Connect with us

News

Tesla FSD v12 shifts away from ‘rules-based’ approach

Credit: @sabber_dev/Twitter

Published

on

Walter Isaacson’s Elon Musk biography is set to be published on Tuesday, and a new preview of the book illustrates details about Tesla’s development of the upcoming Full Self-Driving (FSD) version 12.

In an additional preview of his Musk biography for CNBC, Isaacson discusses the use of AI in the development of Tesla’s FSD v12, in a shift that took place within the last several months. Isaacson talks about Tesla’s recent development of the upcoming FSD v12, which he and Tesla demonstrate has moved away from a “rules-based” approach.

Notably, FSD v12 is expected to use billions of video frames from real-world driving incidents to train its neural network rather than using thousands of lines of code like previous versions. In a conversation with Musk last December, Tesla Autopilot employee Dhaval Shroff had likened the concept to the popular chatbot ChatGPT, instead for use with driving.

“It’s like ChatGPT, but for cars,” Shroff said. “We process an enormous amount of data on how real human drivers acted in a complex driving situation, and then we train a computer’s neural network to mimic that.”

Advertisement

Surprisingly enough, Tesla only shifted toward this “neural network planner” approach recently. By the beginning of this year, however, the neural network had already analyzed 10 million video clips based on the best-case-scenario drivers the system had access to. Musk instructed employees at the company’s Buffalo, New York facility who were in charge of analyzing the footage to train the AI on things “a five-star Uber driver would do.”

Moving from a rules-based to a network-path-based AI approach allowed FSD to use human driving data to avoid obstacles, even if breaking some rules was necessary. Shroff helped demonstrate the idea to Musk with a demo featuring trash bins, debris, and upturned traffic cones, which the car handled surprisingly well.

“Here’s what happens when we move from rules-based to network-path-based,” Shroff explained. “The car will never get into a collision if you turn this thing on, even in unstructured environments.”

Tesla FSD’s “autosteer on city streets” now available

Advertisement

Musk quickly took to the idea, as can be seen in a recent livestream of Tesla’s FSD v12 software in Palo Alto with Autopilot software director Ashok Elluswamy. He has repeatedly spoken about the upcoming software version’s impressive driving results, despite one small moment in the drive where the car almost ran a red light.

In any case, Musk could argue that the red-light moment is a good case for the need for self-driving software to continually learn. Given that it will constantly be trained from the video data generated by camera footage from real-world drivers, it should theoretically make it safer over time, according to Musk.

During development, Musk also reportedly latched onto the fact that it took over a million video clips for the neural network to begin performing well, though he looks forward to what significantly more data will do for FSD.

Still, critics and regulators have expressed concerns about the faults of human drivers training AI-based driving systems, and Tesla has repeatedly been questioned by the National Highway Traffic Safety Administration (NHTSA) about its Autopilot and FSD beta systems.

Advertisement

According to Isaacson, Tesla plans to release FSD v12 as soon as regulators approve it. Meanwhile, an ongoing study by the National Highway Safety Board is looking to determine if self-driving cars should be permitted to imitate human driving actions that blur traffic rules, such as creeping up at stop signs.

Musk said in April that he expects Tesla to reach full autonomy within a year, though he has also been known to share ambitious targets for the software in the past.

You can read Walter Isaacson’s full account of the development of Tesla FSD v12 here, in a CNBC preview of the upcoming Elon Musk biography.

What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send your tips to us at tips@teslarati.com.

Advertisement

Zach is a renewable energy reporter who has been covering electric vehicles since 2020. He grew up in Fremont, California, and he currently lives in Colorado. His work has appeared in the Chicago Tribune, KRON4 San Francisco, FOX31 Denver, InsideEVs, CleanTechnica, and many other publications. When he isn't covering Tesla or other EV companies, you can find him writing and performing music, drinking a good cup of coffee, or hanging out with his cats, Banks and Freddie. Reach out at zach@teslarati.com, find him on X at @zacharyvisconti, or send us tips at tips@teslarati.com.

