News
Tesla Autopilot impresses in 79-minute ‘Torture Test’ through long winding roads
A Tesla Model 3 owner recently tested Autopilot’s abilities on the long and winding roads of the Blue Ridge Parkway in North Carolina. The 45.9-mile trip lasted 79 minutes, giving the driver a fair amount of insight regarding Autopilot’s performance with update 2020.16.2.1.
In a later comment on Reddit, the host of Youtube’s The Tech of Tech channel stated that he decided to test Autopilot’s capabilities on rainy, foggy roads stretching across the historic Blue Ridge Parkway in North Carolina.
“I did adjust the maximum speed to accommodate visibility due to fog as Autopilot doesn’t know to just slow down- it will shut off if it can’t see far enough for the speed. However, by lowering the maximum speed to what I’d want to drive anyway, Autopilot remains engaged. Other than that, all steering, braking, acceleration, etc. is done by the car and not me,” the driver wrote.
One notable feature of the update focused on Autopilot’s capability to slow down for sharp curves, which results in safer navigation through abrupt turns. For example, on a right turn, the car will drive closer to the right edge of the pavement, which provides a safer environment if two vehicles are driving through the same curve in opposite directions.
Tech of Tech indicates that the addition of the “Slowing for Sharp Curves” feature was the only reason the car could navigate the tricky roads of the Blue Ridge Parkway, which it did in impressive fashion.
Tesla rolled out a similar feature last year, but there were some issues with its performance that made some drivers feel uncomfortable. The company continually improved the capability, which was then released with 2020.16.2.1 to owners around May 16 this year, according to TeslaFi.com.
“This update introduced new slowing behavior. Autopilot slows much more for tight curves than before, and it does so much more smoothly than the aborted slowing behavior introduced when V10 first released,” the host said.
As could be seen in the video of the torture test, Autopilot managed to navigate through the various challenging sections of the Blue Ridge Parkway for 79 straight minutes before the car took a turn a bit too sharply, which required the driver to intervene. This instance was the only mistake that the car’s software made in almost 46 miles of driving, and that’s in questionable weather conditions and on a stretch of road that challenges even human drivers.
Tesla’s Autopilot software continues to improve through the millions of miles of driving data that owners contribute to the company’s Neural Network. Tesla CEO Elon Musk indicated during the company’s Q1 2020 Earnings Call that the Neural Net’s training was coming along nicely and that contributions were being assessed to improve the safety and performance of the company’s driver-assist features.
“We are collecting data from over 1 million intersections every month at this point. This number will grow exponentially as more people get the update, and as more people start driving again. Soon, we will be collecting data from over 1 billion intersections per month. All of those confirmations are training on neural net, essentially, the driver when driving and taking action is effectively labeling — the labeling reality as they drive, and making the neural net better and better,” Musk said.
Watch The Tech of Tech‘s Tesla Model 3 navigate the Blue Ridge Parkway below.
News
Tesla’s Cybercab battery design could make robotaxis even cheaper
Tesla built the Cybercab’s battery pack with spare room for a future switch to LFP.
Tesla built the Cybercab’s battery pack with empty space inside it on purpose, and the reason points to a cheaper robotaxi down the road.
Robert Sumpf, Tesla’s director of mechanical design engineering for battery modules and packs, said the extra room beside and beneath the Cybercab’s modules was left there so the company can move to lithium iron phosphate cells later. “So, we protected this pack to go to an even lower energy dense chemistry as we scale and need to bring LFP into the platform,” Sumpf said in a Cybercab deep dive with Munro Live.
The Cybercab runs on Tesla’s own 4680 cells with a nickel chemistry today. LFP cells store less energy for their size, so a straight swap would cut range. The reserved space lets Tesla add more LFP cells to make up the difference. In return, Tesla would get cells that cost less, run cooler and survive more charge cycles. LFP became the most used EV battery chemistry in the world last year.
Not all of that room was planned for a chemistry change. Lars Moravy, Tesla’s Vice President of Vehicle Engineering, said part of it exists because “we didn’t need as many batteries as we thought we did.” Tesla told Sandy Munro in early 2025 that it was targeting a Cybercab pack under 50 kWh with close to 300 miles of range.
