By Julia, Associate at FORWARD.one
AI commands the spotlight. It attracts capital, headlines, and pitch decks at an unprecedented rate. But capital intensity doesn’t indicate maturity. We’re still in the earliest innings of AI, and the market behaves as if we’re deep into the game. $193B of venture capital has gone into AI in 2025 alone (Pitchbook data). This is a historic concentration of money in a sector. This says more about investors than about the sector’s actual maturity.
Every week, I review around 40 early-stage deeptech decks. 20 of them AI. Most rely on the same foundation: thin wrappers around big-tech models. If your company’s competitive moat disappears the moment API pricing changes, you don’t have a moat, you have a dependency. At FORWARD.one, we invest in defensibility, not convenience.
Our stance is simple: we don’t invest in hype rounds. We look for teams who know what it means to build under pressure. Ideally, second-time founders and people who’ve already learned how to break things, fix them, and keep going. Commercialization-first builders with enough scar tissue to scale industrial technology in a real market, not a trend cycle.
Innovation rarely yields its defining applications during its hype phase. Electricity came long before the lightbulb. The iPhone came years before dominant apps like Instagram. Infrastructure leads; breakthroughs follow. Today’s AI boom feels similar. The models exist. The real value which is durable, monetizable, irreplaceable value will come from founders who build real-life applications beyond the stack. Not only a thin-layer on top of it. Companies that are best positioned to benefit are the ones where an AI model reduces costs and increases productivity in defined sectors with defined target customers, like European manufacturers.
This is why we sometimes delay AI investments. If the market is overheated, we step back. Not because we lack conviction in AI, but because performance requires discipline, and discipline means timing the market with clarity, not enthusiasm.
Meanwhile, the next wave of computational performance is already forming. Quantum computing is moving from theoretical promise into engineered reality. When quantum and AI converge, we’ll see what “intelligence at scale” truly means. FORWARD.one backs the builders preparing for that future, not the noise around today’s hype.
And to be clear, the discipline we demand from founders is the discipline we demand from ourselves. Inside our investment team, we actively apply AI models and automation in our own operations. Not as a toy. Not as a marketing line. But as literacy. You cannot invest in a frontier you don’t understand at a functional level.
We believe in AI. We fund commercial reality. And we follow performance, not plans.
Fizyr: vision AI software for robotics. They use deep learning algorithms to enable robots to cope with unknown objects. Targeting logistics for automated picking and stacking.
QuantWare: A company preparing for that future where quantum computing can accelerate AI by achieving speedups in training of neural networks.
We are keen to invest in AI models able to deliver value for SMEs in the industrial space. Since this could be a solution for the pressing costs and labor shortage in this industry.
Europe is entering a new industrial era where AI, robotics, and quantum are converging. The challenge isn’t invention, Europe has world-class technical research, it’s commercialization. The winners of the next decade will be those who turn deeptech complexity into industrial adoption, fast. That’s why disciplined capital matters. The market is drowning in AI noise, but genuine industrial AI which is applied in factories, supply chains, energy systems, and precision automation is still underserved.
We’re also watching the shift from software-centric AI companies to hardware-enabled intelligence. Robots that learn. Optical systems that optimize. Quantum processors that accelerate. This is where Europe’s engineering advantage is strongest. But these opportunities need investors who understand technology timelines, not just valuation curves.
FORWARD.one was built for this moment. Where commercialization is the real test, and performance is the only currency that counts. We follow performance, not plans.
