Nutonian CEO: Business and Science Face Same Desperate Big-Data Problem
Is “big data” starting to fade away as a tech-business theme? Hardly.
A couple of Boston-area startups in the field just announced Series A funding rounds this week. Sqrrl, a Cambridge, MA-based company founded by former NSA scientists, raised $5.2 million from Atlas Venture and Matrix Partners. The startup has developed a new kind of secure database and is led by CEO Mark Terenzoni, who joined from F5 Networks early this year.
And today, fellow Cambridge startup Nutonian says it has raised $4 million led by Atlas Venture to build out its big-data analytics software platform—with the promise of discovering key patterns and relationships in mountains of data and translating them into insights for businesses. (Of course, that’s a Holy Grail—more on that below.)
One person these two startups have in common is Atlas partner Chris Lynch, who led both investments. Lynch, the former CEO of Vertica Systems and Acopia Networks, certainly seems to be doing his part to keep Boston in the broader big-data discussion. He’s involved in other local startups, including Hadapt, as well as the workspace hack/reduce near Kendall Square.
Nutonian (pronounced “Newtonian”) is based on some pretty deep science. The company’s founder and CEO, Michael Schmidt, did his PhD at Cornell University in computational methods for data mining in physics and biology. In a 2009 paper in Science, Schmidt and his Cornell advisor, Hod Lipson (also an advisor to Nutonian), showed they could develop an algorithm to derive fundamental laws of physics based only on experimental data; their key advance was automating that centuries-old scientific process.
Leave it to a VC like Lynch to see the business value here. “I offered Michael a term sheet the day I met him,” Lynch says. “He is a superstar.” Boston-area tech luminaries Jit Saxena and Cheng Wu invested with him in the startup, and Lynch bullishly predicts that the three of their business accomplishments, combined, “will be a pebble in the ocean against the value and impact Michael and Nutonian make.”
The idea behind Nutonian is to apply Schmidt’s algorithmic approach not only to science but to fundamental business problems—sales forecasts, manufacturing predictions, stock trading, and so on. The 14-person startup, which Schmidt (pictured) started in 2011, says its technology already has been used by thousands of scientific researchers and by people in the life sciences, oil and gas, retail, and finance industries.
This is one of those plays where the technology itself is key. Nutonian’s software, known as Eureqa, is based on a technique called symbolic regression, which is related to artificial-intelligence methods such as machine learning and genetic programming.
I wanted to get a better sense of what makes this company (and founder) tick—and what its real chances of success might be. So here’s a transcript of a Q&A with Schmidt:
Xconomy: This feels like a Holy Grail startup to me—exciting but very challenging. Why is the market timing right?
Michael Schmidt: Nutonian is based on new technology that can detect laws of physics from data. But there are almost no “laws of physics” in most businesses. With the data being collected today we can change that, giving businesses the same advantages that scientists have leveraged for years.
Most of today’s sophisticated [business intelligence] tools make “black box” predictions from your data: they can tell you what’s likely to happen, but not why it will happen or how you can avoid or reproduce it in the future. It’s all just correlation without any deeper causal understanding or meaning.
Our software, Eureqa, was created to solve that problem, by uncovering and explaining the relationships within big data that drive predictions.
The market is ready now because the data storage technology is mature. We can store and access enormous data sources quickly and efficiently. Great companies were started and solved that problem. But, the companies we talk with are spending millions storing that data annually; there’s a desperation to deliver on the promises of big data.
X: Your target customers seem very broad in terms of industry. How do you focus, and where’s your sweet spot?
MS: We’ve had incredible traction in science and medicine—everything from characterizing the formation of galaxies in the universe, to improving cancer detection. Many researchers have cited our software in their results.
We see a broad interest in the industry as well. For example, we’ve seen a major chemical manufacturer scale their production of pharmaceuticals by using Eureqa to pinpoint the limiting factor in their production (e.g., how heating and mixing rates impact yield). We’ve also seen retailers using our technology to anticipate changes in store sales across all their products.
Our goal is to be able to serve anyone that needs to predict, control, or understand phenomenon in the data they collect.
X: Do you see Eureqa mainly as transforming business intelligence, or changing the way we do science, or perhaps morphing into something else that has even greater societal impact?
MS: I want to help the world scale with the amount of data we collect and the things we want to understand. It used to be that a lone scientist could retreat to their chalkboard and come back with a unified theory on how something works. Today we need armies of researchers to make progress.
Fundamentally, I see the problem that scientists face as the same problem facing businesses. There are things that are staggeringly complex to understand. The difference today, and only within the past few years, is that we can now efficiently store and access massive amounts of data and information on them. We will increasingly need to rely on technologies that enable us to detect what’s important and explain the hidden patterns; and that’s where we’re going with Eureqa.