Tesla FSD V14 & Robotics
Today is the day, my tesla finally got pushed the long awaited V14 upgrade. It’s downloading as I type this. There’s a lot of hype about this upgrade. Remember Elon is betting the farm on vision where as Waymo and other competitors continue to focus on Vision + Lidar + other sensors. This is a huge bet with huge outcomes. What tesla lacks in sensors they plan to make up with compute and software (Ai).
Speaking of Ai, feels like Tesla is way ahead of the competition when it comes to real world Ai. Elon recently said that these cars are just robots which look like cars. I agree.
Robotics is also past the inflection point, we are no longer building specialized robots to do a task. Tesla and many competitors are focus on generalized robots like Optimus. These robots need more than just hardware, they need the full suite of things hardware, software (Ai) and energy. Feels like hardware and software will get solved, as always, but the limiting constraint is going to be the energy.
Quick search on Grok, reveals that right now we have a 12 GwH energy gap between what the US produces and what the Hyperscalers have committed to. Where is this energy going to come from? I recently saw an article that CrusoeAi and others are looking to space to build extra compute. They can have 24/7 solar energy which powers their chips in space. Distribute the compute and the push it back down for inference etc. I asked Grok if I wanted to generate 1 GW of energy in space how large the solar array has to be? At 35% conversion, we need 2.21 km² solary array. That seems large but not impossible to build. Especially when you look at the regulations associated with building nuclear at home.
Anyhow, clearly there are plenty of people who are looking into this issue. In the meantime, we are all going to face higher electricity costs. Those who have solar are gonna benefit until this energy constraint is removed.
Happy Sunday