News
SpaceX scraps Florida Starship Mk2 prototype
SpaceX has scrapped the lone Starship prototype built in Florida in 2019, surprising very few.
Beginning a few months after work began on Starship Mk1 at SpaceX’s South Texas production facilities, a separate team in Cocoa, Florida was tasked with building a similar Starship Mk2 prototype. Not much is known about Mk2 relative to its much more publicized sibling but unofficial photos and videos taken over the course of 2019 suggested that SpaceX had effectively completed most of Starship Mk2 by the end of last year. However, built dozens of miles and several waterways away from a practical test facility, actually testing a Starship prototype assembled at SpaceX’s Cocoa facilities was always going to be an uphill battle.
To warrant the cost and effort that would be required to transport something as large as a vertical Starship from Cocoa, Florida to Cape Canaveral, Mk2 would have to be able offer something invaluable during testing. Now eight months after Starship Mk1 was destroyed during one of its first real tests, that was sadly not the case and SpaceX has chosen the simplest route forward – scrapping Mk2 where it sits.
In November 2019, SpaceX installed Starship Mk1 on a test stand in Boca Chica, Texas and began a series of tests. The ship passed an initial ambient temperature pressure test on the 18th but failed spectacularly during its first cryogenic proof test, said by SpaceX to have “pressurize[d] systems to the max.” Excluding Starhopper, Starship Mk1 was about as rough of a prototype as SpaceX could have feasibly built and the fact that it survived any length of time under cryogenic loads and pressures was fairly impressive.
Welded together almost entirely out in elements on the South Texas Gulf coast, the total success of Starship Mk1 (and its similar Mk2 sibling) would have flown in the face of almost every single tenet of modern aerospace production. As noted in a Teslarati article describing the Starship’s demise, the Mk1 production apparatus left plenty of room for improvement.
“[Videos of the failure implicated] the weld connecting the LOX dome to the cylindrical body of Starship’s LOX tank, pointing to a bad weld joint as the likeliest source of the failure. Although that hardware failure is unfortunate, Mk1’s loss will hopefully guide improvements in Starship’s design and manufacturing procedures.”
Teslarati.com — November 20th, 2019
That is precisely what SpaceX did – and was likely already doing – in response to Mk1’s failure. Just two months later, SpaceX successfully tested a steel Starship tank built in upgraded facilities with upgraded methods and reached pressures of 7.1 bar (~103 psi) before failing – likely a 50% improvement or better relative to Mk1. A second tank completed weeks later in late January 2020 reached 7.5 bar, sprung a leak, was repaired, and ultimately soared to 8.5 bar (~125 psi) before failing. Per CEO Elon Musk, that would technically be enough for a Starship to launch humans into orbit with an industry-standard ~40% safety factor.
Finally, SpaceX recently proved that a full-scale, two-tank Starship prototype built with the same methods and facilities as those test tanks could achieve the same results, completing a ~7.5 bar (~110 psi) cryogenic proof test with Starship SN4 on May 10th.
Long story short, the methods SpaceX used to build Starship Mk1 and Mk2 were already proven redundant more than six months ago and buried even deeper in May 2020. Aside from serving as a museum piece, Starship Mk2’s fate was sealed – the only real question was how and when it would be scrapped. For now, SpaceX’s Starship program will be almost exclusively stationed in South Texas, where it appears to be in good hands. Starship SN5 is currently expected to attempt its first wet dress rehearsal (WDR) and static fire tests no earlier than July 17th (today) at 8 am CDT (13:00 UTC).
Check out Teslarati’s Marketplace! We offer Tesla accessories, including for the Tesla Cybertruck and Tesla Model 3.
Tesla revealed plans to develop its own maps internally to fix Navigation issues and give owners more control over their routing.
On the company’s Q3 2026 Earnings Call, a question was asked about FSD Navigation issues, particularly speed limit recognition, parking, and routing. These are issues that surface from version to version and seem to be seriously challenging for Tesla’s AI team to resolve.
Tesla’s Head of AI, Ashok Elluswamy, said in response to the question:
“You’re totally right that we shouldn’t be depending on external people’s maps. So we are building our own maps, but not like the same way as OneMoreMap. We are building an entirely new architecture with essentially AI-based mapping and navigation.”
He continued:
“It’s going to be super awesome. It’s going to be integrated with voice. You can talk to it, ask for different routes. It’s going to learn your personalized preferences, plus the fleet preferences. The collective intelligence of all the cars is going to flow into making the maps and navigation super cool.”
Elluswamy said this issue should be confronted, and the mapping should be available for people by the time the v15 release starts deploying to customers.
“And over the next several months,” he added, “it’s just going to take over all of the navigation functions, parking, destination handling. And it will know you better than you know yourself in some regard for the driving task, and then handle it fully.”