Advertisement
Comments

Elon Musk

Celebrating SpaceX’s Falcon Heavy Tesla Roadster launch, seven years later (Op-Ed)

Seven years later, the question is no longer “What if this works?” It’s “How far does this go?”

Published

on

SpaceX's first Falcon Heavy launch also happened to be a strategic and successful test of Falcon upper stage coast capabilities. (SpaceX)

When Falcon Heavy lifted off in February 2018 with Elon Musk’s personal Tesla Roadster as its payload, SpaceX was at a much different place. So was Tesla. It was unclear whether Falcon Heavy was feasible at all, and Tesla was in the depths of Model 3 production hell.

At the time, Tesla’s market capitalization hovered around $55–60 billion, an amount critics argued was already grossly overvalued. SpaceX, on the other hand, was an aggressive private launch provider known for taking risks that traditional aerospace companies avoided.

The Roadster launch was bold by design. Falcon Heavy’s maiden mission carried no paying payload, no government satellite, just a car drifting past Earth with David Bowie playing in the background. To many, it looked like a stunt. For Elon Musk and the SpaceX team, it was a bold statement: there should be some things in the world that simply inspire people.

Inspire it did, and seven years later, SpaceX and Tesla’s results speak for themselves.

Advertisement
Credit: SpaceX

Today, Tesla is the world’s most valuable automaker, with a market capitalization of roughly $1.54 trillion. The Model Y has become the best-selling car in the world by volume for three consecutive years, a scenario that would have sounded insane in 2018. Tesla has also pushed autonomy to a point where its vehicles can navigate complex real-world environments using vision alone.

And then there is Optimus. What began as a literal man in a suit has evolved into a humanoid robot program that Musk now describes as potential Von Neumann machines: systems capable of building civilizations beyond Earth. Whether that vision takes decades or less, one thing is evident: Tesla is no longer just a car company. It is positioning itself at the intersection of AI, robotics, and manufacturing.

SpaceX’s trajectory has been just as dramatic.

Falcon 9 has become the undisputed workhorse of the global launch industry, having completed more than 600 missions to date. Of those, SpaceX has successfully landed a Falcon booster more than 560 times. The rocket flies more often than all other active launch vehicles combined, routinely lifting off multiple times per week.

Falcon Heavy successfully clears the tower after its maiden launch, February 6, 2018. (Tom Cross)

Falcon 9 has ferried astronauts to and from the International Space Station via Crew Dragon, restored U.S. human spaceflight capability, and even stepped in to safely return NASA astronauts Butch Wilmore and Suni Williams when circumstances demanded it.

Starlink, once a controversial idea, now dominates the satellite communications industry, providing broadband connectivity across the globe and reshaping how space-based networks are deployed. SpaceX itself, following its merger with xAI, is now valued at roughly $1.25 trillion and is widely expected to pursue what could become the largest IPO in history.

Advertisement

And then there is Starship, Elon Musk’s fully reusable launch system designed not just to reach orbit, but to make humans multiplanetary. In 2018, the idea was still aspirational. Today, it is under active development, flight-tested in public view, and central to NASA’s future lunar plans.

In hindsight, Falcon Heavy’s maiden flight with Elon Musk’s personal Tesla Roadster was never really about a car in space. It was a signal that SpaceX and Tesla were willing to think bigger, move faster, and accept risks others wouldn’t.

The Roadster is still out there, orbiting the Sun. Seven years later, the question is no longer “What if this works?” It’s “How far does this go?”

Advertisement
Continue Reading

Energy

Tesla launches Cybertruck vehicle-to-grid program in Texas

The initiative was announced by the official Tesla Energy account on social media platform X.

Published

on

Credit: Tesla

Tesla has launched a vehicle-to-grid (V2G) program in Texas, allowing eligible Cybertruck owners to send energy back to the grid during high-demand events and receive compensation on their utility bills. 

