EQT Ventures is one of the largest VC investors in Europe and has raised €2.6 billion across three funds, and invests from Seed up to Series C. Together, EQT Group has invested in over 300 companies, including Einride, Wolt, Marvel Fusion, and Xeltis, as well as unicorns such as Freepik and Beamery.
To understand how the Firm approaches AI, defensibility, and the evolving landscape of early-stage investing, I spoke with Alexander Fred-Ojala, Head of AI at EQT Ventures.
Fred-Ojala leads the AI team developing the platforms, tools, and algorithms that make EQT's early-stage investing AI and data-driven. He also advises on technical aspects of the investment process and helps portfolio companies strengthen their AI capabilities.
Inside EQT's Motherbrain: the AI engine rewiring venture sourcing
When it comes to AI tooling, EQT is probably best known for its proprietary platform Motherbrain, which the Firm describes as "A powerful synergy of AI and human expertise."
Initially launched in 2016 by EQT Ventures as a way to evaluate large volumes of tech startups, Motherbrain uses AI to scan, model, and track investment opportunities, supporting the full investment lifecycle from sourcing, through due diligence, and into portfolio value creation.
For early-stage investing, there are literally millions of opportunities to assess if you're active across Europe, the US, and globally. But a human cannot analyse all those data points. "Motherbrain has been in operation for over 10 years". Fred-Ojala detailed:
"What we're trying to do with these internal tools is build systems that give us an edge and boost productivity. For Ventures investing data-driven methods can especially generate alpha in sourcing and discovering new opportunities."
Historically, this has been achieved through statistical methods, such as monitoring website traffic and utilising predictive models to forecast whether a company is on a strong trajectory.
"There have been so many internal conversations about opportunities—what we thought of a founder, market notes, past assessments," he shared.
Previously, someone had to record and read all those notes to remember the context. Now LLMs can analyse that and proactively say:
"This team member met the company before, they knew the founder, the discussion was about X. When you reconnect, here's the best angle to take," shared Fred-Ojala.
According to Fred-Ojala, what has changed with generative AI and large language models since GPT-3 (2020) and even more so since ChatGPT is that you can now analyse unstructured data. All textual data online — including conversations about a company, PDFs, market research, academic papers, and even internal data — can be understood by the models.
On the sourcing side, the company has built an operating system to navigate opportunities. A dealmaker logs in and sees all relevant opportunities where digital traces exist. They're stack-ranked using interpretable scoring models — dealmakers can see why something is ranked.
"For example, a team member can do deep agentic assessments of opportunities. This system will analyse all publicly available data on an opportunity and cross-run that with internal data. The main bottleneck for doing this at scale is compute.
The tool also integrates metadata like "which internal EQT person knows this company", "which advisors have met them", helping with network intelligence and leveraging the scale of the EQT platform.
It'll also rank by traction, foundation, and relevance to EQT Ventures— "it'd be unlikely that we invest in 20-year-old companies."
"We want early-stage companies with high potential."
"People still underestimate how transformative AI will be."
I asked whether Fred-Ojala sees Europe as a foundational model region, or whether our biggest opportunities are elsewhere. He admits that although his role focuses on building internal AI systems, he also works closely with deal teams when the Firm assesses AI opportunities.
He contends that, if we look at frontier model development, Europe has leaders:
“Mistral is the standout example: they're only a few months behind the cutting edge, perhaps three to six months, and they're innovating in ways that directly compete with the major labs.
And let's not forget: DeepMind was founded in the UK and still drives a large portion of Google's AI breakthroughs."
However, it is extremely CapEx-intensive to build frontier models, and it may not be necessary for Europe to compete. Instead, he suggests:
"We might actually be at 'peak foundational-model scale' and it wouldn't surprise me if the market corrects. Short-term hype or bubble? Possibly. Long term, though, I think people still underestimate how transformative AI will be."
However, Fred-Ojala and I share the opinion that (at least for now) Europe is strongest in AI applications, where companies like Lovable, Cradle, Leya, and Helsing build vertical, industry-specific solutions on top of foundation models.
He shared:
"Europe can accelerate here because our markets are so diverse and nuanced—AI thrives in environments where it can optimise complex workflows, and Europe is full of them.
That's where I believe Europe's biggest opportunity lies."
