On the road for full autonomy … and market dominance
For those that know me, they know I have no doubt that autonomous cars will eventually take over all cars on the road. In fact, I am pretty sure when level 5 autonomy is fully implemented it will signal a complete social revolution.
Experts will tell you it is always at least 20 years away… But a funny thing happened when Tesla started to roll-out its autonomy features… Those timetables started to tumble down!
Now even very conservative estimates predict that by the middle of next decade full level 5 will not only be possible as it will be the norm in all cars coming out of the factories.
Tesla (with Elon Musk at the helm) is, however, a completely different beast. And you should not bet against them.
Why?
Well because of this:
All Tesla Cars Being Produced Now Have Full Self-Driving Hardware
Tesla are approaching this in the way you would think google would:
if it does not scale it is not worth it.
You see… Google has been testing self-driving cars for over 7 years now. They have accumulated millions of miles and ginormous quantities of data. And yet, with every feature roll-out at Tesla, they can gather more miles driven in a few days then google did in all its lengthy research project.
Also having the hardware collecting data even whilst manual mode is active is done in order to convince legislators of the benefits of allowing self-driven cars . With every accident they log, they will be able to run simulations on how the self-driving mode would have behaved. If it had failed like the human they can feed the neutral net a new training set. If it did succeed where the human had failed they can add that to the body of evidence of how many accidents would have been avoided by a self-driving car. The same applies to every human intervention when the tesla is being self-driven. They can access the logs and discover why the human felt it necessary to intervene.
Moreover, with every new feature they add, they can re-test the system against that huge dataset and check for regressions in performance i. e. Compare the new version against the previous on all real world accidents/near calls in the database and find out edge cases where the previous version outperformed the new one.
In other words — putting big data to work effectively.
Outgoogling Google!
This is why Tesla is equipping all cars with the same hardware even if the purchaser does not buy the option for self-driving. This is why no old school car manufacturer will be able to compete with Tesla unless they change their mindset. Can you imagine Ford putting in thousands of dollars worth of equipment that the customer has not paid for?
No chance in hell!
But Tesla is not crazy. They know they will recoup that investment because even reluctant drivers that do not trust autonomous cars today will eventually be convinced and take advantage of the chance to make their cars fully competent autonomous drivers, available by the simple purchase of a quick software upgrade.
How much would you pay to be able to send your kids away driven on a chauffeur-driven car where you know that that driver will be always in its best performance, always careful, not drunk, not falling asleep and not breaking laws, speed limits, etc?
Is 8k a price worth paying to keep your family safe?
Tesla is betting it is… I would not bet against them.
PS: Please note that the 8K will also enable your car to become a part of Tesla’s “uberization” allowing your car to pickup and drop of paying customers whilst you are at work or on vacations and that will more than offset the initial price of the software upgrade. In fact, according to Tesla Masterplan Part Deux, this might even “at times potentially exceeding the monthly loan or lease cost of the car”
