News
SpaceX Starship prototype charred but intact after catching fire [photos]
SpaceX teams have finally safed the fourth full-scale Starship prototype nearly two days after a Raptor engine test caught it on fire, an anomaly that left the massive rocket charred and damaged – but still intact.
While SN4’s survival is a welcome and unexpected outcome, the fire that broke out near the base of the rocket caused damage that will have to be repaired, while the fault that allowed that fire to occur in the first place will also need to be rectified. Had the same events transpired during the ship’s inaugural flight test, things could have gone even further south after the rocket lifted off, carrying it away from remotely-controlled water jets used to suppress unplanned fires on the pad.
Thankfully, SpaceX’s focus on testing, testing, and testing some more meant that Starship SN4’s minor self-immolation occurred on the ground when the stakes – while high – were much lower than they would have been with an airborne rocket. The problems uncovered will, of course, need to be fixed, inevitably delaying the ship’s first flight test, but odds are that SN4 now has a better shot at success thanks to those hiccups.
Thanks to the fact that Starships are constructed almost entirely out of steel, a little (or a lot of) fire shouldn’t theoretically be much of a problem. However, SpaceX has taken a rather freeform approach to its early Starship SNx prototypes, opting to bolt, weld, or tape on the vast majority of external hardware with little or no protection from the elements, including fires ignited by the ships themselves.
With SN4, it appears that the pressure jump experiences immediately after Raptor ignition (the ship’s third such test) shook some methane-related plumbing loose. Raptor continued to burn for another five or so seconds after that minor failure, shutting down as planned – but not before it ignited the methane the burst pipe was leaking. Additionally, after that new plume of boiling liquid methane caught fire, the fire it sustained proceeded to ignite insulation wrapped around the rocket’s launch. It burned vigorously, likely helping to damage wiring, ultimately causing SpaceX to partially lose control of the rocket and preventing attempts to inspect and fix the damage for two full days.
It’s safe to say that SpaceX is probably going to prioritize avoiding the series of events that caused May 19th’s anomaly from here on out, considering that things could have easily gone much worse. Thankfully, whatever control SpaceX or the rocket itself retained after wire damage allowed it to safely offload its flammable propellant and vent expanding gases to prevent SN4’s tanks from bursting. Installing highly flammable insulation approximately 10 feet away from an active Raptor engine and giant controlled fire and explosion was also inadvisable and probably won’t be repeated.
Thankfully, the damage is clearly minimal and Starship SN4 survived the ordeal otherwise unscathed. With any luck, inspections and repairs will be quick and easy and SpaceX – as NASASpaceflight reporter Michael Baylor notes – will be able to complete an identical static fire test without starting a fire on Starship SN4. SpaceX has requested a new road closure (signifying planned testing) on May 28th with backup windows on May 29th and June 1st.
Thanks to Starship SN4’s unplanned delays, it now looks quite likely that SpaceX’s next full-scale Starship prototype (SN5) will be completed – or nearly so – by the time that its predecessor is cleared for flight. “Too many Starships” is certainly a welcome problem to have.
News
Swedish unions consider police report over Tesla Megapack Supercharger
The Tesla Megapack Supercharger opened shortly before Christmas in Arlandastad, outside Stockholm.
Swedish labor unions are considering whether to file a police report related to a newly opened Tesla Megapack Supercharger near Stockholm, citing questions about how electricity is supplied to the site. The matter has also been referred to Sweden’s energy regulator.
Tesla Megapack Supercharger
The Tesla Megapack Supercharger opened shortly before Christmas in Arlandastad, outside Stockholm. Unlike traditional charging stations, the site is powered by an on-site Megapack battery rather than a direct grid connection. Typical grid connections for Tesla charging sites in Sweden have seen challenges for nearly two years due to union blockades.
Swedish labor union IF Metall has submitted a report to the Energy Market Inspectorate, asking the authority to assess whether electricity supplied to the battery system meets regulatory requirements, as noted in a report from Dagens Arbete (DA). The Tesla Megapack on the site is charged using electricity supplied by a local company, though the specific provider has not been publicly identified.
Peter Lydell, an ombudsman at IF Metall, issued a comment about the Tesla Megapack Supercharger. “The legislation states that only companies that engage in electricity trading may supply electricity to other parties. You may not supply electricity without a permit, then you are engaging in illegal electricity trading. That is why we have reported this… This is about a company that helps Tesla circumvent the conflict measures that exist. It is clear that it is troublesome and it can also have consequences,” Lydell said.
Police report under consideration
The Swedish Electricians’ Association has also examined the Tesla Megapack Supercharger and documented its power setup. As per materials submitted to the Energy Market Inspectorate, electrical cables were reportedly routed from a property located approximately 500 meters from the charging site.
