Low-cost humanoid anchors
Unitree official pages list R1 AIR from US$4,900 / R1 from US$5,900, G1 from US$13.5K, and H2 at US$29,900, lowering the cost of experimentation.
Charlie memo · public-safe
Unitree / 宇树: low-cost access curve, not yet deployment economics
Unitree matters because it changes the experiment cost curve. The public evidence is strongest around price-access and developer platform reach, not around customer-confirmed industrial ROI.
Draft updated 2026-06-13. This page is a working surface for review before the final publishing flow.
🟢
Platform-enablement signal
H2 Plus references NVIDIA Jetson T5000, Isaac GR00T, Isaac TeleOp and Isaac Sim; G1-D adds data acquisition, labeling, model training and inference tools.
What it changes
The evidence supports cost-access and developer-platform expansion, not industrial deployment economics. Low price may broaden adoption or compress hardware value capture.
Missing S5 proof
No reviewed official disclosure yet of humanoid shipment scale, named industrial customer KPI, uptime, intervention rate, support cost, gross margin, repeat orders, or customer payback.
Core read
Unitree 的信号不是“已经证明工业 deployment economics”，而是低价 humanoid / quadruped platform 可能把开发者、实验室和早期客户的试错成本大幅降低。R1、G1、H2 的公开价格与 NVIDIA / Isaac / data workflow 叙事，说明它更像 cost-access curve。
What would upgrade it
需要看到具名工业客户的部署数量、任务 KPI、uptime、intervention rate、support cost、 repeat order、gross margin 或 customer payback。否则低价硬件更可能证明 adoption surface， 但不能直接证明价值捕获。