Tesla Summon got insanely good in FSD v14.3.2 — Navigation? Not so much
Navigation is perhaps the most notable complaint with FSD, and although it has gotten better, it is still flawed. Much of the time, FSD will take people on routes that are not preferred by the driver, but Tesla tried to combat this with a routing feature in the Summer 2026 Update.
Tesla currently relies on Google for map data, but it is sort of counterintuitive to believe the company would rely on a third party for this. These improvements will be welcome, as routing and navigation have been such a pain point for so many owners for so long.
Quotes via Motley Fool
Tesla Full Self-Driving (Supervised) showed a record attach rate at delivery for new vehicles in the third quarter of 2026.
75 percent of new vehicle deliveries in North America included a subscription to FSD software, up from 55 percent in the prior quarter, a 20-percentage-point sequential gain in the company’s most mature market.
Tesla Q3 2026 Earnings results: in-line revenues, slight beat on EPS
Tesla shed light on the Q3 Shareholder Deck that 75 percent of new deliveries came with an FSD subscription:
“FSD (Supervised)⁴ attach rate continued to grow, reaching over 75% of new vehicle deliveries in North America – our most mature market – a nearly twenty percentage point sequential increase. We believe that FSD is driving demand in markets where it is available, and we continue to work with regulators to get it approved in other markets. “
The higher attach rate accompanied strong growth in the overall FSD consumer base, as active subscriptions rose to 1.75 million, a 67 percent year-over-year increase. For the first time, more than half of paid customers are on the monthly subscription, which is priced at $99 per month in the U.S.
People are buying FSD with a car attached, not the other way around
— Elon Musk (@elonmusk) October 21, 2026
The transition to subscriptions has accelerated since Tesla discontinued the one-time purchase option in most markets earlier this year. The recurring model improves revenue visibility and margins compared with large upfront payments. With three-quarters of North American deliveries now including FSD, the software is now an evidently large contributor to Tesla’s economics and ancillary revenue.
Geographic expansion remains important for further gains. FSD is now available in nine countries in the EU, and more than 100,000 customers have driven over 160 million kilometers. Tesla is pursuing broader EU-wide approval, with a related vote expected in December. Similar progress in additional markets could support higher attach rates beyond North America.
These results came as Tesla had a strong quarterly delivery performance of 486,532 vehicles in the quarter, which came with its largest passenger-vehicle backlog in four years. Battery supply continued to limit volume, but the company has pointed to FSD as a clear difference-maker that is helping sustain demand. As supervised capabilities improve, the attach rate will likely continue to improve.
At face value, it is a key measure of how Tesla is converting sales into autonomy progress.
Update: 10/22 10:51a – Para 1 consumers changed to “customers” for accuracy.
Tesla Cybercab units finally got a definitive date for when they will be able to be purchased by customers, ending speculation regarding whether the fully autonomous vehicle would be available in the near term.
During Tesla’s third-quarter 2026 earnings call on October 21, Vice President of Engineering Lars Moravy said the company is aiming to begin delivering its purpose-built Cybercab robotaxi to customers before year-end.
🚨 In short, Cybercabs are slated to go to customers by end of the year. All units are going into Robotaxi fleet currently, per @larsmoravy
Cybercab will be “transport-as-a-service,” but vast majority of Cybercab will be “Tesla-owned,” @elonmusk says https://t.co/R8Qu8QWdmp
— TESLARATI (@Teslarati) October 21, 2026
“We’re trying to get Cybercabs to customers by the end of this year, 2026,” Moravy said. He added that production is underway on the line at Gigafactory Texas, but “all of those are currently going to our own internal Robotaxi fleet.”
Moravy explained that as operations scale, Tesla will “mainly deliver fully designed autonomous vehicles to our customers,” with the Cybercab as the first such vehicle. Initial customer deliveries will start small while the company gains experience with delivery logistics, ownership models, fleet management, and operating autonomous vehicles outside its direct control.
CEO Elon Musk reinforced the limited nature of private sales, describing the Cybercab primarily as a “transport-as-a-service” offering rather than a personal ownership vehicle like the Model 3, Model Y, or upcoming Roadster. “The vast majority of the Cybercab stuff will be Tesla-owned and transport-as-a-service,” Musk said.
The Cybercab is a two-seater vehicle developed by Tesla without a steering wheel or pedals. It relies completely on Tesla’s Full Self-Driving suite and has been in production since February. Early units have been put in Tesla’s Robotaxi service as of early September, when the company launched rides to the public. Tesla previously indicated it would aim to sell Cybercabs for about $30,000.
Unsupervised Full Self-Driving is expected to ship to consumer cars during Q2 2027, Ashok Elluswamy, Tesla’s Head of AI, indicated during the call. Early buyers or fleet operators may receive vehicles whose autonomous operation remains geographically restricted due to lagging regulatory agencies. There are some hurdles that remain ahead of Cybercab deliveries.
Tesla is building its own maps to fix Navigation issues
Tesla buyers are buying for Full Self-Driving and not for the car, data shows