The pack also breaks from the structural design Tesla used in some Model Y and Cybertruck builds. Instead of potting hundreds of cells into one block, the Cybercab uses four separate, individually sealed modules, so a bad module can be replaced without scrapping the whole pack. Sumpf was blunt about the older approach, giving the Model Y structural battery “a grade D” because it was hard to build and nearly impossible to repair. He estimated a fully structural pack gains about 10% more cell volume, but said “the issue is the service cost.”
That cost matters more for a car that rarely sits parked. Cybercab chief engineer Eric Earley said replacing an entire pack “before half a million miles is a killer on cost,” which is why remanufacturing and serviceability were built into the design from the start.
The same Munro Live tour produced Moravy’s admission that the Cybercab line at Giga Texas is now outpacing Robotaxi deployments, along with a look at the robotaxi’s unorthodox paint process. Tesla is aiming to deliver the first Cybercabs to outside customers by the end of 2026, though Elon Musk has said the vast majority will stay owned by Tesla and run as a transport service.
Sumpf did not give a date for an LFP version, name a cell supplier or say how a chemistry change would affect range. For now, the 4680 cars keep piling up: Texas DMV records from Tuesday list 544 Cybercabs authorized for Tesla’s Robotaxi fleet, more than the 507 Model Ys on the same roster for the first time.
News
Starlink is ‘biggest competitive advantage,’ but it’s temporary, United CEO says
United Airlines CEO Scott Kirby said that SpaceX’s Starlink satellite internet is its leading competitive advantage in the U.S. airline industry. However, in recent comments, while he described the service as “the biggest competitive advantage” the carrier currently holds, he also acknowledged that the edge is temporary.
“…everybody eventually is going to have high-speed low earth orbit internet connectivity. It’s going to be a must-have. We are going to have a few years of being the only airline that has it at scale, and that’s our mission,” Kirby said. Until everyone has it, United aims to capitalize on its head start by converting trial users into long-term loyal customers.
🚨 United CEO Scott Kirby calls Starlink “the biggest competitive advantage,” but says it’s temporary.
“I think it’s the biggest competitive advantage. And it’s temporary. Like it’s not, everybody eventually is going to have high speed low earth orbit internet connectivity. It’s going to be a must-have. But we are going to have a few years of being the only airline that has it at scale, and that’s our mission.”
“I think we’re going to convince customers that United is the best airline and we’re going to get customers to switch their loyalty, to get our credit card, switch their loyalty to United. Not because of Starlink. They’re going to try us for Starlink and they’re going to stay with all the other technology, and everything else is better, and the technology in particular is so much better than technology other, and it’s accelerating because we’ve got the right tech stack, and because we’re like, our gap is getting wider and wider with our technology every day.”
Next: LAX becomes United’s first 100% Starlink hub in the first half of next year, and the United app will show your Starlink odds for each flight. https://t.co/hToyFWaWj7
— TESLARATI (@Teslarati) October 27, 2026
United has moved faster than its rivals on Starlink installations. As of mid-to-late October 2026, the system was on roughly 43 percent of its fleet, with plans to reach 61 percent by year-end, 87 percent by June 2027, and full fleet coverage by the end of next year.
The airline already operates more Starlink-equipped flights than the rest of the U.S. industry combined. Los Angeles International Airport is scheduled to become United’s first 100 percent Starlink hub in the first half of 2027, and the carrier’s app will display the likelihood that any given flight has the service.
Kirby has pointed to early evidence that the offering influences booking decisions. Some passengers loyal to other airlines are already switching flights for Starlink connectivity, something the airline, along with American Airlines, has tried to leverage against competing airlines. Both companies used status-match programs, and some of those customers are pursuing company-specific credit cards, which signals an intent to stay.
Once onboard, these customers encounter the airline’s broader investments in aircraft, larger seat-back screens, oversized bins, connection tools, and a mobile app that Kirby says outperforms competitors. Wi-Fi satisfaction scores on Starlink-equipped planes have been reported to be nearly double those on aircraft with other systems.
