News
SpaceX CEO Elon Musk reveals next-generation Starlink satellite details
SpaceX CEO Elon Musk has revealed the first technical details about the company’s next-generation Starlink ‘Gen2’ satellite design, confirming that it will far outmatch the current generation of satellites by almost every measure.
Speaking in an onsite interview and Starbase tour with YouTuber Tim Dodd (The Everyday Astronaut), Musk – largely unprovoked – revealed that SpaceX has already built at least one functional Starlink Gen2/V2.0 satellite prototype and shipped it to the South Texas Starship factory, where it is currently being stored. More importantly, Musk also provided the first direct specifications for the next-generation spacecraft, stating that each Starlink V2.0 satellite will weigh about 1.25 tons (~2750 lb), measure about seven meters (~23 ft) long, and be almost an order of magnitude more capable than the “Starlink 1” satellites they’ll ultimately supersede.
Almost ten months after SpaceX first revealed its updated plans for a next-generation, 30,000-satellite constellation, those details have confirmed a few key points of speculation about the future of Starlink.
The 1st #SpaceX #Starlink Gen 2 #satellite has been produced. It’s 7 meters long & 1.2 tons, @elonmusk says. Note: that’s 4-5X more massive than Gen 1!
Musk adds, the new version will be almost an order of magnitude more capable than Starlink 1 in terms of useful data throughput.— Stan Shull (@stanshull) May 26, 2022
Back in August 2021, I surmised that just like it has with Falcon 9, SpaceX would again try to optimize its new Starlink V2.0 satellite design to take maximum advantage of Starship’s launch performance. In an updated Starlink Gen2 filing, the company conveniently revealed that a version of the constellation optimized for Starship would be structured such that the rocket could launch an entire orbital plane (one ring of satellites spaced evenly around the Earth) in one go. In that constellation variant, all but ~500 (1.5%) of almost 30,000 spacecraft would be stationed in planes of 110 or 120 satellites, meaning that it was safe to assume that SpaceX meant that every Starship would nominally carry 110-120 satellites. Using Musk’s latest optimistic Starship performance estimate of 150 tons to low Earth orbit (LEO), that all but guaranteed that a Starship-optimized Starlink V2.0 satellite would weigh up to 1250 kilograms.
Musk has now explicitly confirmed that each Starlink V2.0 satellite will weigh… “about one and a quarter tons” or 1250 kilograms. Starlink V1.0 and V1.5 satellites weigh around 260 and 310 kilograms, respectively, meaning that Starlink V2.0 satellites will be about a bit more than four times heavier than V1.5 and a bit less than five times heavier than V1.0.
Musk also revealed that V2.0 satellites will be “almost an order of magnitude more capable than Starlink 1.” He refused to call that capability bandwidth or throughput, the traditional method of describing a communication satellite’s total performance, but Starlink V1.0 satellites are believed to have a total bandwidth of 18 gigabits per second (18 Gbps). As of today, it’s unknown if Starlink V1.5 – a significant upgrade – also added more bandwidth, nor if Musk was referring to that latest Starlink V1.x iteration. But even if he was comparing V2.0 with the earliest V1.0 satellites, it’s possible that each Starlink V2.0 satellite could add around 140-160 Gbps to the 30,000-satellite constellation.
Ultimately, specific numbers aren’t needed to emphasize the importance of the details Musk provided. If true, they mean that Starlink V2.0 will pack roughly twice as much usable bandwidth into a given unit of satellite mass compared to V1.x. Combined with the fact that Starship could offer ~10 times as much performance to LEO as Falcon 9, a single Starship launch could theoretically expand total network capacity roughly twenty times more than one Falcon 9 launch. For example, each Falcon 9 launch of 60 260-kilogram Starlink V1.0 satellites added about 1080 Gbps of instantaneous bandwidth to the constellation. A Starship launch of 120 1250-kilogram Starlink V2.0 satellites could add around 19,000 Gbps (19 terabits per second).
