@katecrawford This massively fails on 1st principles.
Winning miles-share from ICE is net good from a climate perspective
Running EVs at 50% utilization (robotaxi) rather than 5% utilization (personally owned) is definitely net good
Energy cost of inference is ~0 relative to cost of motion
Winning miles-share from ICE is net good from a climate perspective
Running EVs at 50% utilization (robotaxi) rather than 5% utilization (personally owned) is definitely net good
Energy cost of inference is ~0 relative to cost of motion
@katecrawford There are 2 billion cars on the road, even if we went to 100% EV sales this year miles share to ICE would only reduce ~4%
Every robotaxi sold in will take 10x the number of miles from ICE
Every robotaxi sold in will take 10x the number of miles from ICE
@katecrawford Any concern about the energy impact of extracting the raw materials for EV batteries (on which the scholarship is actually quite bad) should gladly trade off compute energy to maximize utilization of those materials.
@katecrawford And the Tesla inference chip draws 72W.
The Model 3 drivetrain at average speeds draws 5,000W
The Model 3 drivetrain at average speeds draws 5,000W
Loading suggestions...