If Tesla were to make billions of bots, I think the production cost and selling price would probably fall far below $25k. Huge if, of course.
For comparison, an entire car can be built today for around $25k.
The bot will have one or two orders of magnitude less parts, mass, and build operations than a car. The bot’s mass is ~70kg, which is ~30x less than a Model Y. We don’t know the part count, but it should be much less than the multi-thousand part count for a car. This implies much lower bill of materials cost and indirect costs thereof, including supply chain, logistics, floor space, and fabrication/assembly operations.
Also, if the bot eventually becomes capable enough for there to be demand for billions of units, then the army of bots itself will help build more bots with very low labor cost in an exponential self-replication cycle of growth. Assuming Tesla did not fake their Investor Day demo video, they are apparently already working on this.
At production scale in the billions, economies of scale and Wright’s law would reduce costs as well. Fixed costs would become a tiny fraction of the overall cost structure and learning curve effects would result in extremely optimized design and manufacturing techniques, just like we see for other products with similarly huge scale, like rice, computer chips, or aluminum cans.
This leads me to believe the cost has the potential to fall to a few thousand dollars per bot. If so, Tesla probably would not sell the bot for as high as $25k unless they’re targeting like 80% gross margin. It’s hard to really model though. If this thing actually works as hoped then the entire world is going to get weird really fast.