Tomas Jansson, ombudsman and deputy head of negotiations at the Swedish Electricians’ Association, stated that the union was assessing whether to file a police report related to the Tesla Megapack Supercharger. He also confirmed that the electricians’ union was coordinating with IF Metall about the matter. “We have a close collaboration with IF Metall, and we are currently investigating this. We support IF Metall in their fight for fair conditions at Tesla,” Jansson said.
News
Tesla HW4.5 spotted in new Model Y, triggers speculation
Owners taking delivery of recent Model Y builds have identified components labeled “AP45.”
Tesla’s Hardware 4.5 computer appears to have surfaced in newly delivered Model Y vehicles, prompting fresh speculation about an interim upgrade ahead of the company’s upcoming AI5 chip.
Owners taking delivery of recent Model Y builds have identified components labeled “AP45,” suggesting Tesla may have quietly started rolling out revised autonomy hardware.
Hardware 4.5 appears in new Model Y units
The potential Hardware 4.5 sighting was first reported by Model Y owner @Eric5un, who shared details of a Fremont-built 2026 Model Y AWD Premium delivered this January. As per the Model Y owner, the vehicle includes a new front camera housing and a 16-inch center display, along with an Autopilot computer labeled “AP45” and part number 2261336-02-A.
The Tesla owner later explained that he confirmed the part number by briefly pulling down the upper carpet liner below the Model Y’s glovebox. Other owners soon reported similar findings. One Model Y Performance owner noted that their December build also appeared to include Hardware 4.5, while another owner of an Austin-built Model Y Performance reported spotting the same “AP45” hardware.
These sightings suggest that Tesla may already be installing revised FSD computers in its new Model Y batches, despite the company not yet making any formal announcements about Hardware 4.5.
What Hardware 4.5 could represent
Clues about Hardware 4.5 have surfaced previously in Tesla’s Electronic Parts Catalog. As reported by NotATeslaApp, the catalog has listed a component described as “CAR COMPUTER – LEFT HAND DRIVE – PROVISIONED – HARDWARE 4.5.” The component, which features the part number 2261336-S2-A, is priced at $2,300.00.
Longtime Tesla hacker @greentheonly has noted that Tesla software has contained references to a possible three-SoC architecture for some time. Previous generations of Tesla’s FSD computer, including Hardware 3 and Hardware 4, use a dual-SoC design for redundancy. A three-SoC layout could allow for higher inference throughput and improved fault tolerance.
Such an architecture could also serve as a bridge to AI5, Tesla’s next-generation autonomy chip expected to enter production later in 2026. As Tesla’s neural networks grow larger and more computationally demanding, Hardware 4.5 may provide additional headroom for vehicles built before AI5 becomes widely available.
Elon Musk
Elon Musk’s Grokipedia is getting cited by OpenAI’s ChatGPT
Some responses generated by OpenAI’s ChatGPT have recently referenced information from Grokipedia.
Some responses generated by OpenAI’s ChatGPT have recently referenced information from Grokipedia, an AI-generated encyclopedia developed by rival xAI, which was founded by Elon Musk. The citations appeared across a limited set of queries.
Reports about the matter were initially reported by The Guardian.
Grokipedia references in ChatGPT
Grokipedia launched in October as part of xAI’s effort to build an alternative to Wikipedia, which has become less centrist over the years. Unlike Wikipedia, which is moderated and edited by humans, Grokipedia is purely AI-powered, allowing it to approach topics with as little bias as possible, at least in theory. This model has also allowed Grokipedia to grow its article base quickly, with recent reports indicating that it has created over 6 million articles, more than 80% of English Wikipedia.
The Guardian reported that ChatGPT cited Grokipedia nine times across responses to more than a dozen user questions during its tests. As per the publication, the Grokipedia citations did not appear when ChatGPT was asked about high-profile or widely documented topics. Instead, Grokipedia was referenced in responses to more obscure historical or biographical claims. The pattern suggested selective use rather than broad reliance on the source, at least for now.
Broader Grokipedia use
The Guardian also noted that Grokipedia citations were not exclusive to ChatGPT. Anthropic’s AI assistant Claude reportedly showed similar references to Grokipedia in some responses, highlighting a broader issue around how large language models identify and weigh publicly available information.
In a statement to The Guardian, an OpenAI spokesperson stated that ChatGPT “aims to draw from a broad range of publicly available sources and viewpoints.” “We apply safety filters to reduce the risk of surfacing links associated with high-severity harms, and ChatGPT clearly shows which sources informed a response through citations,” the spokesperson stated.
Anthropic, for its part, did not respond to a request for comment on the matter. As for xAI, the artificial intelligence startup simply responded with a short comment that stated, “Legacy media lies.”
Swedish unions consider police report over Tesla Megapack Supercharger
Tesla HW4.5 spotted in new Model Y, triggers speculation