Even despite those massive advantages, SpaceX’s Starlink Gen2 ambitions still leave it no slack whatsoever. If the FCC approves its license request, SpaceX would need to launch half of the constellation within six years – equivalent to around 130 Starship launches or 22 Starship launches per year. In comparison, Falcon 9 – a rocket that’s ten times smaller, less reusable, and has been flying since 2010 – did not achieve 22 launches in one year until 2020. For Starship to have any hope of achieving the cadence Starlink Gen2 requires, SpaceX would have to ramp up launches of the largest rocket ever built at a truly miraculous pace and suffer very few failures or setbacks along the way.
As immense as the challenge may be, the potential rewards are just as high. A constellation of 30,000 Starlink V2.0 satellites – if spaced evenly around the Earth – could have a total bandwidth of ~1250 terabits per second (Tbps) available over land (excluding Antarctica) at any given second. Even if half of that bandwidth is needed for backhaul and routing, the total installed bandwidth of global internet infrastructure was estimated to be 600 Tbps in 2020. Starlink will always be bottlenecked by the number of satellites that can be simultaneously available over any single point on Earth, so the constellation will never be able to match a ground network 1:1 with the same installed capacity, but it’s safe to assume that Starlink Gen2 could serve tens or even hundreds of millions of users located anywhere on Earth if SpaceX is able to build it.
Elon Musk flew under the radar with his recent purchase of a $1 billion energy company, according to Federal Trade Commission (FTC) documents.
Transaction number 202612350 listed Tesla and SpaceX frontman Elon Musk as the acquiring party and CF APR Super Holdings LLC as the seller, with New APR Energy, LLC as the acquired entity. The deal, which closed without public announcement, came to light on May 14.
BREAKING: Elon Musk acquires Jacksonville power company APR Energy in a deal valued at more than $1,000,000,000.00.
— Polymarket Money (@PolymarketMoney) July 15, 2026
Analysts inferred the deal’s scale from minority stakeholder disclosures, including one report of a 5 percent interest sold for approximately $50.4 million. Fortress Investment Group had purchased APR’s assets in late 2024, rebranded the operation as New APR Energy, and subsequently transferred ownership to Musk.
APR Energy specializes in rapidly deployable power infrastructure. The company maintains one of the world’s largest fleets of mobile gas and diesel turbines, with more than 1.1 gigawatts of generation capacity. Its modular units, which are often trailer-mounted, enable turnkey installations ranging from 20 MW to over 500 MW.
APR provides full engineering, procurement, construction, operation, and maintenance services for behind-the-meter power plants, serving everything from data centers, utilities, and industrial clients.
The firm has expanded aggressively to meet surging demand, recently adding turbines and deploying over 100 MW for a major AI hyperscaler. Its solutions bridge critical gaps where grid interconnections face delays of two to five years, according to Yahoo.
The acquisition means something more for Musk. As he continues to expand projects in artificial intelligence, especially xAI, his AI venture, there is a greater need to supply energy-intensive supercomputing clusters, including the Colossus project, with what they need: reliable and high-capacity power.
Ownership of APR provides immediate access to flexible generation assets that can be deployed adjacent to data centers, reducing dependence on a strained infrastructure. It also complements Tesla’s energy storage business, so Musk will be able to pull from his own entities to address the rapid scaling demands of AI training and compute.
Tesla had a petition protesting a recall to fix a potential issue with 2017-2023 Model Y and Model 3 vehicles’ headlights was denied, as the National Highway Traffic Safety Administration (NHTSA) disagreed with the company’s opinion of things.
The recall covers approximately 19,917 Model Y and Model 3 vehicles built from 2017 to 2023. Tesla initially submitted a noncompliance report for the headlights on these vehicles on March 15, 2024. Tesla then petitioned for an exemption from the fix, which violated FMVSS No. 108 (40 CFR 571.108), arguing that the “noncompliance is inconsequential as it relates to motor vehicle safety.
🚨 Tesla was denied a petition by the NHTSA to avoid a recall of 19,900 2017-2023 Model 3 and Model Y vehicles.
