Google has announced the launch of GPUs attached to preemptible VMs, offering a 50% discount – but with a catch.

As with the preemptible VMs, first announced in 2015 but with prices significantly lowered in August last year, resources can be used for a maximum of 24 hours. In addition, Google Compute Engine can shut them down with a 30 second warning. The ideal place for these instances is distributed, fault-tolerant workloads – hence the substantial discount.

Users will be able to attach NVIDIA K80 and NVIDIA P100 GPUs to preemptible VMs for $0.22 and $0.73 per GPU hour respectively.

Google adds that the preemptible GPUs will be “a particularly good fit for large-scale machine learning and other computational batch workloads as customers can harness the power of GPUs to run distributed batch workloads at predictably affordable prices.”

As is always the way with these announcements, Google rolled out a happy customer – in this instance, healthcare technology provider Silicon Therapeutics. “Preemptible GPU instances from CSP give us the best combination of affordable pricing, easy access and sufficient scalability,” said CSO Woody Sherman. “Preemptible GPU instances have advantages over the other discounted cloud offerings we have explored, such as consistent pricing and transparent terms.

“This greatly improves our ability to plan large simulations, control costs and ensure we get the throughput needed to make decisions that impact our projects in a timely fashion,” added Sherman.

You can find out more in a blog post here.