It turns out that robotics, especially robotaxis, is a really hard AI problem, even as the latest advances with digital AIs like GPT-4 stun and amaze many close observers, like myself. Tesla has been trying in earnest for over six years to crack self-driving. Sadly, the latest versions of its “Full Self-Driving” software still require frequent human interventions in order to drive safely, comfortably, and politely. As much as I’d like to believe full autonomy is right around the corner, the rate of progress on the FSD software has been fairly linear and not the exponential trend I’d need to see to believe that robotaxis will happen anytime soon.
Without robotaxis, it’s hard for me to see much opportunity for growth beyond what is presumably already priced in to its valuation of over $750 billion. As boring as it is to say, our default assumption should be that the price of a stock is roughly “correct” and we should only overturn that assumption if we see a clear, compelling reason the stock is mispriced. I don’t see any particular upside potential or downside risk that the market isn’t already presumably pricing in.
What about the humanoid Optimus robot, Tesla’s other robotics project? I think it’s certainly an idea worth trying and it’s fun that Tesla has the capital to embark on such long-term and speculative R&D adventures. However, Optimus is even more ambitious than robotaxis and is at a much earlier stage of development. Only if Tesla succeeds with robotaxis should investors start modelling Optimus when they think about valuation.
It brings me no comfort to write these words. In fact, I write them with a heavy heart. As someone who is excited about frontier technologies, particularly AI, it is deeply disappointing to see the field of AI robotics run aground. I used to believe we were on the cusp of a wondrous robotic future. Now, I no longer do.
Tesla’s competitors like Waymo, Cruise, Zoox, and Mobileye are not in any better a position than Tesla with regard to self-driving cars. After years of R&D and billions invested, self-driving cars seem nearly as elusive today as they did a decade ago. The technology is stuck in a rut. We are not progressing as fast as we need to be to make self-driving cars a going concern.
One day, the research questions that need to be solved in order to commercialize full autonomy will be solved. Self-driving cars will eventually exist. But, for investors, eventually might as well be never.