Antimetal is putting AI to work to root out cloud cost inefficiencies

One constant theme we’ve been hearing from the big cloud infrastructure vendors this year is helping customers increase the efficiency of their cloud spend. Cloud bills have grown enormously as companies move more workloads. Combine that with a high level of economic anxiety, and a pronounced switch from a growth to efficiency mindset, and customers have been looking for ways to cut costs.

The cloud companies themselves and other startups have been working on this problem in various ways, but Antimetal, a New York City-based early-stage startup decided to attack the problem with a proprietary machine learning model, concentrating on AWS for starters, but with plans to expand to other major cloud platforms from Google and Microsoft soon.

Today, the company announced a $4.3 million seed investment to continue building the solution they’ve been working on since launching the company last year. The company already has a working product with more than 30 paying customers

“In short, we're using artificial intelligence and machine learning to save companies on their AWS bills,” company co-founder and CEO Matthew Parkhurst told TechCrunch.

“Our model constantly scans thousands of different data points to dynamically scale that coverage and uncover basically any savings opportunities that they might have on their account. We can save customers up to 75% on their AWS bill in less than five minutes, which can translate to hundreds of thousands or millions of dollars annually for some of these customers.”

Typically, companies will contract with AWS for a number of years to bring down costs in exchange for some certainty, but if they aren’t putting all of those resources to work, they have been stuck paying for something that they’re not using. While AWS has an online marketplace for selling these unused resources, Antimetal took it a step further by creating one that uses AI to find the most efficient way to sell them.

“We have an online algorithm which uses AI to study the market dynamics, then resize and relist these instances. And just what we've seen in our private beta it's about three times faster than just going to the marketplace listing it and hoping it'll sell,” Parkhurst said. He says that the AWS marketplace takes 90 days on average to sell spare instances, and Antimetal appears to help do it much faster.

Parkhurst and co-founder and CTO Shreyas Iyer are cognizant of building a defensible product against whatever AWS (and the other cloud vendors) are doing to help customers analyze and bring down costs. Iyer says that while Amazon is clearly working with customers to reduce costs, it’s not a core part of their business and they are looking to partners like Antimetal for help.

“Incentives are really aligned. AWS wants people to engage in reserved instances and buy them. The fact that we're moving a reserved instance from one customer to another doesn't fundamentally change the fact that lock-in to AWS is still really high,” Iyer said. In other words, it shouldn’t matter to Amazon who owns the instances, only that they're in use, regardless of the customer.

The company currently has five employees with plans to double that number in the coming year with the new funds. Iyer says that building a diverse company is a core value for him and Parkhurst, and it’s never too early to start thinking about it.

“Diversity is particularly important, especially at this early stage because it builds a foundation and sets the tone going forward. And so it's something we've been keeping in mind, in particular, trying to reach out to a lot of different forums to hire people from all walks of life, all types of backgrounds,” he said.

Today’s $4.3 million investment was led by Framework Ventures with participation from Chapter One, IDEO CoLab Ventures and other unnamed investors.