Solving real-world artificial intelligence - whether for autonomous driving, real-world robotics, or advanced reasoning - requires an almost unfathomable amount of computational power. To meet this challenge, Tesla has been developing its own custom AI training hardware while simultaneously purchasing hardware in the open market.
Now, the next-generation Dojo 2 chip has reportedly entered mass production with the world’s largest semiconductor manufacturer, TSMC. While many may consider this a side quest, expanding Tesla’s computing base will be necessary to achieve exascale supercomputing, which will be crucial for all of Tesla’s AI ambitions.
Elon Musk called Dojo 2 “a good computer,” and then followed up with a classic computer performance joke - Dojo 2 can indeed play Crysis at a billion frames per second.
While Tesla has effectively utilized powerful third-party GPUs to train its models to date, the Dojo supercomputer is a ground-up, application-specific solution designed for a single purpose. It will efficiently process massive amounts of video data for training neural networks. The Dojo 2 chip itself is the key that unlocks this potential.
Dojo 2 will train the vision-based neural nets that FSD relies on, allowing Tesla to process video from its massive global fleet of vehicles even faster. As Tesla continues to improve FSD, one of the biggest challenges has been the intake of video for handling difficult edge cases.
Hundreds of thousands of miles of training data may pass by before an edge case is identified and trained on, but it all needs to be analyzed, labeled, and processed, which is key for Dojo 2. Each new useful piece of training data will help Tesla proceed down the march of 9s, making FSD just that little bit better every time.
This process requires massive amounts of compute and training time - but it is an absolute necessity to improve FSD. Of course, this goes beyond just FSD in vehicles. Tesla’s humanoid robot, Optimus, also runs on FSD to navigate and interact with the physical world.
While it may be a custom version of FSD, it remains FSD at its core, which means the same neural nets that analyze the environment and build a 3D map of the world for your car perform the same work for Optimus.
Not a Tesla App
Custom Approach to AI Hardware
Dojo 2’s power doesn’t just come from raw compute; it comes from a series of architectural choices that make it excel at training FSD and differentiate it from general-purpose hardware, or even other AI-specific hardware.
To this end, Tesla is using TSMC’s new Integrated Fan-Out with Silicon-on-Wafer (InFO-SoW) packaging technology. For massive AI workloads, heat and the speed at which data moves between chips are often the biggest bottlenecks.
This new packaging technique allows for high-bandwidth connections directly between processing dies, which lowers latency and dramatically improves heat dissipation, all key to building massive and dense compute clusters.
Unlike general-purpose chips, Dojo 2 is designed with a custom instruction set, specifically built to train FSD. The cores are specifically made to accelerate the exact mathematical operations, like matrix multiples and systolic arrays, which form the backbone of Tesla’s vision-based neural networks.
By building its own hardware, Tesla can then integrate its own software and compilers directly with the silicon, optimizing for specific workloads and avoiding the performance penalties that can result from using third-party software, such as Nvidia’s CUDA.
The start of Dojo 2 may seem like a side quest for some, but it’s actually a key step for Tesla’s AI technologies that give them an advantage over the competition using off-the-shelf hardware. They’ll need to continue investing in custom hardware to improve FSD at a reasonable pace, rather than the current glacial pace we’ve seen over the last few months.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
XAI’s Grok feature was announced to be a part of Tesla’s 2 2025.26 software update. However, it will only be available for AMD-based vehicles, excluding older, Intel-based cars (how to tell which you have). While Grok is by far the most exciting feature in Tesla’s Summer Update, many were hoping it would arrive on all Tesla vehicles, especially given that it’s primarily a cloud-based feature. However, there may still be hope for Intel owners, based on some careful language by Tesla.
A series of recent clues from Tesla’s website, as well as a repost from Elon, suggest that this hardware limitation likely isn’t permanent, and it may just be a matter of time before Grok is available on Intel MCUs.
Reading Between the Lines
The first clues come directly from Tesla’s official homepage. The main banner on the homepage describing the feature stipulates that the current requirements are subject to change, and more keenly, that Grok “currently requires an AMD processor”. That sounds like temporary, flexible language to accommodate Grok making its way to Intel vehicles in the near future.
