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.”
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.”
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.
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.
Lifestyle
Tesla Semi hauls fresh Cybercab batch as Robotaxi era takes hold
A Tesla Semi was filmed hauling Cybercab units out of Giga Texas for the first time.
A Tesla Semi loaded with Cybercab units was recently filmed leaving Gigafactory Texas, marking what appears to be the first documented delivery run of Tesla’s autonomous two-seater. The footage shows multiple Cybercabs secured on a flatbed trailer being hauled by a production Tesla Semi, a truck rated for a gross combination weight of 82,000 lbs. The location is consistent with Giga Texas in Austin, where Cybercab production has been ramping since February 2026.
The sighting follows a wave of Cybercab activity at the Austin facility. In late April, drone operator Joe Tegtmeyer spotted approximately 60 Cybercabs parked in two organized groups in the factory’s outbound lot, the largest concentration observed to date. Units being staged in an outbound lot is a standard pre-delivery step, and the Semi footage is the logical next frame in that sequence.
En route with @tesla_semi pic.twitter.com/ZfuOjaeLH1
— Tesla Robotaxi (@robotaxi) May 7, 2026
This is not the first time Tesla has used its own Semi to move Tesla products. When the Semi was unveiled in 2017, Musk noted it would be used for Tesla’s own operations, and over the years Semi prototypes were spotted carrying cargo ranging from concrete weights to Tesla vehicles being delivered to consumers. In 2023, a Semi was photographed transporting a Cybertruck on a trailer ahead of that vehicle’s delivery launch.
The Cybercab itself was first revealed publicly at Tesla’s “We, Robot” event on October 10, 2024, at Warner Bros. Studios in Burbank, where 20 pre-production units gave attendees rides around the studio lot. Musk stated at the event that Tesla intends to produce the Cybercab before 2027. The first production unit rolled off the Giga Texas line on February 17, 2026, with Musk posting on X: “Congratulations to the Tesla team on making the first production Cybercab.”
Tesla’s annual production goal is 2 million Cybercabs per year once multiple factories reach full design capacity, with the company targeting a price under $30,000 per unit. Tesla has confirmed plans to expand its robotaxi service to seven cities in the first half of 2026, including Dallas, Houston, Phoenix, Miami, Orlando, Tampa, and Las Vegas, building on the unsupervised service already running in Austin. Musk has said he expects robotaxis to cover between a quarter and half of the United States by end of year.
Cybertruck
Tesla Cybertruck too safe for even Musk’s biggest critics to ignore
Krassenstein’s decision reveals that superior safety isn’t a partisan issue. For parents prioritizing family protection over personal or political grudges, the Cybertruck has become too safe to ignore.
The Tesla Cybertruck is an extremely polarizing vehicle because of its potential symbolism as a political stance instead of just a pickup truck — or at least that is what many would want you to believe.
Of course, the Cybertruck is an icon of Tesla culture, and it is one of those things that never has a middle ground: you love it, or you don’t.
But maybe there is an establishment of that “grey area” happening.
In a striking illustration of engineering triumph over political tribalism, prominent Elon Musk critic Brian Krassenstein has purchased a Tesla Cybertruck, openly citing its exceptional safety as the deciding factor for his family.
The announcement on X triggered predictable backlash, yet it underscores a growing reality: the Cybertruck’s safety credentials are proving impossible for even Musk’s fiercest detractors to dismiss.
I might get hate for this too but I bought a Cybertruck.
With a young family, safety was important and so is not polluting the atmosphere with $5 a gallon gasoline. pic.twitter.com/XJqFqR6O9r
— Brian Krassenstein (@krassenstein) May 6, 2026
Krassenstein, who has repeatedly clashed with Musk over issues ranging from content moderation and “wokeness” to public health figures, made no attempt to hide his reservations. In his May 6 post, he acknowledged the coming criticism: “I might get hate for this too but I bought a Cybertruck.”
He stressed that the decision had “nothing to do with Elon or politics,” pointing instead to practical advantages—his existing Tesla charger, eligibility for Full Self-Driving upgrades, a returning-owner discount, and crucially, the vehicle’s strong safety profile.
With gasoline prices hovering near $5 a gallon in some areas, he also highlighted the environmental benefit of switching from a polluting combustion engine.
The numbers, data, and awards validate Krassenstein’s choice.
