Elon Musk has shed some light on what he sees as the “only path to success” for broad full self-driving capability. His thoughts on Twitter came as a response to Whole Mars Blog, who has been speaking out against some of the smear campaigns put forth by some of Tesla’s competitors.
They have no idea how hard FSD is. Only path to success imo is hardcore real-world AI software with dedicated NN inference acceleration ASICs in car, multibillion dollar NN training supercluster and 10+ billion miles of vehicle data. Good luck.
— Elon Musk (@elonmusk) March 16, 2022
Green Hills Software CEO Dan O’Dowd, one such competitor, has been very vocal in his criticism of Elon Musk, Tesla’s FSD software, and Tesla’s achievements in artificial intelligence. It’s this competitor that Whole Mars is referring to in his initial tweets above.
You may have seen a recent full-page ad in the New York Times from Green Hills Software criticizing Tesla FSD. The software CEO tweeted earlier this year that he placed a full-page ad in the New York Times in hopes of banning Tesla’s FSD vehicles from the roads. What the software CEO neglected to mention was that in 2021 his company announced a partnership with Apex.AI.
I think it's only fair and honest for you to disclose which Tesla competitors your company Green Hills Software have taken money from.
— Dave Lee (@heydave7) January 16, 2022
The ad in the New York Times called Tesla’s FSD software the “worst software ever sold by a Fortune 500 company.” The ad focused on The Dawn Project, which isn’t Green Hills Software but a separate project.
In his initial posts, Whole Mars noted that Green Hills Software announced that BMW iX would use its tech, but in reality it appears BMW is not using it.
Suddenly it makes sense why Dan O’Dowd is getting so upset and funding a smear campaign, doesn’t it?
They’re losing customers left and right as OEMs realize they need to step up their game to compete with Tesla. The same broken tech won’t get them where they’re going.
— Whole Mars Catalog (@WholeMarsBlog) March 15, 2022
Whole Mars added his thoughts on what is inspiring the Tesla attacks: “They’re losing customers left and right as OEMs realize they need to step up their game to compete with Tesla. The same broken tech won’t get them where they’re going.”
In response, Elon Musk said, “They have no idea how hard FSD is. Only path to success imo is hardcore real-world AI software with dedicated NN inference acceleration ASICs in car, multibillion dollar NN training supercluster and 10+ billion miles of vehicle data. Good luck.”
Tesla enthusiast and blogger Gail Alfar wanted to know what Elon meant, and in particular, what ASICs were. ASIC is an acronym for Application-Specific Integrated Circuits, or computer chips that are designed to do only one thing. Many in the Tesla community had excellent answers regarding this, and the thread opened up further conversation about Tesla’s AI.
I love going through these explanations, and talking to friends about this. I’m learning – thanks to you all!
— Gail Alfar (Austin) (@GailAlfarATX) March 16, 2022
What Elon was saying was that the only path that he can see to success is hardcore, real-world AI software that has dedicated neural network inference acceleration ASIC chips in the car along with a multi-billion-dollar neural network training supercluster and over 10 billion miles of vehicle data. Tesla has all of these. Its competitors do not. We will see who is right in the coming years.
For those like me who are constantly learning new nuggets of information, AI inference accelerators are specialized hardware designed to accelerate basic machine learning computations while improving performance, reducing latency, and cutting the cost of deploying machine learning–based applications. If you want a deep dive into this, read more here.