"Unique data is really the key" – this veteran tech VC shares which AI is really worth investing in

Sitting in the Helsinki office of pan-European VC OpenOcean, General Partner Tom Henriksson is reflecting on his long history with the databases undergirding the modern internet. OpenOcean roughly emerged from the connections formed around an initial stake in MySQL, the open-source database software which is still to be found behind some of the most used web apps and websites today. That investment, around the turn of the millennium, would prove prescient.

“We took a big leap,” said Henriksson. “And obviously it became something excellent, with a billion dollar exit to Sun Microsystems in 2008.”

Investing in core web processes

With that background, OpenOcean today are a highly specialised investor in business-to-business data software, with positions in dozens of companies around Europe and a handful in North America. The firm just announced the €100m close of its fourth fund, with a final target of €130m, much of it to focus on AI.

In the early days of OpenOcean and before, says Henriksson, there were few core web infrastructure tools managing large and complicated databases. Now that’s all changed, with an abundance of such database tools, rapidly expanding in terms of data and processing capacity. And that’s not even to mention the advent of AI-backed data management and processing tools.

Among its investments, OpenOcean looks out for companies that help other businesses get the most out of AI. Henriksson points to two companies in OpenOcean’s portfolio – MindsDB and Cambri – which both offer businesses tools to get the most out of the data they have, improving efficiency and designing more climate-friendly products and production processes. With their Article 8 fund, OpenOcean have an investment approach emphasising what Henriksson calls “responsible software.”

“We invest only in software that actually improves things. So will it be on the right side of privacy, not the wrong side of privacy? Will it not increase consumption of energy, rather make a data centre more efficient, consumption-wise, etc. That is deeply ingrained in us.”

The goal, he says, is to make investments he would be happy sitting at the dinner table with his family talking about.

So many AI 'solutions,' but which actually solve something?

With all that in mind – and considering the thousands of companies OpenOcean looks into per year - Impact Loop wanted to know how he approaches investments related to AI. This is of course relevant to the impact space, given the proliferation of AI-backed tools claiming to be able to help fight and adapt to the climate crisis.

The first thing in evaluating the viability of the project, he says, is establishing whether the proposition actually has the right data.

“Unique data is really the key to a long term-edge,” says Henrikkson. “If you don't have that, (it’s) not a very sustainable position.”

From there, things get granular very quickly, and there are no shortcuts. What’s really required is for AI specialists to get under the hood of a prospective investment’s algorithms and databases. For that, OpenOcean has two AI experts on staff – General Partner Ekaterina Almasque and AI Venture Partner Crystal van Oosterom - who have between them decades in AI development.

They and the other partners also spend a lot of time evaluating the people behind the projects, assessing their knowledge of the various AI verticals and how much they seem to understand the problem they are trying to solve, as well as the environments in which their projects would operate.

“We get into, you know, Where do they come from? What do they really know? What do they know about the theory? What have they been trained on, and so on. And only after all of that, we come to what it is that they have innovated on,” says Henrikkson.

This kind of rigour may not sound like the kind of cheat-code for evaluating AI-backed impact companies that the headline of this article may left you hoping for, but the VC says its essential to learn if a company actually knows what it’s doing, “because otherwise you'll invest in something that sounds good and then actually either cannot technically perform or is just not that unique."