Tesla (TSLA), SpaceX, and xAI CEO Elon Musk appeared on the All-In Podcast a few days ago. He gave out some important information on the status and plans on AI-based products his companies are working on.
“Your car is going to feel like it’s sentient by the end of the year,” Musk said. This is where the Tesla FSD AI is pushing self-driving technology.
During this part of the talk (both full and partial videos below), Elon Musk also gave us an updated timeline of the FSD v14 deployment to customer-owned Tesla cars.
– Advertisement –
Tesla FSD v14 Rollout Timeline
In the 2nd week of August, Elon Musk announced that FSD v14 is going to be launched in around 6 weeks. However, during the All-In Podcast session, Musk rescheduled the FSD v14 rollout timeline to the end of the year (2025).
Musk is generally known for being over-optimistic with his delivery timelines. FSD v14 has faced several delays this year. However, it’s confirmed that Tesla employees are internally testing this critical version of the automaker’s self-driving AI software.
So, according to Elon Musk’s following remarks during the podcast, the FSD v14 rollout is delayed once again and should be expected around Christmas holidays.
I am confident that the current chips, AI4 chips that are in the cars, will achieve self-driving safety that is at least that is 2-3 times that of humans, maybe even 10X.
And the software that will be released for that is coming out over the next few months. So, version 14 will be the biggest upgrade in Tesla software since version 12.
We are increasing parameter count by an order of magnitude. There’s a lot of reinforcement learning that’s been used.
Ah, like you can think of AI, sort of as a way of compressing reality, and some of those compression steps were too lossy, and we addressed the lossiness in the compression steps. These are all software updates that’ll go out. So, just over the air updates.
Your car is going to feel like it’s sentient by the end of the year.
– Advertisement –
If this clip doesn't get you excited for Tesla FSD v14 I don't know what will!
ALSO: As a huge Tesla fan and owner of an AI4-equipped Tesla vehicle, I'm really happy that @elonmusk reiterated how AI4 remains extremely capable and will remain so for the foreseeable future. I've… pic.twitter.com/sarBOymqfT
According Musk, the upcoming AI5 chip is around 40 times better than the existing AI4 (HW4) silicon chip installed in Tesla vehicles delivered to customers since Jan 2023.
The good news for FSD v14-based Tesla owners is that FSD v14 will work on their cars. However, the true performance of the next generation of Tesla Full Self-Driving will be unlocked by AI5 and AI6 chips.
Musk told the audience that FSD on AI4 chips is still going to be 2-3 times safer than an average human driver. It may even improve by 10X compared to humans.
Here’s what he said on AI5 and AI4 chips during the All-In Podcast:
At Tesla, we basically had two chip programs. Dojo on the training side and AI4, which is our inference chip.
AI4 is currently shipping in all vehicles. And we’re finalizing the design of AI5 which will be an immense jump from AI4.
By some metrics, the improvement in AI5 will be 40 times better than AI4. And, this is because we work so closely at a fine-grained level on the AI software and the AI hardware. So, we know exactly where the limiting factors are.
And, so, effectively, the AI hardware and software teams are co-designing the chip.
To be precise, the 40x is on if you say like compared to the worst limitation on AI4, which is running the SoftMax operation. Ah, we currently have to run SoftMax in around 40 steps in emulation mode, whereas that’ll just be done in a few steps natively in AI5.
– Advertisement –
AI5 chip will also easily handle mixed-precision models. So, you don’t have to… it’ll dynamically handle mixed precision. There’s a bunch of, sort of technical stuff that AI5 will do a lot better.
In terms of nominal raw compute, it’s eight times more compute, about 9 times more memory, roughly 5 times more memory bandwidth.
So, but because we’re addressing some core limitations in AI4, you multiply that by 8x computer improvement by another 5x improvement because of optimization at a very fine grain silicon level of things that are currently sub-optimal in AI4.
That’s where you get the 40x improvement.
Video: Tesla and SpaceX CEO Elon Musk talks on the All-In Podcast.
Author of more than 1500 articles on Tesla, SpaceX, and EVs. His work has been liked and tweeted by Elon Musk and other prominent influencers. You can reach him on Twitter @IqtidarAlii
