Google is pursuing Amazon Web Services in a fresh place — providing access to “on demand” supercomputing resources.

The service declared in May, empowers individuals to buy computer processing that Google is not using at a steep reduction, provided the customer is willing to concede the resources on short notice.

A.W.S. has a similar short term use plan, called Spot Cases. It gives itself to supercomputing activities, especially for businesses and universities that do not need to purchase multimillion-dollar machines of their own, and has been used in places like drug discovery and learning about new stuff.

The inclusion of Google as a competitor probably will boost how many ways while lowering costs, these resources are used.

Google additionally is apparently stressing its abilities in machine learning, a robust instrument for locating new designs in large data sets.

Using cancer researchers at the Broad Institute, Google used 51,200 computing of ways they’re expressed, the centers to look at the interrelationships between human genes, the cell lines from some 500 kinds of cancer, and The thought was to sort through billions of data points immediately, looking for promising places for research workers to seek treatments and drugs.

The investigation, which about 30 years would have run on one computer server, took a few hours, said the temporary manager of Information Technology at Broad, Chris Dwan.

“This is not like computing a few cab excursions,” said Mr. Dwan. “This is genuinely hefty computing in complex biological systems, steering system research.”

Losing the machines for a brief while in these activities that are supercomputing isn’t fatal, so long as the job was created to back information up and restart. Often, the number crunching in such jobs is comparatively short, while sorting out what the results mean can take.

Cycle Computing, the business that specializes in cloud supercomputing and set up the job for Broad, has done jobs on A.W.S. that use 150,000 centers.

The chief executive of Cycle, Jason Stowe, noted that as Broad did running a supercomputing job, costs maybe one third as much as it did in 2012. The improvement that is intriguing is in the growing utilization of machine learning to direct research that is real.

“What occurs in Life Sciences, where they’ve used a lot of computers for quite a long time, tends to lead what occurs elsewhere,” he said. Now it is not only evaluation, it is directing future determinations.”