Tesla’s Fully Agile Rapid Innovation

In 2008, Nextbigfuture wrote how a massive technological transformation was possible even without greater than human-level general artificial intelligence or molecular nanotechnology. I proposed that a mundane technological singularity without super-technology could transform the world. A lot of systems and processes would have to be redesigned. The mundane singularity could be 100 to 1000 times faster in terms of production and various capabilities. However, it now turns out that Elon Musk and Tesla will be able to deliver 100 times faster factories or production by applying Agile methodologies to hardware manufacturing.

Agile has the shortest (fastest) responsible iteration cycle. You iterate once and then replan. So you so the best thing this week and then replan for next week.

Tesla has the goal of walk-up to-do lists. A list of to-do items that should be understandable when you walk up to it. Self-organized teams then work on the tests.

There are machine learning stacks that measure and determine if any contribution was a net benefit to the company. There is an automated measurement of all contributions.

There is also full testing suites for every car by setting it to factory mode. The full automated testing for every car frees people to make changes and innovation. This is because full testing is low-cost. Traditional manufacturing has a partial testing suite once every thousand cars and full testing only after the annual design of a new model. The automated test suite has very high standards. Tesla is willing to innovate on the chassis and every part of the car. This means the full automated tests will ensure that the build quality will stay world-class but it could shift from 10 times better to just world-class from car to car.

In 2016, Elon talked about changing factories every two years to get ten times better productivity in ten years. Now, Tesla is iterating and making changes 27 times a week.

There is evidence that Tesla Shanghai already has 1.1 million cars per year peak production capacity. There is another $188 million expansion that should be completed by April 2022. Tesla Shanghai had 500,000 planned cars per year capacity in 2018.

Tesla has added in Gigapresses to cast the entire front and rear segments of the car. These changes save 300 robots, a thousand parts and 30% of the factory space.

In 2020, Elon Musk said optimizing the layout and operations of a factory can increase the speed by at least ten times and up to 100 times.

In practice, there are limitations revamping production lines without building an entirely new factory (like Berlin and Austin) or stopping production and rebuilding part of a facility.

Tesla has dozens of production cells making every part of the car. Each is acceptable quality but they are iterating and trying to make better versions of the part and try to ramp up more and more production from each cell.

Tesla is adding new technology like dry cell batteries and gigapresses and new alloys to enable radical factory and production speed improvements.

The 4680 batteries and front and rear casting will be fully implemented in 2023. This will enable faster production from Tesla Shanghai and the other factories. Tesla Shanghai should reach 2 million cars per year by the end of 2022. This should be further improved to 4 million cars per year by 2024-2025.

SOURCES- Best in Tesla, Tesla Daily, Joe Justice
Written by Brian Wang, Nextbigfuture.com

4 thoughts on “Tesla’s Fully Agile Rapid Innovation”

  1. There's the whole full digital model of the car using modern systems, which is a Tesla advantage, coupled with their own versioning. They were very agile in use of microcontrollers such that they often swiftly moved between different IC's/PCB's as the semiconductor shortage developed over the last two years. They don't lock down the design to a model year. Because EV's are electric and software driven machines, the fungibility of the parts is a huge advantage.

    Reply
  2. Interesting, the "production cells" sound like they are the equivalent of parallel production lines. Upgrading one at a time is more expensive than the traditional shut down and retool model, it increases logistical complexity at least in terms of needed parts if the production values vary significantly between the cells. But, yeah, this sounds like another good approach.

    Reply

Leave a Comment