Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

What makes you think Tesla can't make Optimus?


They can't even deliver the self-driving car they keep promising; why should we believe that they have some special expertise in automation which will allow them to build better robots than anyone else?


Robotaxi is rolled out and we are not seeing crazy videos like we do with Waymo (who admittedly is doing more miles).

What is exactly with their AI that does not satisfy your criteria? xAI seems to be SOTA, IIRC unbeated in coding performance for whole half year now or so.


> Robotaxi is rolled out and we are not seeing crazy videos like we do with Waymo (who admittedly is doing more miles).

Robotaxi is two orders of magnitude worse than Waymo: https://arstechnica.com/cars/2025/09/teslas-robotaxi-test-th...

> xAI seems to be SOTA, IIRC unbeated in coding performance for whole half year now or so

(1) Not according to any of the listings I've seen. I mean, don't get me wrong, they're occasionally at the top, but nobody is persistently at the top for more than about 6 weeks.

(2) xAI is not Tesla, and Musk has faced legal action because his work on xAI took talent and resources away from Tesla: https://www.theverge.com/2024/6/13/24177975/tesla-shareholde...


The Tesla Optimus is, unlike Optimus Prime, smaller than a car. This means it has less room for battery and compute. The gap between the cars and the robots suggests that Koomey's law[0] will still require 5-10 years to go from compute that can run in and operate a self-driving car (by any constant standard[1]) and compute that can run in the power envelope available to a humanoid robot which can get into any old car and drive it equally well.

And that's just the ability to drive a car. To be what Musk is selling Optimus as, it would have to be much more capable than simply driving a car.

--

None of this precludes him selling Optimus robots as remote control units, where the AI is only the bare minimum for walking without falling over. This would already be transformative to a lot of industries, and if such robots would have a TCO in factory conditions of, say, $24k/year, could easily lead to a lot of American factories being filled with such robots remotely controlled by people in other countries very happy to get paid $12k/year, and a whole bunch of very confused and also very unemployed Americans.

I suspect a lot of people will feel cheated if this is what happens… but also, I have not seen any specific minimum requirement for what is actually shipped for this payout. Musk being the troll he is, I wouldn't put it past him to buy out Hasbro, ship a million of these, and claim victory: https://www.amazon.com/Optimus-Auto-Converting-Transformers-...

[0] https://en.wikipedia.org/wiki/Koomey%27s_law

[1] https://en.wikipedia.org/wiki/Self-driving_car#Definitions




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: