Our customers see a decrease in cost by 50% to train their models.

We offer 3 configurations with 2 pricing models

vCPUs
RAM
Storage
Nvidia's A100 SXM GPU
A100 SXM 40GB
vCPUs
15.5
RAM
200GiB
Storage
512 GiB SSD
Predictable spot
$1.05/gpu/hr
Guaranteed uptime
$1.42/gpu/hr
Nvidia's A100 SXM GPU
A100 SXM 80GB
vCPUs
30
RAM
225GiB
Storage
2.5 TiB SSD
Predictable spot
$1.45/gpu/hr
Guaranteed uptime
$1.97/gpu/hr
Nvidia's H100 SXM GPU
H100 SXM 80GB
vCPUs
26
RAM
225GiB
Storage
3.25 TiB SSD
Predictable spot
$2.79/gpu/hr
Guaranteed uptime
$3.85/gpu/hr
27% discount
Predictable spot
$1.05/gpu/hr
$1.45/gpu/hr
$2.79/gpu/hr
Guaranteed uptime
$1.42/gpu/hr
$1.97/gpu/hr
$3.85/gpu/hr
  • Get started with $200 in free compute.
  • No payment required for data movement.
  • GPUs come pre-installed with popular ML frameworks.
Reserve GPUs Today

Prices listed in the chart above are “Plus applicable sales tax”. GPU instances are reserved weekly, either with predictable spot (3hrs/day downtime) or guaranteed uptime.

Self-service (On-demand) instances coming soon!

Pricing comparison

Short term reserved capacity on 1 week minimum term.

Predictable spot
High uptime
A100 SXM 40GB
Chip type/config
A100 SXM 40GB
A100 SXM 80GB
H100 SXM 80GB
Predictable interruption
(5–8pm CT)
$1.05/gpu/hr
$1.45/gpu/hr
$2.79/gpu/hr
Traditional cloud providers

Classic spot pricing — uncertainty when interruptions will occur

$1.15/gpu/hr
$1.61/gpu/hr
$3.93/gpu/hr
$1.57/gpu/hr
$1.57/gpu/hr
$6.91/gpu/hr
Not offered
$1.02/gpu/hr
$6.98/gpu/hr

Across providers, 90% of spot instances are interrupted within 4 hours. Our servers are guaranteed to run for 21 hours of the day, with a predictable 3 hour interruption.

A100 SXM 40GB
Chip type/config
A100 SXM 40GB
A100 SXM 80GB
H100 SXM 80GB
Our offering
$1.42/gpu/hr
$1.97/gpu/hr
$3.85/gpu/hr
Traditional cloud providers
$4.10/gpu/hr
$5.12/gpu/hr
$12.29/gpu/hr
$3.67/gpu/hr
$5.02/gpu/hr
Not offered
$3.40/gpu/hr
$4.10/gpu/hr
$6.98/gpu/hr
GPU cloud alternatives
$2.06/gpu/hr
$2.21/gpu/hr
$4.76/gpu/hr
$3.09/gpu/hr
$3.18/gpu/hr
$5.95/gpu/hr
Not offered
$2.55/gpu/hr
$4.69/gpu/hr

Get started in 3 easy steps

1

Sign up for your short term rental.

sign up
2

Schedule or checkpoint your training scripts.

3

Log in to your machine and get going!

Build AI has created a new pricing model — predictable spot

There are 3 types of pricing models in the industry today and Build AI is creating the 4th.

1

On-demand

Expensive

Customer charged only when GPUs are actively being used.

$3.72/hr
2

Reserved

Slight discount

Customer charged whether they use the GPUs or not: 1mo, 3mo, 6mo, 1yr, or 3yr commitments.

$2.62/hr
3

Spot

Cheap

Customer gets discount because their instances get shut down when higher paying customer comes along.

$1.36/hr
4

Predictable spot

Cheapest

We marry the low-price of spot instances with the predictability of on-demand and reserved instances.

$1.05/hr

Predictable spot is the lowest price on the market, powered by renewable energy, & built with the planet in mind.

Get started