Meet Aaron Rosenberg: The man DeepMind researchers call before they quit

SPEAKER_1 (0:01)

This podcast is brought to you by Chargebee. As an AI founder, you're doing the hard work of reimagining entire industries, repeatedly identifying use cases that abundant intelligence and domain specific know-how can solve and shaping the future in the process. Chargebee enables founders just like you to monetize that work as inventively as you build your products. Whether you want to charge based on usage credits, outcomes, seats, or a hybrid model that combines all of those alongside complex enterprise contracts, Chargebee can help. It is the billing, metering, and monetization infrastructure purpose built for this era. Learn more at chargebee.com/sifted.

That's chargebee.com/sifted.

John Thornhill (0:45)

Hello and welcome back to the Sifted Podcast. I'm your host, John Thornhill. Joining me today in London is Aaron Rosenberg, partner at Radical Ventures and former head of strategy and operations at Google DeepMind. Last month, Toronto headquartered Radical Ventures became the largest venture capital fund in Canadian history, having reached the first close of more than $1 billion for a new growth stage fund.

Since 2023, Aaron has led the firm's UK presence, which has involved betting on many of his former DeepMind colleagues at an early stage. His investments include Frontier Biotech startup, Latent Labs, Neolab Inherent, which builds agents for scientific research, and super forecasting company, Mantic, all of which were founded by DeepMind alumni.

We will get into how DeepMind is shaping the UK's AI ecosystem, Radical Ventures' unique proposition, and what trends Aaron is seeing among Europe's emerging cohort of Neolabs. But first, welcome Aaron.

Aaron Rosenberg (1:44)

Thank you so much for having me.

John Thornhill (1:46)

You've had several different roles in the tech industry. Can you talk us through how each one led to the other?

Aaron Rosenberg (1:53)

Sure. Yeah, happy to do so. So immediately people will perhaps discern that I'm not from here. I am from New York originally and started my career in consulting, of all things, at McKinsey. Spent a few years there, ended up having the privilege to work with the head of the firm as his chief of staff, and then went to graduate school where I grew increasingly fascinated by this field of machine learning.

At the time, this was in 2015, all the exciting advances were happening in research labs, primarily academic and a few corporate. I had a friend of a friend at DeepMind, and so started conversations there that culminated in getting to meet Demis Hassabis, the founder and CEO. And it just so happened that he was looking for someone kind of like myself, someone who had served in a deputy capacity.

Extremely fortunately for myself, he hired me, and that's what brought me to London nine years ago.

John Thornhill (2:46)

And so what was that role exactly?

Aaron Rosenberg (2:48)

Was to work with him in the office of CEO. I think the official title was principal, somewhat nondescript. And I was responsible for a wide array of activities. At the time, DeepMind was 500 or so research scientists and engineers.

And so there was a lot of operational strategic work to be done that lay outside of their expertise. And so I gravitated to those kinds of activities, which included opening new offices in new locations, helping to set up our internal governance group around ethics and safety, and then increasingly thinking about how to commercialize the research in various ways.

John Thornhill (3:27)

That's quite a prism through which to view the AI industry. But how long did you work there, and why did you decide to make the jump to Radical?

Aaron Rosenberg (3:34)

So I was there for about six years in total. The first three as Demis' deputy, and then three as head of strategy and operations, as you mentioned. And I really had a phenomenal experience. I thought I could be there for the rest of my life, honestly.

In many respects, it was like living in the future. We were seeing all the advances that were very much on the frontier. And I guess it was around the time that we were spinning out Isomorphic Labs, which was an incredible project to which I was able to contribute in a small way alongside many other amazing people, that I was catching up with a classmate of mine, one of the only other people in business school back in 2015 who was excited about AI. He had gone on to an interesting stint in autonomous vehicles and then had joined this firm, Radical Ventures, to start the San Francisco office with Fei-Fei Li as a partner, esteemed computer scientist at Stanford for anyone who's unaware. And when he mentioned that he was doing so alongside her, I thought that was quite exceptional. There are not that many venture firms that have someone like Fei-Fei in their midst. So I figured I should get to know more of these people, and so I got to know the managing partner and his co-founder and learned about their history, having been machine learning founders themselves, having built and sold a series of companies, and taking the proceeds from the last one to start the firm they wish had existed when they were entrepreneurs, one that was dedicated exclusively to AI, when that was not really a thing. Now everyone, the whole world of venture is artificial intelligence. In 2017, it was very much not that way. And one of their co-founders had been a student of Jeff Hinton's, and so they had basically started the firm around Jeff's lab at the University of Toronto, co-founded the Vector Institute in Toronto, all these significant institutions in the world of AI.