Watch the 19 min abridged version here
Shortage of talent - The entire presentation was a recruitment drive to hire AI engineers. If Tesla is having issues, take a look at your portfolio companies who leverage AI that isn’t just an off-the-shelf Amazon API. Imagine the problems they are having recruiting. The global war for talent is real, as an industry, we need to find ways to upskill our young people.
Full Self Driving (FSD) - is really hard. Elon has been promising Level 4/5 since 2016 but it’s still not here. I don’t blame him, it is just a really hard problem to solve (watch the video to have some sense of how hard it is to move from 2D to Vector space 3D modelling)
Tesla has had to build its own Exaflop supercomputer to train its models called Dojo and has to design its own 7nm chip. Their chip hasGPU-level compute with 2x the I/O bandwidth to handle huge amounts of data. Tesla is being forced to become full-stack not because they want to but because that is the only way to control this beast.
Wall Street’s best guess right now is Level by 2025, which is still a long time away.FSD is going to be a commodity
This is a controversial view because Tesla bulls will have you believe that this will an enduring moat. I don’t think so. As they “crack the code”, they may have an advantage for some years but over time this becomes commoditzed. In cars, we have seen everything from electric windows to auto-park being available only in high-end motors initially but quickly becomes mass market. In AI terms, everything from speech recognition to OCR was all supposed to be a moat until it wasn’t. Tesla knows this which is why they also said they will license their tech out once its available.
This is not an investible area IMHO and VCs will lose a lot of moneyRobots - “In the future physical work will be a choice” - Musk
To most, the robot is a distraction to a company that already has a lot of work to do. However, to me it’s just a recruitment tool. Give someone an audacious mission like going to Mars and they will help you launch a satellite. Give someone a mission to create sentient robots, they will help you create a car that makes a left turn without getting hitValuations will matter, but only eventually
Telsa 2021 revenue forecasts have only gone up by 15% since Jan 2018. That means that analysts were broadly right in their predictions. However, the stock is up 12x since then. Whilst valuations will matter eventually, some stocks can defy gravity for longer than shorts can be solvent.