Tesla uses Optimus internally
Sanctuary AI
Tesla can treat Optimus like an internal factory tool before it ever has to be a sellable product. That changes the early game from enterprise selling to learning by doing. If Tesla puts robots on its own lines in Fremont and later Texas, every shift becomes training data, every failure becomes an engineering fix, and every extra unit produced lowers cost through the same manufacturing system Tesla already uses for cars, batteries, and power electronics.
-
Tesla is already organizing around in house deployment and scale. Its Fremont factory is described as a hub for Optimus production, Tesla has factory roles tied to Optimus manufacturing and production support, and it has said its first large scale Optimus factory line is planned for Fremont with a 1 million robot per year design target.
-
That is a different burden from Figure, Agility, or Sanctuary. Figure has raised about $1.75B at a $39B valuation and Agility about $641M at a $2.12B valuation, but both still need outside customers and reference deployments to prove ROI. Tesla can first prove ROI to itself by replacing or augmenting labor inside its own plants.
-
The data loop is the real moat. Tesla is hiring Optimus policy engineers to build datasets from humans, bots, and simulation, and robotics integration roles that support Optimus across factories and warehouses. That means deployment is not only about labor savings, it is also about generating the messy real world manipulation data humanoids need to improve.
If this model works, the humanoid market will likely look more like automotive manufacturing than enterprise software. The winners will be the companies that can finance long ramps, run robots in their own operations, and turn internal usage into lower cost and better autonomy before broad commercial rollout.