Interview
What iPhone Packaging Reveals About the Robotics Opportunity
- Anticipates continued heavy investment in industrial capacity by both U.S. and Chinese companies, with the market expected to mature from a "cottage industry" of custom machines toward general-purpose form factors like humanoids before eventually re-specializing.
- Predicts that most current customers are purchasing robots for the first time, with 99% of existing market pitches failing to address end-use applications, while expecting the 38,000 industrial robot units deployed in the U.S. in 2025 to be dwarfed by future demand driven by generally capable robots.
- Plans to manufacture robots at scale in Watertown, Massachusetts, verticalize component production in-house over the next few years, and eventually shift supply chain reliance from OEMs to in-house manufacturing to ensure a mature U.S. robotic supply chain.
- Aims to train frontier AI models and deploy a network of robots connected to human tele-operators ("tutors") to teach new tasks, creating a data flywheel from economically valuable work performed in real customer sites.
- Targets the deployment of "Sunny" and "Cassie" robots to perform 80% to 90% of manufacturing and logistics work, with expected reliability for three to ten years, while acknowledging current hardware limitations regarding off-industrial terrain like gravel.
- Foresees the industry cost structure shifting from high upfront capital expenditure to ongoing operational costs for inference and teleoperation as the business scales, enabling a "great equalizer" effect for U.S. competitiveness.
- Expects to hire researchers and engineers across all technical stack layers, including a team of 100 Sunny robots in Boston to serve as a data collection farm for real-world reinforcement learning.
- Projects a sales cycle averaging 40 to 60 days from first contact to signed contract, followed by approximately 40 days to get robots working on-site, with production stand-up occurring within hours of arrival.
- Plans to ship code updates daily or in major releases every two weeks, iterating through two robot versions to avoid being stuck on V1, while aiming to save customers $100,000 to $100 million annually within two months without requiring initial customer capex.
- Identifies the average age of manual labor workers as 55, with a high turnover rate and imminent retirement creating immediate demand for replacement labor, while noting that home robot deployment remains a significantly tougher technical and economic challenge than factory settings.
- Envisions expanding into construction and landfill material reclaiming with legged robots as part of a diverse "pantheon of embodiments," while maintaining a strategy of avoiding custom "Rube Goldberg machines" in favor of scalable, general-purpose substrates.
- Maintains a neutral stance on market entry points despite acknowledging technical difficulties in home deployment, relying on trade-offs across stack layers initially that will decrease as the platform matures, with the expectation that the global supply chain remains highly interconnected in 2026.