Dear hustlers, founders, operators and visionaries,
Today’s guest is Roland Siegwart, Professor of Autonomous Systems at ETH Zurich, who initiated ETH’s robotics master program and helped catalyze dozens of deep tech spin-outs in robotics and autonomous systems. He has been at ETH since 2006 and has been directly involved in shaping funding structures, talent programs, and company formation long before teams formally incorporate.
🎧 Tune in now on Spotify, Apple, YouTube and share your thoughts! In the meantime: Follow the Gradient and stay tuned!
PS: Has this e-mail been forwarded to you? Sign up here.
Why you should listen
You should listen to this if you are building deep tech or robotics and feel the tension between long development cycles and the pressure to scale fast.
As the conversation unfolded, it became clear why early revenue, painful market focus, and leadership changes matter more than vision alone in hardware-heavy companies.
What we talk about
00:00 Introduction
03:51 How Switzerland built a self sustaining robotics ecosystem
05:02 Turning university research into real world companies
07:39 Why Europe struggles to scale deep tech globally
09:46 Funding strategies for long hardware driven timelines
12:16 The painful gap between lab prototypes and products
14:09 Early signals that founders can survive the transition
15:37 Finding a first market with real customer pain
19:00 When pilots turn into scalable robotics businesses
20:47 Bringing business leadership into technical teams
23:24 When founders must step aside to let companies grow
31:49 Navigating dual use and ethical responsibility in robotics
Our main take away’s
Deep tech fails when capital replaces urgency: Siegwart argues that abundant early VC in Switzerland reduced pressure to find paying customers, which later exposes startups that optimized for grants and awards instead of revenue.
Hardware startups must plan for a 10x effort gap after the prototype: He explains that moving from proof of concept to market-ready systems often takes at least ten times more resources, with the last 10 percent of product readiness consuming as much time as the first 90 percent.
