Industry Brief, September 5, 2026: Defense plant robots, August jobs, robot data

By MillBrief Editorial · September 5, 2026 · Retrospective edition, published September 23, 2026

This is a retrospective edition. The brief did not run on September 5, 2026; we wrote it on September 23, 2026 using only reporting that was published on or shortly before September 5, 2026, and it makes no forecasts.

This is a retrospective edition covering September 5, 2026, using only reporting published on or up to two days before that date. A Saturday edition with three items: public money for robots in the military’s own plants, a modest but steady month for factory hiring, and a sign of how much investors now pay for the data robots learn from. Our own words, links to the original outlet, vendor numbers labelled as vendor numbers.

ARM Institute wins nearly $90 million for 10 projects in military plants

The Robot Report reports that the Advanced Robotics for Manufacturing (ARM) Institute, the Pittsburgh-based consortium, has received nearly $90 million for 10 projects led by the institute and its members. The money comes through the Organic Industrial Base (OIB) Modernization Challenge run by the Office of the Under Secretary of War’s Manufacturing Technology Office. The OIB is the network of arsenals, depots and ammunition plants that build and maintain US military equipment.

Fifteen member organisations are to deliver working solutions within two years at 12 military manufacturing sites. The selected projects fall into three areas: safer energetics production and better quality using AI, drones to speed mission readiness, and robotics to remove bottlenecks. Each plan includes a workforce component, led by the institute with its workforce partners, to identify the skills needed to run the new systems and to upskill or recruit where necessary.

Why it matters: This is a different story from the institute’s call for concept papers that we covered in August: that was the invitation, and these are funded projects. The pairing of each project with a workforce plan is the part worth copying in a small shop: budget for training and ownership of the system, not just the hardware, as our list of hidden costs of automation explains.

Source: The Robot Report, September 5, 2026.

Factories add 16,000 jobs in August, and July is revised up

Manufacturing Dive reports Bureau of Labor Statistics data released September 4: manufacturing added 16,000 jobs in August, bringing sector employment to nearly 12.64 million, against 12.62 million a year earlier. July was revised from an initial 5,000 jobs to 14,000. Machinery manufacturing added the most workers (6,100), while motor vehicles and parts lost 4,500. Across all industries, 162,000 jobs were added.

The article puts unemployed manufacturing workers at 497,000, a rate of 3.3%, compared with 555,000 in August 2025. Separate BLS turnover data showed July manufacturing job openings up about 36% from a year earlier to 580,000, with 180,000 quits and 88,000 layoffs and discharges. Scott Paul of the Alliance for American Manufacturing said factory employment is up 58,000 since the December jobs report, while naming interest rates, China policy and energy price shocks as possible roadblocks. The Institute for Supply Management’s employment index eased to 51.2% in August; ISM says a reading above 50.3% over time is generally consistent with rising BLS manufacturing employment.

Why it matters: For a small manufacturer, the number that bites is 580,000 open factory jobs (July) against 497,000 unemployed factory workers (August): the months differ, but openings outnumber people looking. Where hiring for a repetitive station keeps failing, that station is a candidate for automation; our guide to what to automate first covers how to pick it, and the payback guide shows how to value hours you cannot staff, not just hours you pay for.

Source: Manufacturing Dive, September 4, 2026.

Robot-data startup XDOF is in talks at about a $1.2 billion valuation

TechCrunch reports, citing several people with knowledge of the deal, that XDOF, which collects real-world teleoperation data for training general-purpose robots, is in late-stage talks for a Series B led by 8VC at a valuation of about $1.2 billion. The terms are not final, and XDOF and 8VC did not respond to the outlet. TechCrunch could not learn how much is being raised or whether the valuation includes the new money. The people said annualized revenue is approaching $50 million. The company raised a $70 million Series A in June.

XDOF was founded in 2024 by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO), who earlier built GELLO, a low-cost system for controlling a robot arm remotely to generate training data. It combines remote teleoperation with human collectors wearing sensors while doing tasks such as folding clothes and flattening boxes, and it is working with Berkeley’s AI Research lab on a large training dataset called ABC. It plans to hire and train teams of data collectors worldwide, both teleoperators and people wearing body sensors. The company has previously said it works with 20 customers, including several frontier AI labs.

Why it matters: You will not buy from XDOF, but the price investors put on robot training data tells you where the bottleneck is: robots that learn tasks from demonstrations are still short of demonstrations. When a vendor pitches a learning-based cell, ask whose data trained it, how much of it resembles your parts and lighting, and whether collecting your own demonstrations is part of the job. It pairs well with our engineers’ notes on what they wish buyers knew.

Source: TechCrunch, September 4, 2026.