Agentic AI will rewrite every knowledge field
Fred-Ojala has zero doubt that we're in a paradigm shift rather than simply an AI bubble, and maybe the biggest in technology history. He says it's comparable to electricity and the internet:
"We're creating external cognitive engines that solve problems for us. That changes everything."
He sees white-collar work and digital workflows as the start, asserting,
"In a few years, it will feel archaic to type on a keyboard and stare at a static screen. We'll instruct systems and manage agentic workflows instead.
Look at software engineering: OpenAI just launched an agent-creation framework. Internally, they prototyped it in six weeks, and Codex, their own code-generation agent, wrote 80 per cent of the code.
That would've been science fiction two years ago."
From here on, agentic work will spread into every knowledge field, including biotech, chemistry, physics, and maths:
"GPT-5 Pro solves novel math problems. DeepMind's AlphaEvolve discovers new algorithms. These are goosebump moments."
Regarding a bubble, he asserts that, if anything, the short-term bubble is on the "the CapEx side — data-centre build-outs — more than venture funding itself."
Crypto looked for problems. AI solves them
Like many of us, Fred-Ojala has lived through the blockchain hype cycles — the ICO days, Web3, and NFTs — business models dependent on platform economics.
Before joining EQT Ventures, Fred-Ojala served as Research Director of the Data Lab at UC Berkeley's Sutardja Centre for Entrepreneurship and Technology (SCET) and co-founded the Berkeley Blockchain Xcelerator, whose alumni have collectively raised over $600 million.
He admits that while crypto often felt like a tech looking for a problem. AI is the opposite.
"It solves problems now. Even if progress stopped today, distributing and integrating current capabilities would create massive value.
The main bottleneck isn't technology—it's people. Changing workflows, mindsets, and habits takes time. That's why enterprise ROI measurements often look messy."
The rise of new power centres
As our conversation shifted from hype cycles to real-world impact, Fred-Ojala pointed out that the next wave of AI value won't be confined to familiar hubs like London, Paris, or Berlin.
Fred-Ojala explained that EQT recently analysed AI activity across Europe and found that innovation is emerging far beyond the usual regional power centres.
Instead, AI accelerates research in physics, chemistry, materials, and optimisation, which opens new physical-world applications and the opportunity for highly durable businesses.
Beyond that, "we ended up creating three categories," he explained. The first is Full-Stack Powerhouses—regions with both a high number of AI startups and strong investment volume. "Stockholm is the standout here, with more than 50 AI startups and over €205 million invested," he noted.
The second group, Founder Factories, consists of cities with high startup density but comparatively lower funding. "Tallinn actually has the most AI startups per capita in Europe, with 360 companies per million residents" he said.
Finally, there are the Money Magnets, ecosystems with fewer startups but significant capital inflows. "Heidelberg and Cambridge fall into this category," Fred-Ojala explained.
"They're deep-tech academic cities that consistently attract large checks."
Complexity is no longer a moat
When asked how he distinguishes defensibility from hype in today's AI market, Fred-Ojala was clear:
"One major shift is that technical complexity is no longer a moat, especially in software," he said.
With tools like Lovable, he noted, "you can prototype a Slack-like SaaS product quickly, then have engineers refine what the AI can't finish."
Instead, the attributes that make an AI startup truly fundable are changing.
He contends that velocity matters: "the teams that adopt AI workflows the fastest will outpace everyone.
”Distribution and brand are key assets "because mindshare compounds globally, and deep domain expertise is now a critical differentiator, the kind of vertical focus that frontier labs won't casually replace."
Ultimately, his advice to founders is straightforward:
"Solve one or two problems exceptionally well. Don't try to solve 50. Generic wrappers won't last."
"We're only scratching the surface"
When asked where the AI landscape is heading over the next three to five years, Fred-Ojala cautioned that forecasting exponential progress is always difficult—but the direction of travel is clear.
"Services industries will be heavily disrupted," he said, pointing to a future where end-to-end GenAI workflows take over large swathes of customer support, legal, finance, and accounting."
He also expects software itself to become increasingly fluid:
"We'll see real-time generated interfaces—software that reshapes itself around the user."
Fred-Ojala believes several long-dormant technologies are poised for a comeback.
"AR and VR will resurge," he said, adding that humanoid robotics are finally approaching a level of maturity where they can become "meaningful contributors" across industries.
Even if the pace of AI progress slows, he stressed that we're still early:
"We're only scratching the surface of current capabilities."
Would you like to write the first comment?
Login to post comments