A simple error can lead to costly, wasted cloud resources.

One example can be found in the customer whose keystroke mistake led to the spin-up of a cloud instance much larger than was required. No one noticed, the job ran all weekend, and the result was $300,000 in unnecessary charges.

Tales such as these are motivating businesses of all sizes to take a closer look at cloud costs and find new ways to manage them. One startup company that has based its business on helping customers manage cloud costs more efficiently is Archera.ai Inc., a Seattle-based firm that offers a way for cloud users to save money and time.

“In the cloud, you’re getting billed by the second, so every time you’re not committed to a machine for a given hour, that’s basically money that you’re basically flushing down the toilet,” said Aran Khanna (right), co-founder and chief executive officer of Archera. “One of our key operational initiatives is to make sure, as soon as the customer plugs in, we’re able to find all that low-hanging fruit and, within less than a week, actually go and automatically address it. So, you’re not burning money on demand, but you have flexibility to make those longer-term optimizations like replatforming or moving to managed services.”

Khanna spoke with John Furrier, industry analyst for theCUBE, SiliconANGLE Media’s livestreaming studio, in advance of the “Analytics and Cost Optimization” AWS Startup Showcase on June 8. He was joined by Nikhil Khanna (left), co-founder and chief technology officer of Archera, and they discussed how enterprises can take advantage of new approaches to reduce spending in the cloud. (* Disclosure below.)

Low-friction cost optimization

The need to control cloud cost has grown as companies embrace artificial intelligence. Model training can require significant GPU resources and most engineers are more focused on getting desired results than managing cloud spending.

“Usually, the engineer is not really incentivized to care too much about the underlying cost of it in most cases. They just want to get a cool model out of the other end,” Nikhil Khanna said. “The best solution is something where they’re able to make those sorts of cost optimizations on the rate you’re paying for the actual machine level. We’re sort of a low-friction way that you can get savings without having to disrupt your engineers.”

Archera’s model is based on a guaranteed buy-back provision that lets users sell commitments for cloud spending soon after purchase.

“If you are a deep learning company and you have an AI workload that might take three months to train, but you don’t want to buy hour-by-hour, we offer this flexible commitment called a Guaranteed Reserved Instance where you can commit for 30 days and then at any point after those 30 days you can click a button and send that commitment right back to us,” Aran Khanna said. “That’s really the core financial offering that we monetize through this free platform.”

Archera is accessible through the AWS Marketplace and will soon be in general availability for Microsoft Azure as well. The company is hiring across North America as interest in controlling cloud outlays continues to grow.

“We’re seeing a huge amount of demand for offerings like ours and the short-term commitments that we offer in order to fuel these deep learning workloads and be able to get cost reductions on them,” Nikhil Khanna said. “There’s a gap that we saw in the market that we realized we could help fill.”

Watch the complete video interview below, and be sure to check out SiliconANGLE’s and theCUBE’s upcoming “Analytics and Cost Optimization” AWS Startup Showcase on June 8.

(* Disclosure: Archera.ai Inc. sponsored this segment of theCUBE. Neither Archera nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE