Tesla has fallen behind BYD in terms of vehicle sales. Not to worry because Tesla is a AI & Robotics company
In 2025 calendar year BYD overtook Tesla, in terms of the most numbers of EVs sold worldwide.
Tesla delivered 1.63 million vehicles globally in 2025, a 9% fall from 1.79 million in 2024
BYD, on the other hand, delivered 2.26 million EVs in 2025
The elimination of the $7,500 U.S. federal tax incentive that seems to have played a big role in the decline of EV sales in the U.S. in the fourth quarter, which obviously affected Tesla numbers negatively.
…..
Now, Tesla bulls justify this by saying that Tesla is more than an EV company. It’s transitioning to an AI and robotics company
But the bulk of Tesla’s income comes from its EV business. For instance, Tesla generated $28 billion in revenue in the third quarter, of which $21.2 billion came from selling EVs.
Again, I understand that this is simply a snapshot of the present, not what’s to come in the future.
…..
A logic that doesn’t make sense to me
Tesla has a data advantage for self-driving cars. But they do not have this advantage for Robotics. But the entire thesis/argument is about Tesla transitioning to AI and Robotics, but how are they expected to crack Robotics if they do not have the data advantage for Robotics i.e. they do not have any advantage over other Robot companies.
Now, if you say that, Robots can be trained in virtual/ simulated worlds, then my response is “If robots can be trained on simulated/ virtual worlds, then self-driving can also be trained in the same manner”. So, Tesla’s data advantage is not a good thesis.
So, either:
Tesla has a data advantage for self-driving car, in which case Tesla does not have a data advantage for humanoid robots (unless they have been collecting humanoid robot centric data for the last decade unknown to public knowledge). This means that Tesla will dominate autonomous driving, but there will be aggressive competition for autonomous humanoid robots, with no guarantee that Tesla’s Optimus will come out on top.
OR
Humanoid robots can be trained in simulated virtual worlds, in which case self-driving cars can also be trained in a similar manner in theory. In this case Tesla does not have the data advantage.
..
Personally, I lean into option 1, based on my knowledge of machine learning works.
SOURCES and FURTHER EXPLORATION:
Global EV sales reach 20.7 million units in 2025, growing by 20% - Rho Motion
Tesla Fourth Quarter 2025 Production, Deliveries & Deployments
ADDENDUM
1. EV Sales Data in 2025
Global: 20.7 million, +20%
China: 12.9 million, +17% (Around 63% of the total EV’s were sold in China)
Europe: 4.3 million, +33%
North America: 1.8 million, -4%
Rest of World: 1.7 million, +48%
2. Tesla launched fully autonomous ride hailing services in Austin
Tesla has recently launched fully autonomous (unsupervised) robotaxi rides in Austin, Texas, without a human safety monitor physically inside the vehicle. This is part of a limited pilot program that began with safety drivers in June 2025 and expanded to public, driverless rides in January 2026 (Tesla launches robotaxi rides without safety monitors in Austin - Automotive News)