The 2025 Cybertruck earned the Insurance Institute for Highway Safety’s (IIHS) elite Top Safety Pick+ award—the only pickup truck to achieve this highest rating. It delivered “Good” scores across every crashworthiness category, including the challenging updated moderate overlap front crash test, while excelling in crash avoidance and mitigation systems.
The National Highway Traffic Safety Administration (NHTSA) awarded it a perfect 5-star overall rating, with top marks in frontal, side, and rollover categories. No other pickup truck holds both distinctions simultaneously.
Tesla Cybertruck crash test rating situation revealed by NHTSA, IIHS
Beyond lab results, the Cybertruck’s stainless-steel exoskeleton and ultra-rigid structure have demonstrated remarkable real-world resilience. Owners have reported surviving high-speed collisions with minimal cabin intrusion.
In one widely discussed incident, a Cybertruck endured a 70 mph sideswipe on the interstate; the driver reported barely feeling the impact while the other vehicle was heavily damaged.
Tesla’s crash demonstrations and independent analyses consistently show how the vehicle’s design prioritizes occupant protection through a fortified passenger cell rather than traditional crumple zones, giving families superior safeguarding in many common crash scenarios.
The online pile-on following Krassenstein’s post focused on aesthetics, politics, and perceived hypocrisy rather than the data. Critics called the angular truck “ugly” or accused him of selling out.
Yet his purchase highlights an inconvenient truth for polarized discourse: when objective safety metrics—IIHS awards, NHTSA ratings, and documented crash performance—point decisively toward one vehicle, even Musk’s biggest critics are forced to confront its merits.
Krassenstein’s decision reveals that superior safety isn’t a partisan issue. For parents prioritizing family protection over personal or political grudges, the Cybertruck has become too safe to ignore.
SpaceXAI announced today that it had signed an agreement with Anthropic to give the company access to its Colossus 1 data center in Memphis, Tennessee.
It is a monumental deal as Anthropic will gain access to all of the compute at the plant, delivering more than 300 megawatts of power and over 220,000 NVIDIA GPUs within the month.
Anthropic’s Claude AI account on X announced the partnership:
“We’ve agreed to a partnership with SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.”
The company is also:
- Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans;
- Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and
- Substantially raising its API rate limits for Opus models.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity.
This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
— Claude (@claudeai) May 6, 2026
SpaceX also published its own release on the new agreement, noting that it is “the only organization with the launch cadence, mass-to-orbit economics, and constellation operations experience to make orbital compute a near-term engineering program rather than a research concept.”
CEO Elon Musk also commented on the partnership and shed light on intense meetings he had with senior members of Anthropic last week, stating, “nobody set on my evil detector.”
Same here.
By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed.
Everyone I met was highly competent and cared a great deal about…
— Elon Musk (@elonmusk) May 6, 2026
This has turned the argument that SpaceX is as much an AI company as a space exploration company into a very valid argument:
SpaceX is following in Tesla’s footsteps in a way nobody expected
Nevertheless, this is an incredibly valuable and important move in the grand scheme of things. AI scaling is fundamentally bottlenecked by compute, and demand for Claude has surged, bringing terrestrial power grids, land, and cooling operations hitting limits everywhere.
Anthropic has been aggressively signing multiple large-scale deals to be competitive in the space, including:
- Up to 5GW with Amazon
- 5GW with Google and Broadcom
- Strategic $30b Azure deal with Microsoft/NVIDIA
- $50b U.S. infrastructure investment with Fluidstack
Access to Colossus 1 gives Anthropic immediate relief on NVIDIA GPU capacity. For SpaceXAI, it turns its rapid buildout into revenue. It also showcases its ability to deliver at world-leading speed and scale.
Most importantly, it plants the seed that its much larger vision, orbital AI compute, is totally viable.
Starlink V3 satellites could enable SpaceX’s orbital computing plans: Musk
Within the month, Anthropic will begin using 100 percent of Colossus 1’s compute, directly expanding capacity for Claude Pro and Max subscribers and the API. This means fewer limits, faster responses, and support for heavier workloads.
In the long term, meaning 2026 and beyond, there will be a continued rollout of other multi-GW deals Anthropic has signed, and an early exploration of orbital compute with SpaceXAI.
Tesla Semi hauls fresh Cybercab batch as Robotaxi era takes hold
Tesla Cybertruck too safe for even Musk’s biggest critics to ignore
