What actually breaks deep tech startups isn’t lack of vision, but running out of patience before reality catches up.

Follow the Gradient sits down with Roland Siegwart, Professor of Autonomous Systems at ETH Zurich and one of Europe’s most influential robotics mentors. Over two decades, Roland has helped shape ETH’s spin-out culture and advised companies like Anybotics, BlueBotics, and 7Sense from research to real-world deployment .

This conversation explores how deep tech companies are really built when timelines are long, capital is patient but finite, and founders must constantly trade ambition against survival.

We talk about:

Why abundant early-stage capital can quietly reduce urgency in deep tech startups
The hidden cost of the last 10 percent from prototype to market-ready system
How early revenue from “unsexy” applications can enable bigger long-term bets
Why Europe’s strength in hardware systems also makes scaling slower and harder
The role experienced operators play in grounding PhD-driven founding teams
When founders must step aside to prevent becoming the company’s bottleneck

Rather than offering formulas, this episode examines how founders learn to make irreversible decisions with incomplete information, balancing engineering rigor with the pressure to move before the window closes.