The initiative, dubbed Powershare Grid Support, was announced by the official Tesla Energy account on social media platform X.

Texas’ Cybertruck V2G program

In its post on X, Tesla Energy confirmed that vehicle-to-grid functionality is “coming soon,” starting with select Texas markets. Under the new Powershare Grid Support program, owners of the Cybertruck equipped with Powershare home backup hardware can opt in through the Tesla app and participate in short-notice grid stress events.

During these events, the Cybertruck automatically discharges excess energy back to the grid, supporting local utilities such as CenterPoint Energy and Oncor. In return, participants receive compensation in the form of bill credits. Tesla noted that the program is currently invitation-only as part of an early adopter rollout.

Advertisement

The launch builds on the Cybertruck’s existing Powershare capability, which allows the vehicle to provide up to 11.5 kW of power for home backup. Tesla added that the program is expected to expand to California next, with eligibility tied to utilities such as PG&E, SCE, and SDG&E.

Powershare Grid Support

To participate in Texas, Cybertruck owners must live in areas served by CenterPoint Energy or Oncor, have Powershare equipment installed, enroll in the Tesla Electric Drive plan, and opt in through the Tesla app. Once enrolled, vehicles would be able to contribute power during high-demand events, helping stabilize the grid.

Tesla noted that events may occur with little notice, so participants are encouraged to keep their Cybertrucks plugged in when at home and to manage their discharge limits based on personal needs. Compensation varies depending on the electricity plan, similar to how Powerwall owners in some regions have earned substantial credits by participating in Virtual Power Plant (VPP) programs.

Continue Reading

News

Samsung nears Tesla AI chip ramp with early approval at TX factory

This marks a key step towards the tech giant’s production of Tesla’s next-generation AI5 chips in the United States.

Published

on

Tesla-Chips-HW3-1
Image used with permission for Teslarati. (Credit: Tom Cross)

Samsung has received temporary approval to begin limited operations at its semiconductor plant in Taylor, Texas.

This marks a key step towards the tech giant’s production of Tesla’s next-generation AI5 chips in the United States.

Samsung clears early operations hurdle

As noted in a report from Korea JoongAng Daily, Samsung Electronics has secured temporary certificates of occupancy (TCOs) for a portion of its semiconductor facility in Taylor. This should allow the facility to start operations ahead of full completion later this year.

City officials confirmed that approximately 88,000 square feet of Samsung’s Fab 1 building has received temporary approval, with additional areas expected to follow. The overall timeline for permitting the remaining sections has not yet been finalized.

Advertisement

Samsung’s Taylor facility is expected to manufacture Tesla’s AI5 chips once mass production begins in the second half of the year. The facility is also expected to produce Tesla’s upcoming AI6 chips. 

Tesla CEO Elon Musk recently stated that the design for AI5 is nearly complete, and the development of AI6 is already underway. Musk has previously outlined an aggressive roadmap targeting nine-month design cycles for successive generations of its AI chips.

Samsung’s U.S. expansion

Construction at the Taylor site remains on schedule. Reports indicate Samsung plans to begin testing extreme ultraviolet (EUV) lithography equipment next month, a critical step for producing advanced 2-nanometer semiconductors.

Samsung is expected to complete 6 million square feet of floor space at the site by the end of this year, with an additional 1 million square feet planned by 2028. The full campus spans more than 1,200 acres.

Advertisement

Beyond Tesla, Samsung Foundry is also pursuing additional U.S. customers as demand for AI and high-performance computing chips accelerates. Company executives have stated that Samsung is looking to achieve more than 130% growth in 2-nanometer chip orders this year.

One of Samsung’s biggest rivals, TSMC, is also looking to expand its footprint in the United States, with reports suggesting that the company is considering expanding its Arizona facility to as many as 11 total plants. TSMC is also expected to produce Tesla’s AI5 chips. 

Advertisement
Continue Reading