EO StudioUtopia or Dystopia? Humanoid Robot NEO Has Arrived at Our Home | 1X, Bernt Børnich
CHAPTERS
- 0:00 – 1:00
Why failure-tolerant culture drives real innovation
Bernt frames failure as an inevitable byproduct of doing work that’s never been done before—and therefore a prerequisite for innovation. The only unacceptable failures are not trying hard enough or failing to reflect and learn under pressure.
- •Innovation means many attempts will be wrong in hindsight
- •Celebrate failures that come from genuine effort and learning
- •Reflection after failure is what converts mistakes into progress
- •Maintaining this mindset under deadlines and pressure is the real test
- 1:00 – 2:03
Meet 1X and NEO: humanoid robots designed for the home
Bernt introduces himself and 1X’s mission: creating “abundance of labor” via intelligent humanoids that live with people. The company name and “1X” idea reflect a focus on natural, human-speed motion rather than sped-up demos.
- •1X builds humanoid robots specifically for home environments
- •“1X” branding pushes for natural, real-time movement in demos
- •Mission: free humans to focus on meaningful, human activities
- •Humanoids must be useful—not toys—while remaining affordable
- 2:03 – 2:33
Origin story: early fascination with computers and ASIMO’s human-like promise
Bernt traces his motivation back to childhood computing and the magic of software affecting the physical world. Honda’s ASIMO inspired him because it wasn’t just automation—it was a social, interactive robot presence.
- •Early computing sparked interest in controlling the physical world with code
- •ASIMO felt like ‘Star Wars’ because of human interaction, not factory tasks
- •The dream: robots handling daily chores like fetching items and helping at home
- •Humanoid usefulness depends on interaction as much as mechanics
- 2:33 – 3:33
Why ASIMO-style robotics struggled outside the lab
He explains that classical robotics approaches rely on rigid assumptions about the environment—assumptions that collapse in messy, unpredictable homes. True intelligence and alignment require robots to live and learn among people in real settings.
- •Traditional robotics depends on controlled, assumption-heavy environments
- •Homes are “creative chaos,” where assumptions break constantly
- •Robots must be safe, affordable, and genuinely capable at real work
- •To align with humans, robots need real-world exposure and learning among us
- 3:33 – 4:35
Building 1X early: a small field, a Norway-based talent magnet, and belief before hype
Bernt describes the company’s early years when humanoid robotics was a tiny community and 1X could gather top talent into a tight-knit team. The upside of being early was focus and camaraderie; the downside was constant fundraising difficulty.
- •10-year journey starting with a small, highly concentrated talent pool
- •Norway provided space to build with little external noise
- •Team intensity: “living and breathing” the problem together
- •Being early meant fewer investors and more skepticism
- 4:35 – 5:05
The COVID fundraising cliff and the two ways deep tech companies lose
A near-closed round collapsed when COVID hit, forcing painful downsizing. Bernt distills survival into two existential risks: losing speed (velocity) or running out of money.
- •Fundraising was one signature away when COVID shut markets down
- •Team cuts were emotionally costly, especially for relocated talent
- •Deep tech survival constraints: maintain velocity and avoid cash-out
- •Pressure makes culture and execution harder—but more important
- 5:05 – 5:35
A new robotics paradigm: tendon/cable drives as a long-term moat
1X committed early to tendon-driven, compliant actuation—an approach that’s difficult and time-consuming, but hard for others to copy quickly. This sets the foundation for safer, more natural interaction in unstructured environments.
- •Shift from conventional designs to tendon/cable-driven systems
- •The approach isn’t new in concept, but rare to execute deeply at scale
- •Long development cycles create defensibility (‘moat’)
- •Design aims for natural movement and safer real-world interaction
- 5:35 – 7:36
Safety through low energy: why high-gear industrial robots don’t belong in homes
Bernt explains the physics: high gear ratios store large kinetic energy, making impacts dangerous and hard to stop quickly. Since real-world manipulation is full of collisions, home robots must be low-energy, compliant, and forgiving.
