My primary thesis for investing in Tesla is that it’s a play on the falling cost of batteries. But it’s also a play on artificial intelligence. Last week Tesla made a big bet on AI by adding hardware they believe can support fully autonomous driving. But the bet goes beyond that, the play on AI is why they won’t be needing to raise a lot of capital.
Recently Tesla has indicated that it’s designing cars to be easier to be put together by robots. Today on the conference call Elon Musk discussed what is code named “alien dreadnaught.” The idea is that general assembly still requires a lot of manual workers instead of robots. From what I can tell, what Elon Musk calls alien dreadnaught 1.0, and is predicted to be completed in Summer of 2018 is when Teslas will be built by robots and will not be touched by a human. This is a game changer, it will lower labor costs dramatically.
When asked about raising capital Ron Baron asserted that Tesla has gone “capital light.” On the Q3'16 conference call Jason Wheeler stated that Tesla is entering a new paradigm. One where step change in capacity, doesn’t equal a step change in capital. They have clearly signaled they are working to improve capital efficiency. It’s not just that robots are cheaper than manual workers, robots are faster. Which leads into Elon Musk recent statements about improving volumetric efficiency, and efforts on the machine that builds the machine.
I think the general public and Wall Street analysts are skeptical. However, AI researchers are probably much less skeptical because there have been significant advances in computer vision in recent years. Specifically deep learning, which is a sub-field of artificial intelligence, is showing particular promise. For instance, Microsoft recently used deep learning to achieve human level speech recognition. Google is in the process of changing all their translations services to use deep learning. But more applicable to robotic manufacturing I point you towards ImageNet, an academic computer vision competition where computers are exceeding humans at object identification. Also, check out this blog post by Google about robotic grasping or video below. This task is directly related to computer vision and more flexible robotic tasks such as installing interior items. Elon Musk has made a side bet on AI with his non-profit OpenAI which has a goal of building a household robot using deep learning. Deep learning is the reason Elon Musk refers to self driving vehicles as a “solved problem” and has allowed pre-orders for self driving. The overall impression of these announcements is deep learning is exceeding human capabilities and it’s moving from research into production products.
It has been asserted by Clay Christiansen that the automobile didn’t disrupt the horse and carriage, but rather the assembly line that disrupted the horse and carriage. Similarly, it will be the cost reductions in electric vehicles from economies of scale and deep learning that will disrupt the auto industry.
Check out Tesla news feed
