Nvidia A100 vs Nvidia H100

The H100 is newer and has 73% more bandwidth, with the same 80 GB of memory. The A100 is typically cheaper.

At a glance

Memory

Identical

A100: 80 GB

H100: 80 GB

Memory bandwidth

73% higher on H100

A100: 1.94 TB/s

H100: 3.35 TB/s

Cheapest on-demand

51% cheaper on A100

A100: $0.89 /GPU/hr

H100: $1.80 /GPU/hr

Price comparison

Median price per GPU per hour by billing type, with the cheapest listing under it.

Billing	A100	H100	Difference
On-demand	$1.88 /GPU/hr from $0.89	$3.80 /GPU/hr from $1.80	51% cheaper on A100
Reserved	$1.67 /GPU/hr from $0.66 (24 mo)	$3.32 /GPU/hr from $1.53 (1 mo)	50% cheaper on A100
Spot	$1.19 /GPU/hr from $0.68	$2.25 /GPU/hr from $1.29	47% cheaper on A100

Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Medians are across providers, one vote each, so a large catalog does not outweigh a small one. Verify before provisioning. More on how we price.

Price history

A100 vs H100 price history

Weekly median on-demand price per GPU per hour

Last 12 months

Aggregating historical prices...

Specs comparison

Spec	A100	H100	Difference
Memory
Capacity	80 GB HBM2e 40 GB on SXM, PCIe	80 GB HBM3 94 GB on NVL	Same headline SKU
Bandwidth	1,935 GB/s 2,039 GB/s on SXM 1,555 GB/s on SXM, PCIe	3,350 GB/s 2,000 GB/s on PCIe 3,938 GB/s on NVL	73% higher on H100
Bus width	5,120-bit	5,120-bit	Identical
Compute · dense
FP8	Not supported	1,979 TFLOPS	H100 only
FP16 / BF16	312 TFLOPS	989.5 TFLOPS	3.2x higher on H100
INT8	624 TOPS	1,979 TOPS	3.2x higher on H100
FP32	19.5 TFLOPS	67 TFLOPS	3.4x higher on H100
FP64	9.7 TFLOPS	34 TFLOPS	3.5x higher on H100
Platform
Interconnect	NVLink 600 GB/s	NVLink 900 GB/s NVLink 600 GB/s on PCIe, NVL	50% higher on H100
Board power	300 W	700 W	2.3x higher on H100
Released	Q2 2020	Q3 2022	2 years newer on H100
Precision
Supported formats	FP16BF16TF32FP32FP64INT4INT8	FP8FP16BF16TF32FP32FP64INT8

Figures are for each model's headline configuration. Where a form factor carries a different number, it is named under the figure.

Compute: Vendor peak figures, dense and per GPU. Not measured throughput.

Source: A100 datasheet and H100 datasheet.

Where to rent them

Providers that rent these GPUs, with each one's cheapest rate. Providers that rent both are listed first.

Provider	A100	H100
Jarvislabs India	$0.89 /GPU/hr On-demand	$3.49 /GPU/hr On-demand SXM
GPU.ai UAE	$1.07 /GPU/hr On-demand	$1.98 /GPU/hr On-demand PCIe
Hyperstack UK	$1.35 /GPU/hr On-demand PCIe	$2.50 /GPU/hr On-demand PCIe
Spheron Singapore	$1.43 /GPU/hr On-demand PCIe	$2.64 /GPU/hr On-demand PCIe
Massed Compute USA	$1.38 /GPU/hr On-demand SXM	$2.73 /GPU/hr On-demand
Runpod USA	$1.00 /GPU/hr On-demand SXM	--
Thunder Compute USA	$1.09 /GPU/hr On-demand	--
Oblivus USA	$1.47 /GPU/hr On-demand PCIe	--
HyperAI Hong Kong	--	$1.80 /GPU/hr On-demand
Sesterce France	--	$2.75 /GPU/hr On-demand
Scaleway France	--	$3.21 /GPU/hr On-demand PCIe