Two humanoid robot programs put dates on the calendar this week instead of another demo reel. Tesla confirmed that Optimus production has begun on lines in Fremont, California, according to Interesting Engineering. AGCO plans to deploy an Agility Robotics Digit humanoid at a Kansas factory, with the pilot expected to start in 2027, The Robot Report reported. That gives buyers two concrete dates to weigh against the rest of the category's announcements: one on the supply side, one on the demand side.
Tesla: production line, not prototype line
Interesting Engineering reports that Tesla has begun manufacturing Optimus units on dedicated production lines in Fremont. Elon Musk is targeting 20,000 Optimus units a week by 2027, the outlet reported.
TechCrunch Robotics reported that Tesla started Optimus production even as the company's profit fell 43%. That a capital-intensive robotics ramp is proceeding alongside a steep profit decline is worth tracking on its own: either Tesla sees the robot program as a growth engine worth the spend regardless, or the ramp is simply on schedule and indifferent to near-term financial pressure. Both readings matter for anyone judging how durable this production commitment is. Buyers can track specs and pricing for Optimus against rivals on the humanoid comparison page and the Optimus Gen 2 page.
A production line running is not the same as a sold, shipped, deployed unit. Tesla has not disclosed what the Fremont lines are producing toward: internal use, limited external pilots, or stock ahead of the 2027 target. Until Tesla publishes unit deliveries or named customer deployments, "production has started" is a manufacturing milestone, not a sales one.
AGCO: a named industrial pilot, further out
AGCO's plan is narrower but more concrete in a different way. The Robot Report says AGCO intends to deploy a Digit humanoid at a Kansas factory, with the pilot expected to begin in 2027. A named industrial customer committing to a factory pilot, with a date, is a different kind of signal from Tesla's volume target: it tells buyers what Digit will actually be asked to do and where, rather than how many units a vendor hopes to build. See the Digit spec page for the platform's current capabilities.
The two announcements answer different questions. Tesla's is a supply-side claim: it can build at volume. AGCO's is a demand-side claim: a real operator will put one to work on a factory floor. A platform that ships in volume with no named deployments is a different bet than one with a confirmed pilot but no stated production scale. Buyers deciding what to track ahead of 2027 should weigh both.
Compute is still the bottleneck, not the hardware
Production and deployment dates only matter if the robots can do useful work once they arrive. Here the news is incremental, not dramatic. Interesting Engineering reports that Bear Robotics cut the time needed to run a robotics foundation model, reducing model processing delay 8x in internal tests using Nvidia technology. Faster on-device inference is one of the practical constraints between a humanoid that can walk and wave and one that can react to a factory floor in real time. If that result holds up outside internal testing, it is relevant to anyone assessing deployment readiness generally.
Nvidia is pushing the infrastructure side too. At the company's GTC event in Berlin, Jensen Huang unveiled the DSX Pod, a plug-in data center, Wired reported, positioning it as key to unlocking autonomous humanoid robotics alongside Europe's broader AI infrastructure build-out. That is a compute story, not a robot story: it signals that the industry still treats the autonomy stack, not the hardware, as the bottleneck for deployments like AGCO's.
None of this is specific to Optimus or Digit. Both programs depend on the same underlying trend toward faster on-device processing and more available compute. It is context for the 2027 timeline, not a guarantee either program hits it.
Track the dates, not the marketing
Not every humanoid headline this week carries the same weight. Gizmodo reported that Unitree released a new advertisement for its robot, with the company saying to use the robot "safely and in a friendly manner." That is marketing, not a production or deployment milestone. It is a useful contrast: Tesla and AGCO's news this week comes with lines and pilots attached to dates, while a robot ad is the kind of content this category has produced for years without moving anyone closer to a buying decision.
For buyers and investors, the takeaways are narrow. Tesla has a production milestone and a 2027 volume target, not yet a named customer deployment. AGCO has a named 2027 deployment, not a disclosed unit count. Track both dates, not the marketing around them, and check current specs and pricing on the humanoid robot price tracker as more platforms confirm their own timelines.