Tesla ran a hidden second floor making its own parts to ship 50 changes a week

Factory tours of Tesla's Fremont plant showed the presses and robots on the ground floor — but the 'secret' second floor, unknown to most visitors, built many of the battery, power electronics and drivetrain systems, with robots so fast their arms were carbon-fibre rather than steel. Greg Reichow, Tesla's VP of production during the Roadster-to-Model S and X ramp, says the goal was never just the best EV but the best premium car that happened to be one, which meant integrating technologies that didn't exist as supplier parts.

The first advantage was speed: outsourced components require resident engineers in someone else's factory and thin decision-making on the spot, while in-house manufacturing let the whole team improve continuously — up to 50 changes shipped per week, at the cost of some manufacturing efficiency. The second was depth: building its own safety fuse and switch system is what made 'Ludicrous Mode' possible, solving the physics problem of letting massive current flow normally while cutting it off in milliseconds during a spike.

Reichow's framing was a layered capacity strategy, not universal in-sourcing: start with your own pilot line beside the engineers to learn and iterate fast, then layer contract manufacturing on top for scale. Commodity parts that won't change significantly should stay bought — build only what carries unique intellectual property or changes weekly.

Suppliers couldn't match Tesla's pace: the accelerated timeline was the binding constraint.

Co-locating production with engineering turned factory feedback into same-day design changes.

In-house fuse development solved a current-and-safety problem no off-the-shelf part could, enabling Ludicrous Mode.

Owning a pilot line keeps contract manufacturers honest about change costs.

Vertically integrate where change velocity is high: the pain of building it yourself is the tuition for product knowledge you can't buy from a supplier.

The approach carried Tesla from low-volume Roadster production to fully integrated Model S and X manufacturing, with Reichow arguing the pattern — own the fast-changing core, buy the commodity rest — as a general rule for hardware companies.