Tesla's Optimus production is running well behind the company's own targets. Citing The Information, The Verge reports that Tesla built "several hundred robots a week" last month, far short of the 20,000-per-week goal the company has set for Optimus manufacturing. The Verge also reports that hitting that figure is proving difficult, and that Tesla has repurposed part of its Model S production line to build Optimus units. Agility Robotics, maker of the Digit humanoid, is answering a different question this week: not how many robots it can build, but what shape they should be. The two stories together are a useful check on how far humanoid robotics still has to go before the hardware matches the pitch.
Optimus production falls well short of target
Several hundred units a week is a pilot-line number, not a mass-manufacturing one. Production volume is the clearest test of how close a robot is to commercial scale, and the current number is nowhere near the 20,000-per-week target Tesla has set for itself. Retooling part of the Model S line to build Optimus is a reasonable stopgap at low volume, but it is also a sign that Optimus does not yet have dedicated manufacturing infrastructure built for the scale Tesla has talked about.
None of this means Optimus is stalled. It means the gap between a robot that can be demonstrated and a robot that ships at the stated volume is wider than the targets suggest. Anyone evaluating the Tesla Optimus Gen 2 program should weigh near-term delivery claims against what is actually coming off the line, not the production goal on a slide. The price tracker and humanoid spec sheet are useful references for comparing what is shipping against what is promised.
Agility Robotics hedges on the legged form
A separate report from The Robot Report points to a different kind of caution, this one about form factor rather than volume. Agility Robotics, maker of the Agility Robotics Digit humanoid, is reportedly exploring a range of robot designs, including wheeled robots, alongside its legged platform. The company frames this as a response to different operating environments and customer needs, not an abandonment of legs.
That is a notable move from a company built around bipedal locomotion. Legs cross stairs, curbs and uneven terrain that wheels cannot, which is the core case for legged humanoids in general and Digit in particular. But legs also add cost and complexity that a flat warehouse or distribution floor does not require. If Agility ships wheeled variants for those environments, it suggests the company sees flat-floor logistics work as better served by simpler hardware, at least for now, rather than one legged design for every job.
The software layer is hedging too
The Robot Report also notes that General Robotics is composing specialized capabilities that help robots adapt across tasks and form factors, rather than training one general-purpose model to do everything. Paired with Agility's hardware hedge, the direction is consistent: rather than one humanoid shape and one general-purpose brain handling every job, part of the industry is building toward fleets matched to the specific task and environment.
What this means for buyers
Two questions are worth keeping separate when evaluating a humanoid program. The first is delivery credibility: does the manufacturer's stated production capacity match what is actually coming off the line? Tesla's gap between 20,000-per-week and "several hundred" is the sharpest current example. The second is form-factor fit: is a legged, wheeled or hybrid design right for the deployment environment, rather than assuming legs are always the better choice.
Scale matters too. According to the IFR, 5 million robots are now at work in factories worldwide, with China leading installations and the U.S. and Germany also significant markets. That base is overwhelmingly conventional industrial automation, not humanoids, a reminder of how early the humanoid category still is by comparison. The programs worth watching are the ones hitting realistic production numbers and matching hardware to the task, not the ones with the biggest keynote claims. For side-by-side specs, see the humanoid comparison page.