Industry Brief, September 9, 2026: Humanoid price tags, robot setup time, FCC fallout
This is a retrospective edition. The brief did not run on September 9, 2026; we wrote it on September 23, 2026 using only reporting that was published on or shortly before September 9, 2026, and it makes no forecasts.
This is a retrospective edition covering September 9, 2026, using only reporting published on or up to two days before that date. Today: a humanoid maker puts illustrative prices in a securities filing, a flexible-automation vendor opens a lab to turn its installed base into training data, a software startup claims robot onboarding in hours rather than weeks, and a European warehouse robot maker spells out who the new FCC rules really hit. Our own words, links to the original outlet, vendor numbers labelled as vendor numbers.
Agility’s filing puts a number on a humanoid: about $400,000 over five years
Business Insider (read here in its Yahoo Finance UK syndication) went through a securities filing Agility Robotics made on Tuesday. Under one business model in the filing, a customer buying the next-generation Digit v5 would pay about $200,000 for the robot, a roughly $20,000 one-time deployment fee, and then about $36,000 a year for software and maintenance.
Over the five-year useful life Agility assumes for its model, that comes to roughly $400,000 ($200,000 + $20,000 + 5 × $36,000). The company calls the figures illustrative: contracts are negotiated, and the five years is a modelling assumption, not a statement that the robot stops working. The filing also says Digit v5 is expected to reach customer breakeven in about 1.1 years, citing productivity, uptime and worker availability, but Agility did not show the calculation and did not respond to the outlet’s request for comment.
Other company figures in the piece: the robot is designed to work about 20 hours in every 24 and lift up to 50 lb (about 22.7 kg). The current Digit v4 is deployed at nine customer facilities. For contrast, the article notes Unitree advertises its G1 humanoid from $13,500, though those robots are mostly used for research.
Why it matters: This is one of the few humanoid price lists in public, and its structure is more useful than its total. Roughly 45% of the five-year figure ($180,000 of $400,000) is recurring software and maintenance, which is exactly the kind of line that gets left out of a first budget. Treat the 1.1-year breakeven as a vendor planning assumption and ask what hourly labour value, utilisation and shift pattern it assumes before comparing it with your own payback math or with a cobot’s all-in cost.
Source: Business Insider, via Yahoo Finance UK, September 9, 2026.
Vention opens a physical AI lab fed by its own installed base
The Robot Report covers the opening of Vention’s Physical AI Lab in Montreal, focused on robotic manipulation in manufacturing. CEO Etienne Lacroix’s argument is about data: the company says it moves several hundred robot cells a year and that they all collect industrial manipulation data it had not properly used until now. Vention’s own figures: more than 28,000 machines deployed, a community of more than 6,000 factories, 90 of the Fortune 500 as customers, and physical-AI revenue up 400% over the past year.
The lab is led by Dr. Jimmy Li, a McGill robotics and machine learning researcher, with Cohere chief AI officer Dr. Joelle Pineau as external advisor. Vention launched a modular pipeline called GRIIP in February 2026 (scene digitalization, segmentation, pose estimation, grasp selection, motion planning) and plans a public SDK; Lacroix said it will be open-sourced and shown at IMTS. He named kitting as a top use case, and the company says it is working with a large automotive OEM on final-assembly tasks.
Why it matters: Lacroix’s own framing is the right test: the hard part is no longer proving a robot can do a task in a lab, it is making it reliable and economical across many factories. Vention says it expects foundation models to cut deployment cost and complexity, but that is an expectation, not a result. If a vendor pitches AI-driven picking or kitting, ask for reliability data from production cells, not demos, and check who owns the data your cell generates. Kitting is a sensible candidate on any what to automate first list because the task is repetitive and easy to measure.
Source: The Robot Report, September 9, 2026.
General Robotics says its software cuts robot onboarding from a month to hours
GeekWire reports that General Robotics, a Redmond, Wash., startup founded by former Microsoft researchers, says advances in its GRID platform have cut onboarding a new robot from about a month to as little as two hours. It calls the approach “Auto Engineering”: each onboarded robot and diagnosed failure feeds back into the system. CEO Ashish Kapoor said the old model needed a team of experts working on the customer’s behalf, which he described as not scalable and expensive.
The company was founded in 2023, has about 50 staff, and has raised nearly $34 million. Kapoor said it has roughly a dozen customers in manufacturing, logistics, energy and defense, with revenue in the millions of dollars. The platform works with arms, humanoids, quadrupeds, wheeled robots and drones, and has NVIDIA’s Isaac Sim built in. Kapoor’s view is that robot makers build good hardware but often lack the expertise to put it to work in a specific site.
Why it matters: Setup and integration engineering can be a large share of a small cell’s cost, so a real month-to-hours reduction would matter. But this is the company’s claim, with no independent test and no detail on which robots or tasks the two-hour figure applies to. Ask any vendor making this kind of claim what “onboarded” means (first motion, or production rate at target quality), then compare it with the integrator hours you would otherwise pay for.
Source: GeekWire, September 9, 2026.
Exotec: the FCC rules hit new robots, and origin alone is no exemption
Robotics & Automation News carries a statement from Arthur Bellamy, chief revenue officer of French warehouse robot maker Exotec, on the FCC’s tightened rules for foreign-made robots. His reading: the rules mainly affect new foreign-produced mobile robots that need a new FCC authorisation, while already-authorised systems can keep operating. European suppliers, he says, should not assume their origin exempts them; the outcome depends on the product, where it is made, and whether it already holds US authorisation.
He points to the FCC’s earlier drone measures, where conditional approvals went to companies such as Mobilicom and AIR6 Systems but none to Chinese firms. He also argues the US rules could make Europe relatively more attractive to Chinese suppliers, while saying it is too early to predict a big rise in competition. The outlet notes Europe has no blanket ban on foreign-made warehouse robots.
Why it matters: This is a competitor’s view, so read it as such, but the practical point stands for US buyers: the question is not only who sells you a system but where its robots were made and whether they are already authorised. Before signing for new mobile robots, ask the supplier or integrator for the authorisation status and country of manufacture of each robot model in the quote. It belongs on your integrator question list.
Source: Robotics & Automation News, September 9, 2026.
Sources
- Want a humanoid in your factory? Here's how much one could cost you. (Business Insider (via Yahoo Finance UK), 2026-09-09)
- Vention opens Physical AI Lab for manufacturing in Montreal (The Robot Report, 2026-09-09)
- General Robotics, led by Microsoft vets, says its AI has cut robot setup from a month to hours (GeekWire, 2026-09-09)
- US robotics restrictions could drive more Chinese suppliers toward Europe, warns Exotec (Robotics & Automation News, 2026-09-09)