference: $0.01 (0%)
AWS
$2.74/GPU/hr
on-demand
—Google Cloud
$2.24/GPU/hr
on-demand
—Difference: $0.50 (18%)
AWS
$3.43/GPU/hr
on-demand
—Google Cloud
$2.42/GPU/hr
on-demand
—Difference: $1.01 (30%)
AWS
$6.88/GPU/hr
on-demand
—Google Cloud
$4.77/GPU/hr
on-demand
—Difference: $2.11 (31%)
AWS
$7.91/GPU/hr
on-demand
—Google Cloud
$5.20/GPU/hr
on-demand
—Difference: $2.71 (34%)
AWS
$14.24/GPU/hr
on-demand
—Google Cloud
$8.68/GPU/hr
on-demand
—Difference: $5.56 (39%)
AWS
$1.01/GPU/hr
on-demand
—Google Cloud
--
—AWS
$17.80/GPU/hr
on-demand
—Google Cloud
--
—AWS
$1.64/GPU/hr
on-demand
—Google Cloud
--
—AWS
$0.23/GPU/hr
on-demand
—Google Cloud
--
—AWS
$0.76/GPU/hr
on-demand
—Google Cloud
--
—AWS
$1.86/GPU/hr
on-demand
—Google Cloud
--
—AWS
--
—Google Cloud
$2.22/GPU/hr
on-demand
—AWS
--
—Google Cloud
$0.98/GPU/hr
on-demand
—AWS
$0.38/GPU/hr
on-demand
—Google Cloud
--
—AWS
$2.52/GPU/hr
on-demand
—Google Cloud
--
—AWS
$3.36/GPU/hr
on-demand
—Google Cloud
--
—AWS
$0.42/GPU/hr
on-demand
—Google Cloud
--
—AWS
--
—Google Cloud
$5.26/GPU/hr
on-demand
—AWS
--
—Google Cloud
$8.76/GPU/hr
on-demand
—AWS
--
—Google Cloud
$1.96/GPU/hr
on-demand
—AWS
--
—Google Cloud
$17.56/GPU/hr
on-demand
—AWS
--
—Google Cloud
$3.46/GPU/hr
on-demand
—AWS
$1.34/GPU/hr
on-demand
—Google Cloud
--
—AWS
--
—Google Cloud
--
—AWS
$1.65/GPU/hr
on-demand
—Google Cloud
--
—AWS
$1.98/GPU/hr
on-demand
—Google Cloud
--
—| GPU Model | AWS on-demand $/GPU/hr | Google Cloud on-demand $/GPU/hr | Difference | Cheaper |
|---|---|---|---|---|
| L4 | $0.80/GPU/hr on-demand | $1.13/GPU/hr on-demand | $0.33 (29%) | AWS |
| T4 | $0.53/GPU/hr on-demand | $0.73/GPU/hr on-demand | $0.20 (28%) | AWS |
| V100 | $3.06/GPU/hr on-demand | $3.05/GPU/hr on-demand | $0.01 (0%) | Google Cloud |
| A100 | $2.74/GPU/hr on-demand | $2.24/GPU/hr on-demand | $0.50 (18%) | Google Cloud |
| A100-80GB | $3.43/GPU/hr on-demand | $2.42/GPU/hr on-demand | $1.01 (30%) | Google Cloud |
| H100 | $6.88/GPU/hr on-demand | $4.77/GPU/hr on-demand | $2.11 (31%) | Google Cloud |
| H200 | $7.91/GPU/hr on-demand | $5.20/GPU/hr on-demand | $2.71 (34%) | Google Cloud |
| B200 | $14.24/GPU/hr on-demand | $8.68/GPU/hr on-demand | $5.56 (39%) | Google Cloud |
| A10G | $1.01/GPU/hr on-demand | -- | -- | -- |
| B300 | $17.80/GPU/hr on-demand | -- | -- | -- |
| Gaudi HL-205 | $1.64/GPU/hr on-demand | -- | -- | -- |
| Inferentia | $0.23/GPU/hr on-demand | -- | -- | -- |
| Inferentia2 | $0.76/GPU/hr on-demand | -- | -- | -- |
| L40S | $1.86/GPU/hr on-demand | -- | -- | -- |
| P100 | -- | $2.22/GPU/hr on-demand | -- | -- |
| P4 | -- | $0.98/GPU/hr on-demand | -- | -- |
| Radeon Pro V520 | $0.38/GPU/hr on-demand | -- | -- | -- |
| RTX PRO 4500 | $2.52/GPU/hr on-demand | -- | -- | -- |
| RTX PRO 6000 | $3.36/GPU/hr on-demand | -- | -- | -- |
| T4g | $0.42/GPU/hr on-demand | -- | -- | -- |
| TPU v2 | -- | $5.26/GPU/hr on-demand | -- | -- |
| TPU v3 | -- | $8.76/GPU/hr on-demand | -- | -- |
| TPU v5 Lite | -- | $1.96/GPU/hr on-demand | -- | -- |
| TPU v5p | -- | $17.56/GPU/hr on-demand | -- | -- |
| TPU v6e | -- | $3.46/GPU/hr on-demand | -- | -- |
| Trainium | $1.34/GPU/hr on-demand | -- | -- | -- |
| Trainium2 | -- | -- | -- | -- |
| Virtex UltraScale (VU9P) | $1.65/GPU/hr on-demand | -- | -- | -- |
| Virtex UltraScale+ (VU47P) | $1.98/GPU/hr on-demand | -- | -- | -- |
AWS
$0.09/GB
Google Cloud
$0.12/GB
Only on AWS (13)
Only on Google Cloud (7)
Choose AWS if...
Choose Google Cloud if...