Google is marking its latest round of double-digit cloud pricing cuts with the introduction of an entirely new consumption option for infrastructure-as-a-service customers that promises to bring the costs of running certain workloads even lower.

Amazon started the trend a decade ago by offering massive discounts on idle capacity. Google’s new Preemptive Instances represent an ambitious, albeit greatly delayed, attempt to match that.

Both cloud giants are aiming to provide organizations with more choice in consuming cloud infrastructure while squeezing a return out of the surplus infrastructure in their data centers, but the methods by which that goal is achieved are a world apart. Whereas Amazon employs a bidding system designed to maximize the potential savings, Google has taken a more conventional approach that emphasizes predictability.

At least in theory, anyway. Preemptive Instances trade the upwards of 85 percent of cost cuts that can be realized on the retail giant’s platform for a fixed 70 percent pricing reduction that replaces customer demand fluctuations with internal demand fluctuations as the wild card. That is, a virtual machine is shut down when Google needs the capacity rather than after the market rate has exceeded the user’s blanket bid for all instances of that type.

That’s potentially more accommodating to distributed cloud-native applications that can quickly reallocate work if needed, a use case the search giant is actively targeting with an alert system that issues a warning 60 seconds before shutting down a server. That’s half as much time as Amazon provides, but still plenty of time for an automated process to wrap everything up.

The only catch is that a Preemptive Instance is limited to running for up to 24 hours, assuming no preemptive shut-downs, which means that the only services that can feasibly take advantage of available cost savings are those performing work with relatively lax completion expectations. But that still leaves customers with a great many potential applications, such as graphics rendering, scientific modeling and, most importantly, batch analytics.

Google is clearly placing the emphasis on the latter with the addition of a new option to its bdutil command-line tool that enables developers to automatically incorporate Preemptive Instances into their cloud-based Hadoop clusters. The new functionality opens a whole new world of use scenarios on the platform. but it’s not only the search giant’s customers that will be affected by the update.

With both of its biggest rivals now making their surplus capacity available for consumption, Microsoft is bound to follow suit sooner or later, especially considering Hadoop’s central role in its cloud value proposition. The question is how it will manage to set itself apart.