Competitors are moving slowly to do the same. Southwest Airlines began installations in 2025 and expects to complete them soon. Delta is pursuing Amazon’s Leo system, while American expects to start Starlink work in early 2027 and has not yet detailed regional-fleet plans. This timeline gap gives United a multi-year window in which it remains the only major U.S. carrier offering Starlink at scale.
Kirby has framed the strategy as pressing that advantage while simultaneously widening gaps in other technology areas so that customers who try United for the Wi-Fi remain for the overall experience.
The approach reflects a calculated bet that superior in-flight connectivity can accelerate share gains and support longer-term financial targets, including a 10 percent pre-tax margin goal.
High-speed internet is particularly valued by premium and business travelers who can work or stream reliably. As low-Earth-orbit service becomes ubiquitous, the temporary nature of the pure connectivity edge means United’s retention will depend on whether its other product and operational improvements keep pace.
News
Tesla Cybercab’s unorthodox paint process explained
Tesla Cybercab skips the traditional paint shop entirely, marking an unorthodox process for coloring the fully autonomous vehicle. However, the decision was a smart one, as it slashed factory costs, floor space utilization, and complexity, since Tesla relied on specialized alloys and in-mold coloring.
Munro Live was given the opportunity to do a factory walkthrough with Tesla Vice President of Vehicle Engineering Lars Moravy, who explained and detailed how Cybercab production at Giga Texas eliminates one of the automotive manufacturing industry’s most expensive and space-hungry elements.
A conventional paint line can cost hundreds of millions of dollars and require a million man-hours to build. Moravy framed the choice around Tesla’s core rule: “Question the requirements: why do you need paint?”
🚨 Super cool quote from the Cybercab breakdown on the decision to avoid traditional paint:
“One of the most expensive parts of the factory is a paint line… a few hundred million dollars, a million man hours to build one… rule number one at Tesla, question the requirements. Why do you need paint?… the alloys we use in the castings… don’t oxidize and reduce strength over time.” -@larsmoravy
Model Y and Cybertruck bodies are fully coated and painted. On Cybercab, only the side pieces get the anti-corrosion dip coat, and there’s no paint shop at all, which helps cut factory floor space per car by about 50%.
— TESLARATI (@Teslarati) October 26, 2026
The structural answer lies in material selection. The aluminum alloys used in Cybercab’s front and rear castings do not oxidize and lose strength over time, the way many common automotive aluminum alloys do. Only the side modules, built around hot-stamped boron-steel door rings, receive an anti-corrosion e-coat, or dip coat. Front and rear modules skip coating altogether.
Model Y and Model 3 bodies, by contrast, are fully coated and painted.
Exterior colors come from the plastics shop, which Tesla engineers described as “our paint shop” for the Cybercab. Body panels are produced via reaction injection molding, referred to as RIM. Color is then mixed into a polyurethane resin at the mix head immediately before injection, so the panel emerges already colored with a Class-A surface.
Robots trim overflow to control sink marks and flow lines. The process avoids conventional spray-paint defects like orange peel and paint bleed. Scratches can still be repaired with compatible coatings that behave “basically like paint.” Because color is integral rather than a surface layer, shallow scratches do not expose a contrasting undercoat.
The secret behind Tesla’s Cybercab Gold goes well beyond just the color
Tesla has also explored related approaches, including in-mold lamination and painted films on stamped panels.
The manufacturing payoff is substantial because removing the paint shop, combined with the “unboxed” modular process that builds and tests front, rear, side, and center modules in parallel before a short final assembly, cuts factory floor space per vehicle by about 50 percent.
Final assembly itself uses only about 25 automated stations, far fewer than the typical Model Y line, and the overall line runs at about 78 percent automation, Moravy said. The approach also eliminates volatile organic compound emissions associated with traditional spray painting for these panels and shortens the coloring cycle from hours to minutes.
Cybercab’s paint strategy is clearly less about aesthetics than about systematically deleting capital-intensive, energy-heavy, and space-consuming steps once material science and process design make them unnecessary. It is a concrete example of Tesla applying first-principles questioning to one of the auto industry’s most entrenched factory requirements.