Boston Dynamics said Thursday morning that its Stretch warehouse robot passed 1,000 units deployed at customer sites during the second quarter of 2026. The disclosure came inside a routine hardware-refresh press briefing about Stretch v2, the successor unit scheduled to start shipping in Q4 with a new payload envelope and a longer battery duty cycle. The v2 news is fine. The 1,000-unit number is the interesting part, because it is roughly ten times larger than any humanoid maker’s deployed base and roughly twenty times larger than 1X, Figure, and Sanctuary combined.
Stretch is not a humanoid. It is a treaded mobile base with a single articulated arm and a suction end-effector, purpose-built for one task: unloading boxes from the back of a shipping trailer or the far end of a shipping container. Boston Dynamics designed the platform around exactly that task, sold it to logistics operators who were losing money and workers on manual trailer unloading, and shipped it in volume for three years. The customer list runs through DHL, Maersk, GXO, Nippon Express, XPO Logistics, and a rotating handful of large US retailers that Boston Dynamics does not name in briefings.
The 1,000-unit milestone matters because it is the largest deployed base of any purpose-built commercial robot outside of AMR fleet installations and the SCARA/6-axis industrial-arm base, both of which are decades-old categories with their own supply chains. Stretch got to a thousand units in three years. It got there without a humanoid form factor, without a foundation-model policy stack, and without the kind of teleop disclosure 1X had to publish this week for Neo. It got there because the task is bounded, the ROI is legible on a spreadsheet, and the failure mode is graceful. When Stretch cannot unload a box, it stops and waits for a human. It does not fall over.
The Clank read on the robotics conversation of 2026 is that humanoids get the coverage because they are photogenic, and the boring rectangular robots that ship boxes do the actual revenue. Boston Dynamics has known this longer than anyone in the category. Spot, the yellow dog, is the marketing product. Stretch is the actual business. Atlas, the humanoid, now in its all-electric form and running a foundation-model policy stack, is the R&D pipeline for whatever the next actual business will be. The order of ceremony is deliberate, and the fact that the biggest name in industrial robotics is running exactly this internal hierarchy is worth noticing before assuming that everyone else in humanoids has a shorter path to revenue than Boston Dynamics did.
The read for the humanoid category is that the labor-substitution argument still has to be made against a comparison set that already includes purpose-built robots that already work. A trailer-unloading humanoid has to be better than Stretch on the same task. Stretch is not standing still. A dishwasher-loading humanoid has to be better than a dishwasher, which sounds glib until you notice how much of the 1X Neo launch deck this week was about loading dishwashers. The humanoid pitch that everyone in the category is telling investors is some version of “one platform, every task, foundation-model policy stack.” The counter-pitch Boston Dynamics has been quietly making with revenue is “several platforms, each excellent at one task, deployed today, paying for the humanoid R&D.” Both can be right at once. Only one has 1,000 units in the field this quarter.
The follow-on question for every humanoid company with a warehouse pitch is which specific task on which specific customer’s trailer dock is going to justify a humanoid over Stretch v2, and how many hours per week it needs to run to hit ROI. That is the diligence question the next round of humanoid Series C decks is going to have to answer. Nobody has answered it yet in a form that survives contact with a logistics operations director. The people running the pilots know this. The people writing the pitch decks do not always know that the pilot operators know.