Less than ten days since the last FSD update, Tesla has started rolling out FSD v12.3.6. Not only does this update include improvements to FSD itself, but it also adds the new Autopark and High Fidelity Park Assist to vehicles with ultrasonic sensors.
As Tesla continues to ramp up its AI training capacity, the average number of miles driven on FSD continues to climb at an increased rate.
Computing Power
Last week, in the earnings call deck, Tesla revealed that it’s set to increase its core AI infrastructure capacity “in the coming months.” Tesla stated it has quadrupled its capacity since September 2023.
To further improve our end-to-end training capability, we will continue to increase our core AI infrastructure capacity in the coming months. - Tesla
Musk said on Sunday that the company will spend “around $10B” throughout 2024 in “combined training and inference AI, the latter being primarily in car”. “Any company not spending at this level, and doing so efficiently, cannot compete,” he added.
Tesla's computing power has drastically increased
Not a Tesla App
Expansion to China
In an unexpected 24-hour trip to China, Elon Musk visited Beijing to meet with Premier Li Qiang to discuss the rollout of Full Self-Driving (FSD) software and permission to transfer driving data overseas. Since Tesla uses fleet data to continuously improve FSD, being able to use data gathered in and outside of China will be crucial to its continued improvement.
Elon Musk meeting with Premier Li Qiang today in China.
On the same day, Tesla models produced at Giga Shanghai were listed among the EVs that meet China's data security requirements for smart cars, an important hurdle for Tesla to surpass.
When Musk was asked on X about Tesla releasing FSD in China, Musk simply said “It may be possible very soon.”
Regarding additional markets such as Europe and Australia, Musk said he believes that v12 is “ready for supervised FSD in LHD countries” while RHD like the UK or Australia “will take a bit longer”.
Internal FSD Is 3 to 6 Months Ahead
Earlier this month, Tesla disclosed that the team knows “roughly” how the software will perform in “3 to 6 months”. He added that it takes “a few months to fix those bugs and complete safety tests” before releasing the version to the Tesla employees and then to the general public.
Major FSD Updates Coming
On X, Musk teased that the next major update to FSD will be v12.4, saying “it’s awesome,” and that it’s “another big jump in capabilities.” Musk said that Tesla’s computing constraint has been much improved, and likely may be a factor in this further being another big improvement.
In mid-March, Elon Musk announced that Tesla will reintroduce Smart Summon and Autopark to its vehicles. Tesla has now delivered on the Autopark promise, but we’re still waiting on the improved Smart Summon feature. It’s possible we’ll see in FSD v12.4.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
Tesla’s plan to brand its autonomous network of taxicabs has found an interesting little snag. The US Patent and Trademark Office (USPTO) has issued a “nonfinal office action” regarding Tesla’s application to trademark the term “Robotaxi” specifically in connection with autonomous electric vehicles.
This is an initial refusal by USPTO’s examining attorney and is very particular for autonomous electric automobiles. A separate trademark application by Tesla for the term “Robotaxi” as it relates to its upcoming autonomous ride-hailing service is still under examination and has not yet received a similar rebuttal.
Understanding the Refusal
A “nonfinal office action” means the USPTO has found potential issues with the trademark application, as stated, which prevents its immediate approval and subsequent granting to Tesla. Tesla now has a three-month period to file its counterarguments and address the USPTO's concerns.
If Tesla’s response satisfies the examiner, the trademark could be granted.
While the exact content of the office action isn’t detailed in the initial report, such refusals for terms like “Robotaxi” often occur if the USPTO considers the term “merely descriptive” or “generic” for the goods in question. In this particular context, “Robotaxi” could refer to any autonomous taxi vehicle.
Trademark law generally prevents the exclusive registration of terms that competitors would need to use to describe their own similar products. For a term to be trademarked, it typically needs to be distinctive and act as a brand identifier rather than just a descriptive name of the product’s class or type.
Separate Application for Ride-Hailing
Tesla still has a distinct, separate, and still pending application to register “Robotaxi” as a trademark for “transportation services, namely, autonomous ride-hailing services.” The criteria for trademarking a service can differ from those on trademarks for goods, and it’s possible Tesla may have more success securing the name for the service itself, which would allow them to brand the network as “Tesla Robotaxi.”
Why This Matters
Securing a trademark grants exclusive rights to use a brand name in conjunction with specific goods or services. This helps prevent customer confusion and to protect the brand identity.
