Tesla recently launched their new AI training cluster with 10,000 GPUs
NVIDIA
In a recent TIME article adapted from his forthcoming book on Elon Musk, Walter Isaacson gives us a rare look into the innovative strides Tesla is making in the realm of artificial intelligence. Musk's ambitions have consistently set him apart, and Isaacson provides intriguing details, such as Musk's original intent to merge Tesla with OpenAI.
Though the idea didn't pan out, Musk's unyielding vision saw him championing rival AI teams to birth a series of groundbreaking projects. With Tesla vehicles sending back "160 billion frames per day of video" of humans in diverse real-world situations, the scale and scope of the data being harnessed are staggering.
"Tesla's real-world AI is underrated," Musk emphasized, highlighting the company's edge in the AI race. He posed a challenging notion: "Imagine if Tesla and OpenAI had to swap tasks... Who wins? We do."
FSD V12: More than Just Code
The Full Self-Driving v12 live stream from Musk's phone to X has generated more than 45 million views. This wasn't merely an upgrade but a glimpse into a future where technology starts to mimic human intelligence and seamlessly intertwines with it all, thanks to the power of neural nets.
This evolution marks a radical departure for those tracking Tesla's technological trajectory. Musk revealed that they had scrapped over 300,000 lines of hard-coded programming from the car's neural networks to take the helm. The FSD V12's efficacy was laid bare as it tackled construction zones, speed bumps, roundabouts, and traffic with much more ease than we've seen in public releases. Musk encapsulated this transition with his apt exclamation, "This is all nets, baby, nothing but net."
Real-World Challenges & AI Nuances
But it's not just about making the car drive. As Musk demonstrated, it's about making it drive like a human — with all the nuances that come with that. From not jerking abruptly to finding a lane to the smoothness in transitioning from one traffic condition to another, the FSD V12 is designed to observe, learn, and replicate the best of human driving behavior. However, it also underscores the challenges of programming AI for real-world scenarios. As Musk explained, the sparse data on drivers coming to complete stops at stop signs poses an obstacle. With less than 0.5% of drivers fully adhering to this rule, it's a stark reminder that AI is navigating a world where humans often flout regulations.
Looking Ahead: AI & Beyond
Isaacson's insights, combined with the V12 showcase, hint at a future where Tesla isn't just leading the electric vehicle market but is also on the frontlines of the AI revolution. Musk's commitment to AI is evident as he navigates the complexity of running multiple ventures, from SpaceX to Neuralink and the much-anticipated xAI. The next chapter promises even more advancements. Isaacson's book release is tomorrow, September 12th.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
For the first time in quite a while, Tesla has increased the price of one of its vehicle offerings. The Model S Long Range and the Model S Plaid have both received a hefty price increase. However, not all is bad - as Tesla also added a new benefit for buyers.
Price Increase
The price increase for the Model S is $5,000 - currently only in the United States. This price increase will likely impact other markets, including Canada, in the coming days.
The Model S Long Range now starts at $79,990, while the Model S Plaid now starts at $94,990. The Model S and Model X now have the same starting price. Interestingly, that’s also the same pricing point for the Cybertruck AWD and Cyberbeast Trimotor non-Foundation Series.
The vehicle configuration does not appear to have changed, so the new pricing is simply an increase rather than the addition or removal of features. Tesla previously cut the price of the Model S and Model X by 15% in 2023, so this could simply be an adjustment to ensure that the vehicle pricing stays in line with inflation and other factors.
Free Lifetime Supercharging
For those on the fence about ordering a Model S, Tesla has brought back Free Lifetime Supercharging when you order a new Model S on or after December 13, 2024. As always, Free Lifetime Supercharging is restricted to the buyer’s Tesla account and to that specific vehicle. It cannot be transferred to another vehicle or another owner after ownership transfer. It’s worth noting, that it also doesn’t exclude the owner from receiving Supercharger idle fees or congestion fees. There is currently no end date for this promotion.
There are currently no changes to the Model X, neither a price increase nor the addition of Lifetime Superchargering. However, when Tesla makes changes to one of their premium vehicles, it usually affects the other one as well. So be on the lookout for potential changes to the Model X offering in the coming days.
We’ve seen Tesla value Lifetime Supercharging at $5,000, so this falls in line with the price increase we’re seeing. It’s possible that Tesla will begin to bring back Lifetime Supercharging as a perk for buying into their more premium Model S and Model X cars, or this could be another temporary promotion to get buyers who are on the fence to go ahead and make their purchase while this promotion lasts.
Tesla recently showed off a demo of Optimus, its humanoid robot, walking around in moderately challenging terrain—not on a flat surface but on dirt and slopes. These things can be difficult for a humanoid robot, especially during the training cycle.
Most interestingly, Milan Kovac, VP of Engineering for Optimus, clarified what it takes to get Optimus to this stage. Let’s break down what he said.
Optimus is Blind
Optimus is getting seriously good at walking now - it can keep its balance over uneven ground - even while walking blind. Tesla is currently using just the sensors, all powered by a neural net running on the embedded computer.
Essentially, Tesla is building Optimus from the ground up, relying on as much additional data as possible while it trains vision. This is similar to how they train FSD on vehicles, using LiDAR rigs to validate the vision system’s accuracy. While Optimus doesn’t have LiDAR, it relies on all those other sensors on board, many of which will likely become simplified as vision takes over as the primary sensor.
Today, Optimus is walking blind, but it’s able to react almost instantly to changes in the terrain underneath it, even if it falls or slips.
What’s Next?
Next up, Tesla AI will be adding vision to Optimus - helping complete the neural net. Remember, Optimus runs on the same overall AI stack as FSD - in fact, Optimus uses an FSD computer and an offshoot of the FSD stack for vision-based tasks.
Milan mentions they’re planning on adding vision to help the robot plan ahead and improve its walking gait. While the zombie shuffle is iconic and a little bit amusing, getting humanoid robots to walk like humans is actually difficult.
There’s plenty more, too - including better responsiveness to velocity and direction commands and learning to fall and stand back up. Falling while protecting yourself to minimize damage is something natural to humans - but not exactly natural to something like a robot. Training it to do so is essential in keeping the robot, the environment around it, and the people it is interacting with safe.
We’re excited to see what’s coming with Optimus next because it is already getting started in some fashion in Tesla’s factories.