Over on the official Grok FAQ page, Tesla noted that “Grok may become available for additional Tesla vehicles with over-the-air software updates in the future.” This text is also placed immediately after the hardware requirements section. It seems that Tesla may already be working on a Grok version for Intel vehicles, or they’re at least going to try to add it to more vehicles.
Not a Tesla App
Grok
All the heavy lifting for Grok is done on the server, where it transcribes the driver’s audio to text, finds an answer, and then translates text back into audio and sends it to the vehicle. Tesla did feature a Siri-like animation when Grok is talking in the vehicle, and this could be one feature that’s cut from older vehicles.
But Wait, There’s More
Adding fuel to the fire, Elon also reposted the following post from X, which suggests that older vehicles will also receive Grok in the future.
While we doubt that the original Tesla MCU (NVIDIA-based MCU 1) in a 2012 Model S would be updated to support Grok, MCU 1 vehicles can be updated to MCU 2 (Intel). While this is never a certainty that a feature is coming, it is a good hint that Grok may make its way to older vehicles in the future.
Just imagine a 2012 Model S getting Grok 4. 13 years later, it will still be the most intelligent vehicle out there in the world. This is going to be happening by latest next week. pic.twitter.com/3xVFGv9JU0
If Grok is set to expand to additional vehicles in the future, the next most logical candidates are the hundreds of thousands of Teslas on the road today equipped with the Intel Atom processor, also known as MCU2. These vehicles, built years prior to the switch to AMD (MCU 3), represent a huge chunk of the Tesla fleet.
While Intel units aren’t as powerful as the latest AMD units, they’re still a very capable unit that Tesla was able to add weather radar to, a feature many assumed would not be available on Intel MCUs.
For now, Grok remains a US-only feature that just started rolling out yesterday. It’s currently only available to a very small fraction of owners, approximately 0.2%, but that number is expected to increase as Tesla rolls out additional waves. You can track the rollout on our statistics page. For owners with Intel-based vehicles, this is a glimmer of hope that Tesla isn’t abandoning you yet.
It has been a long wait for FSD for European customers, many of whom paid for the feature years ago on now legacy hardware. While the FSD transfer program has come and gone multiple times, there’s something to be said about having it available in North America, where it can be used, and in Europe or other countries, where it still just remains the same as Enhanced Autopilot (differences between Autopilot, EAP, and FSD).
FSD Transfer is a nice goodwill gesture from Tesla that in theory doesn’t cost them anything. Instead, it keeps customers, especially those who have been waiting for years, loyal and happy. It also incentivizes them to upgrade to a newer Tesla with HW4, where FSD will hopefully be achieved.
In a reply to a post on X, Elon agreed with the suggestion that offering FSD transfers in Europe would be a fair solution for those who have already purchased FSD but can’t use its capabilities.
FSD Transfer directly addresses a growing concern for many long-term European Tesla owners. Thousands of customers purchased the full package, often many years ago, with the expectation that FSD would eventually be capable and approved for use. However, the reality is that FSD, even as an advanced driver assistance system (ADAS), continues to be pushed back in Europe.
Without the transfer program, it's a difficult choice: either throw away your original investment in FSD and pay for the package a second time (FSD price history), or subscribe to it in the future.
Offering FSD transfers is a good way for Tesla to meet them halfway. It's a difficult situation, and one that’s being hindered by processes beyond the control of both the customer and Tesla. However, a transfer helps both parties. Tesla sells another vehicle, and the customer gets to keep FSD.
When Will it Be Available?
Based on how FSD transfers have worked in the past following Elon’s announcements, this feature is likely to become available for a limited time period in the coming days or weeks. If it happens, we should expect an announcement from Tesla Europe on X and emails being sent out to Tesla customers.
Once the program is in place, all you need to do is complete your vehicle purchase and then inform your Tesla sales advisor that you’d like to transfer FSD. You don’t even have to sell or trade in your old Tesla; FSD will simply be removed from it as a feature.
Hopefully, Tesla enables FSD Transfers for everyone, regardless of region. It should be an ongoing offer until at least FSD is approved in the given country or region.
![Tesla Software Update 2025.26: Grok, Light Sync, Audio Presets and More [VIDEO]](https://www.notateslaapp.com/img/containers/article_images/tesla-update/light-sync-update.jpeg/e84ab2f1fe12f493a75927db105a9586/light-sync-update.jpg)
