All In Interview - Peter Funkhauser, Bernt Bornich, Amanda McMaster, Jonathan Hurst - Recap and Notes - July 29 2026
July 29 2026
(from Gemini)- Dr. Péter Fankhauser – Co-founder and CEO of ANYbotics
- Four-Legged Form Factor & Harsh Environments: Four-legged "dog-like" robots excel in complex industrial spaces because four legs provide superior balance, dexterity, wide footholds, and stability over slippery floors, stairs, snow, and rain compared to bipedal humanoids. ANYbotics focuses on data collection and routine inspection across critical infrastructure rather than simple human labor replacement.
- Superhuman Industrial Inspection & ROI: Equipped with thermal cameras, acoustic sensors, microphones, and gas detectors, the robots perceive anomalies invisible to humans, such as micro gas leaks and overheating equipment. The primary financial return comes from eliminating facility downtime, which can cost hundreds of thousands of dollars per hour, justifying high-end onboard GPUs and sensors.
- Edge Autonomy vs. Cloud Processing: Real-time operations like obstacle avoidance and image capture verification run directly on the robot's local compute to handle connectivity loss and prevent uploading corrupted data. Cloud compute is reserved for historical trend analysis and predictive maintenance across fleets.
- Extreme Deployments & Explosion-Proof Engineering: The robots operate in harsh environments ranging from sub-zero temperatures in Norway (-20°C) to desert heat (+60°C) and offshore transformer platforms for wind and oil facilities. ANYbotics specifically engineered an "Ex-certified" spark-free robot that safely inspects volatile, explosive environments filled with methane gas.
- Supply Chain Sovereignty & The Chinese Hardware Threat: ANYbotics sources zero percent of its components from China, relying on European and U.S. supply chains to satisfy strict cybersecurity and data privacy demands from critical infrastructure clients. While Chinese manufacturers produce capable and inexpensive hardware, Fankhauser emphasizes that long-term enterprise value lies in end-to-end software integration, inspection intelligence, and trusted data handling.
- Ethical Stance on Military Weaponization: ANYbotics strictly develops commercial inspection systems and actively opposes mounting offensive weapons on mobile robots, co-signing an industry-wide pledge with peers like Boston Dynamics against the weaponization of robotic platforms.
- Bernt Børnich – Founder and CEO of 1X Technologies
- The Neo Bipedal Humanoid & Market Launch: 1X is launching Neo, a consumer and commercial bipedal humanoid designed to generate an abundance of physical labor. Neo is offered through both upfront purchase options and a $500/month subscription model targeted at early adopters.
- Neo as an Open Platform & Skill Ecosystem: Beyond in-house capabilities, 1X is structuring Neo as an open hardware and software platform with an "App Store" model. Third-party developers and foundation AI labs can build custom manipulation skills, integrate external foundation models, and automate domain-specific workflows.
- The Data Pyramid & Cross-Embodiment Training: Solving embodied general intelligence requires training on massive data sets. 1X structures its data hierarchy into teleoperation (high quality, low volume), human sensor-suit capture, egocentric human video, and internet-scale general video. Because Neo matches human kinematics, skin compliance, and physical proportions, 1X can directly pre-train its robotic foundation models on the vast archive of human video across the web.
- Teleoperation as Remote Presence & Data Alignment: Remote teleoperation serves as both a data-gathering tool and a continuous utility for "expert-in-place" scenarios, allowing a specialist or executive to inhabit a robot anywhere globally to handle one-off, complex tasks.
- Timeline to the "Hard Takeoff": Børnich projects that robotics will reach a "hard takeoff"—where autonomous robots build other robots, data centers, chip fabs, and perform resource mining—in less than a decade, with an aggressive target estimate of just three to seven years.
- Amanda McMaster – Interim CEO of Boston Dynamics
- Transition to Commercial Scale & Spot's Footprint: Boston Dynamics has evolved from a pure research lab into a commercial deployment enterprise, with over 500 customers across 46 countries using Spot. Spot serves industrial facilities by performing acoustic imaging, vibration analysis, gauge reading, and thermal checks by day, while doubling as automated perimeter security at night.
- Commercial Models & Reliability Metrics: Spot operates primarily under a capital expenditure (CapEx) model ranging from $100,000 to $300,000 fully equipped, delivering return on investment in under two years. Spot achieves over 3,000 hours of mean time between human interventions, utilizing 90-minute operational cycles and autonomous docking stations.
- Atlas Humanoid & Autonomous Battery Swapping: The next-generation electric Atlas humanoid is slated for a "Robot-as-a-Service" (RaaS) deployment model to offer flexible fleet scaling. Unlike traditional charging downtime, Atlas features a dual-battery architecture where the robot autonomously swaps its own depleted battery pack to enable near 24/7 runtime.
- Dual-Brain Architecture: Boston Dynamics splits robot compute into two layers: an onboard real-time physical controller managing balance, locomotion, and fast dynamic reflexes, paired with a cloud-connected cognitive layer (leveraging partners like Google DeepMind) that manages high-level semantic reasoning and environmental understanding.
- Domestic Manufacturing & Anti-Weaponization: 100% of Boston Dynamics' robots are built in the United States without Chinese components to mitigate IP theft and critical infrastructure data leaks. The company maintains a strict ban on weaponizing its platforms, focusing government and defense engagements exclusively on non-lethal operations like Explosive Ordnance Disposal (EOD) and search-and-rescue.
- Professor Jonathan Hurst – Co-founder and Chief Robot Officer at Agility Robotics
- The Shift from Automation to Versatile Humanoids: Historically, robotics meant rigid, specialized factory automation. Humanoid robotics is now reaching an inflection point because vision and semantic perception have been largely commoditized by multimodal AI, allowing robots to perceive everyday human objects and navigate human-centric environments naturally.
- The Sim-to-Real Challenge & Physical Motor Data: Unlike text and image data, internet-scale datasets do not exist for precise motor torques and dynamic joint forces. Bridging the "sim-to-real gap" requires a combination of teleoperation, physics simulations, world models, and extensive physical trial-and-error in real-world environments.
- Digit V5 & Barrier-Free Human Collaboration: Agility Robotics developed Digit to handle multi-purpose logistics, tote movement, and pallet handling. The upcoming Digit V5 marks the first commercial humanoid certified to step out of caged work cells and operate alongside human warehouse workers without physical safety barriers.
- Economic Parity & Deployment Economics: Digit V5 is designed to operate 20 hours per day (due to high-speed charging) over a five-year lifespan, providing approximately 40,000 lifetime operational hours. This compresses the effective operating cost of physical labor to a few dollars per hour, creating strong ROI margins against human warehouse wages ($20–$40/hour).
- The Long-Term Role of Humanoids: Hurst notes that while specialized fixed machinery will run fully enclosed "lights-out" sortation hubs, versatile humanoids will manage the variable, unstructured tasks across legacy factories, retail aisles, hospitals, construction sites, and last-mile doorstep deliveries to liberate humans from the "3Ds" (dull, dirty, and dangerous work).
Comments