- •Kinetic energy scales with speed squared—risk rises fast
- •Industrial robots use high gear ratios, creating huge internal energy
- •Factory robots succeed because environments are calibrated and predictable
- •Home robots must handle constant collisions safely for people and objects
- 7:36 – 8:06
Early home deployments reveal the ‘social’ nature of household labor
Testing in homes taught 1X that even simple tasks (like opening the fridge) are inherently social because robots must signal intent and move safely around people. Physical labor among humans requires social awareness as part of the task.
- •Most household actions are social when people share space
- •Robots must communicate intent implicitly through motion and timing
- •Safety and task execution can’t be separated in human environments
- •Real-world diversity is a key driver of machine intelligence
- 8:06 – 8:36
The macro case: abundant labor to meet demographic and cost pressures
Bernt compares future labor abundance to today’s energy abundance: it becomes an expected utility. He argues humanoid robots are needed because aging populations and rising costs strain society’s capacity to deliver services.
- •Vision: physical labor becomes as accessible as electricity
- •Aging demographics reduce available caregivers and workers
- •Labor scarcity influences prices of goods and services
- •Goal: accelerate the path to widespread, practical deployment
- 8:36 – 9:37
Manufacturing before the market: vertically building components, tools, and factories
Because key components for 1X’s approach don’t exist off-the-shelf, the company has spent years building not just robots but also the machinery and automation to produce them. A major theme is simplification—lower part counts, lighter designs, and tight feedback loops between engineering and the factory floor.
- •Many competitors integrate off-the-shelf parts; 1X must build its own
- •10 years spent on foundational tech plus production equipment
- •Design-for-manufacture: simplify materials, reduce weight, minimize parts
- •Co-locating design and manufacturing speeds iteration and quality
- 9:37 – 11:37
Failure in public: the ‘Eve’ demo crash and what it taught about culture
Bernt recounts a stage demo where the robot unexpectedly reversed, hit a wall, and fell—an example of how robotics often goes wrong despite preparation. He uses it to reinforce that learning-focused failure tolerance is essential, not optional.
- •Real-world demos expose edge cases and unpredictability
- •Public setbacks are part of pioneering hardware development
- •The lesson: treat mistakes as learning opportunities, not taboo
- •Cultural resilience enables continued experimentation
- 11:37 – 12:38
How robots learn: bootstrapping with internet data, simulation, and teleoperation
He outlines the training pipeline: start with large-scale internet data and simulation, then collect robot-specific data via teleoperation where a human ‘embodies’ the robot. Over time, autonomy improves through iteration and real-world trials.
- •Data stack: internet data → synthetic/simulated data → robot data
- •Teleoperation transfers human skills and intent into robot behavior
- •Early autonomy emerges once sufficient demonstrations exist
- •Scaling learning requires many robots experimenting in the real world
- 12:38 – 14:09
Why consumer adoption must come before enterprise for humanoid robots
Bernt argues that breakthrough technologies often scale through consumers first because enterprise is risk-averse and slow. He uses ChatGPT as an analogy: widespread bottom-up usage creates pressure that eventually forces top-down enterprise adoption.
- •Enterprise friction: IT barriers, risk aversion, organizational red tape
- •Consumer markets enable faster adoption and diverse use discovery
- •ChatGPT example: consumer virality drove workplace demand
- •Humanoids need early adopters and ‘true believers’ to accelerate progress
- 14:09 – 17:09
Lessons for deep tech founders: solve the hard problem, build usefulness, stay excited
He closes with founder advice: deep tech begins far from customers because the market appears once the core problem is solved—but you still must ship usefulness, generate revenue, and align the technology with humans. Sustaining motivation requires choosing problems that are both meaningful and exciting enough to attract talent and endure dark days.
- •Deep tech is primarily about solving a fundamental technical problem
- •You can’t wait a decade in a lab—deploy gradually to prove usefulness
- •Revenue and productization matter to sustain long journeys
- •Choose problems that are fun and exciting to attract world-class teams