If the refusal for the vehicle trademark becomes final, Tesla may be limited in its ability to exclusively name a good (specific vehicle) the “Tesla Robotaxi.” Other manufacturers could also potentially use “robotaxi” descriptively for their own autonomous taxi vehicles.
The ability to trademark “Robotaxi” for the ride-hailing service is arguably more critical for Tesla, as they’re working to establish a unique brand for their autonomous transportation network, which kicks off in Austin next month.
The USPTO’s office action won’t hinder Tesla’s ability to develop or deploy its own vehicles in June - instead, it’ll just impact how Tesla can brand the app and their vehicles, which could cause some last-minute delays if they have to rebrand.
Cybercab and Robovan/Robobus Trademarks
While Tesla is facing challenges with the broader Robotaxi term for vehicles, the company is also seeking to trademark “Cybercab,” “Robovan,” and “Robobus.” Securing a less descriptive name for the vehicle itself often has a higher chance of success with USPTO, as it is far more distinctive than a more general term like “robotaxi.”
Why Didn’t Tesla Do This Years Ago?
Tesla may have waited too long to file a trademark for the term “Robotaxi.” While the company has been discussing a self-driving fleet since 2016, the concept of autonomous taxis has gained a lot more traction in recent years — and competitors like Uber have also begun using the term.
We suspect there was some strategic timing behind these filings. Earlier versions of FSD — particularly those prior to V12 — may have lacked the progress needed to support Tesla’s robotaxi ambitions. Filing for a trademark that isn’t actively in use or about to be used can make it harder to defend or retain.
Moreover, while the idea of autonomous vehicles has been around for years, a clearer public understanding of Tesla’s specific plans has only emerged over the past 18 months. Filing too early can trigger speculation long before the company is ready to reveal details.
Ultimately, whether Tesla secures the rights to “Robotaxi” remains uncertain — but trademarks like “Cybercab” and “Robovan” seem much more likely to stick.
One of Tesla’s greatest weaknesses, as it has quickly become one of the world’s most ubiquitous cars on the planet has always been service. Escalating issues to managers and sometimes even reaching a Tesla Service employee can be a total coin flip, depending on your Service Center.
Tesla is continuing its push to integrate AI across its customer support channels in an effort to improve customer service. According to Raj Jegannathan, Tesla’s VP for IT, AI Infrastructure, Apps, Infosecurity, and Vehicle Service Operations (that’s a lot), Tesla is launching a pilot program for a new AI designed to improve customer interactions with Service.
Tesla Service’s new AI Agent detects comms delays, monitors sentiment, & auto-escalates to leaders. Starts tomorrow at 10 pilot locations. In 2 weeks, type “Escalate” in ‘message center’ to reach managers. Guardrails in place to prevent abuse. We’ll keep improving!
At 10 pilot service locations, this new AI agent will begin working behind the scenes at Tesla Service, to help with customer communications. It will provide three key features:
Detect Communications Delays: The AI will actively monitor service interactions to identify potential delays in communication or progress. These are often a key pain point for customers who reach out to Tesla Service and don’t receive a response for several days, as Service has nothing new to add. The AI can now step in and let the customer know Tesla is still waiting on parts or something else.
Monitor Customer Sentiment: By monitoring the tone and content of the messages between the customer and Service agents, Tesla will be able to identify situations where a customer might be dissatisfied or facing difficulties.
Auto-Escalate: If either a communications delay or negative sentiment is detected, the AI can automatically escalate issues to human managers for review. This helps to address problems before customers need to seek escalation themselves or become upset about an issue.
Customer Escalation Requests
Alongside the new AI tool, Tesla is also introducing a more direct way for customers to get higher-level attention. According to Tesla, within the next two weeks, customers can simply type “Escalate” in order to have their issue routed directly to management.
Raj’s team is currently working on implementing guardrails to prevent abuse, but this will soon make its way to improving Tesla’s service offerings. We’re glad to see Tesla taking steps to identify and correct deficiencies in the process - it has always been a sore tooth for Tesla in the last few years.
![Exclusive: A Look Inside the Tesla Diner [PHOTOS]](https://www.notateslaapp.com/img/containers/article_images/2025/diner-seats.jpg/b0b4cb51b80d1f29298c9ca6b9ad2ae8/diner-seats.jpg)
_300w.png)
