Morgan Stanley is investing millions in the startup that helped crack the Panama Papers, bringing its total funding to $160 million

Advertisement
Morgan Stanley is investing millions in the startup that helped crack the Panama Papers, bringing its total funding to $160 million

D3C_7048

Neo4j

Neo4j CEO and co-founder Emil Eifrem

Advertisement
  • Startup Neo4j announced Thursday it raised $80 million in Series E financing, led by One Peak Partners and Morgan Stanley Expansion Capital.
  • Neo4j is known for developing graph databases, a newer way to store data that can be even faster than the more widely-used relational databases.
  • Since Neo4j was founded 11 years ago, graph databases have become increasingly popular, and even Amazon has introduced its own graph database.
  • Neo4j will use the funding to invest in artificial intelligence, which can help with making predictions in fields like retail or even cancer research.

Not many startups can take credit for helping to power cancer research, bolstering a space mission to bring people to Mars, and exposing government corruption with the Panama Papers.

But Neo4j can. On Thursday, the Silicon Valley database startup announced that it raised $80 million in Series E financing, led by One Peak Partners and Morgan Stanley Expansion Capital. This doubles Neo4j's total funding to $160 million in all.

When Neo4j was founded 11 years ago, it introduced a relatively novel concept: graph databases, something that Emil Eifrem, Neo4j's CEO and co-founder, has been evangelizing his entire career. Now graph databases are growing in use, and even Amazon Web Services has entered the space with its own AWS Neptune.

"Amazon doesn't go into markets that aren't real," Eifrem told Business Insider. "They only choose to invest if it's there. Their customers have told them loudly and clearly they want a graph database. That by and large is a good thing."

Advertisement

So what, exactly, is a graph database?

For context, today's widely-used databases are relational databases, which have been around since Oracle pioneered them in the 1970s. Relational databases organize information into tables by category, much like a spreadsheet.

"I grew up as a professional programmer in the 90s," Eifrem said. "When you put an architecture together, the only database choice was Microsoft, IBM, Oracle. For all these years, we reflexively put any data in a relational database."

But these aren't really built for finding links between two data points among massive databases - the system basically has to comb every single entry looking for commonalities, which takes up time and computer memory.

On the other hand, graph databases are built so that one can easily find the connections. While relational databases look like spreadsheet tables, a graph database looks a little bit more like connect-the-dots. Graph databases are especially effective for uses like fraud detection and automatic recommendations, where the information doesn't fit neatly into a traditional database.

Advertisement

When detecting fraud, graph databases make it easier systems to detect when, for instance, someone's credit card was used simultaneously in both New York and Los Angeles. Likewise, with recommendations, people can see recommendations based on previous activity, like products they previously bought or phrases they recently searched.

graph database

Neo4j

A simple example of a graph database. It's all about the relationships between data points.

It's like how children learn, Eifrem says. His youngest daughter just turned one year old and is starting to learn new words, like "mom" and "dad."

"Every new thing she learns, she learns by relating it to stuff she already learns," Eifrem said. "'Dad' is sort of like 'mom.' Machines are the same way. The context is how things are related to each other. There's a lot of applications where graphs are critical for AI."

The future of graph databases lies in artificial intelligence, Eifrem believes, and that's how Neo4j plans to spend its new cash influx. Neo4j has been working with eBay on a shopping bot to recommend products to customers based on what they've bought in the past. Research organizations are even using Neo4j's databases to study cancer.

Advertisement

"We help the world, but we don't do it ourselves," Eifrem said. "We're tool builders. We enable the world to do it."

And Neo4j has been able to help with some compelling, world-changing projects already in its existence.

When NASA engineers were working on the the Orion spacecraft, which could take astronauts to Mars in the future, they struggled with figuring out why the returning capsule wasn't flipping correctly when it lands on the water. But using Neo4j's databases to scope the problem, they finally solved the problem within hours. Relational databases are certainly useful, Eifrem said, but in many cases, graph databases are more powerful.

Read more: A new way to search for data is helping NASA's biggest brains save millions and get to space faster

Neo4j even helped journalists crack the Panama Papers and Paradise Papers, which showed how the world's elite hid away their massive wealth in offshore bank accounts. The graph databases helped reporters and researchers map out all the connections between various offshore bank accounts. After the Panama Papers went public, former Iceland prime minister Sigmundur Davíð Gunnlaugsson stepped down.

Advertisement

"That's a pretty massive political impact we had," Eifrem said. "They say for investigative journalism, there's a pre-graph database era and a post-graph database era."

By making these kinds of connections, too, it's easier to make predictions. You can learn more about something based on what it's connected to. For example, if your friends on Facebook lean politically to the left or to the right, then it's likely you do as well. Similarly, if lots of people click on a new TV after adding a video game console to their shopping cart, then a shopping website can assume that you will, as well.

With more companies like Amazon entering the graph database space, Neo4j might have a bit more competition, but the idea that Eifrem has been trying to evangelize all these years will also be louder and clearer.

"I will never say that relational bases are dead, and go away," Eifrem said. "Graph is for when there's connections. We're frequently a thousand times faster."

See more: How a nerdy Swedish database startup with $80m in funding cracked the Paradise Papers

Advertisement
{{}}