The NHTSA found that the vehicles’ headlights may exceed maximum lighting levels. Tesla argued it was inconsequential and did not require a recall. pic.twitter.com/m8Jmm1teLL
— TESLARATI (@Teslarati) July 16, 2026
The NHTSA disagreed, stating that Tesla’s conclusion that the headlights do not increase any risk was not an opinion it shared. The agency said it disagreed with Tesla’s assumption that glare is not increased to surrounding traffic. This issue could be highlighted even more in certain weather conditions.
Tesla will be required to remedy the issue, the NHTSA ruled:
“In consideration of the foregoing, NHTSA has decided that Tesla has not met its burden of persuasion that the subject FMVSS No. 108 noncompliance is inconsequential to motor vehicle safety. Accordingly, Tesla’s petition is hereby denied, and Tesla is consequently obligated to provide notification of and free remedy for that noncompliance under 49 U.S.C. 30118 and 30120.”
The issue here appears to be the angle of the headlights and the brightness they emit during operation. The NHTSA report states that:
“Tesla’s headlamp supplier, Marelli Automotive Lighting, tested 25 right-hand and 25 left-hand lamps, and for this sample, found the maximum photometric intensity measured in the 10°U to 90°U and 90°L to 90°R zone was between 136.2 cd and 230.1 cd for the right-hand lamps and between 117.5 cd and 160.3 cd for the left-hand lamps. According to Tesla, these tests revealed that the photometric intensity of the right-hand and left-hand headlamp lower beam on the subject vehicles may measure as much as 230.1 cd in the 10°U to 90°U and 90°L to 90°R zone, exceeding the maximum photometric intensity by 105.1 cd. Additionally, Tesla states that a left-hand lamp tested by a Transport Canada recognized laboratory measured a maximum of 171.27 cd in the 10°U to 90°U and 90°L to 90°R zone. Despite these measurements exceeding the allowed photometric maximum of 125 cd, Tesla believes that the subject noncompliance is inconsequential to motor vehicle safety.”
Tesla also argued at some points that the headlights had not been deemed responsible for any complaints, accidents, or injuries related to the noncompliance.
Lifestyle
NTSB findings on fatal Tesla crash tell a very different story
The NTSB confirmed the driver, not Tesla’s FSD, caused the fatal Texas house crash.
The National Transportation Safety Board released preliminary findings Wednesday confirming that a Tesla driver, not the vehicle’s software, caused a fatal crash in Katy, Texas in June. The driver, 44-year-old Michael Butler, had engaged Full Self-Driving Supervised mode on Rose Hollow Lane, a residential street with a 30 mph speed limit, before manually overriding the system by pressing the accelerator pedal all the way to 100%. Data recovered from the 2025 Tesla Model 3 showed the vehicle was traveling over 70 miles per hour when it struck a home and killed 76-year-old Martha Avila, who was inside. Weather was clear, the road was dry, and it was daylight.
Texas man charged in fatal Tesla crash where he blamed Autopilot
Butler told authorities he had passed out at the wheel. But security camera footage obtained by the NTSB told a different story, and showed the car accelerating through an intersection before leaving the road entirely. Police also found that Butler’s phone had Google searches including the terms “Tesla FSD not aggressive enough 2026” and “Tesla FSD too timid,” raising serious questions about how he was using the system before the crash. Butler has since been charged with manslaughter. The victim’s family has filed a lawsuit against both Butler and Tesla, alleging negligence.
The NTSB findings aligned directly with what Tesla VP of AI Software Ashok Elluswamy had already stated publicly on X in the weeks after the crash, writing that “the driver manually overrode self-driving by pressing the accelerator all the way to 100%.” The data confirmed his account.
Yup. In this case, the driver manually overrode self-driving by pressing the accelerator all the way to 100% of the accel pedal in this residential area. They reached a speed of 73 mph during the crash, and had the accelerator pressed even after the crash.
— Ashok Elluswamy (@aelluswamy) June 22, 2026
Elon Musk secretly acquires $1B energy company to power the AI future
Tesla has to fix a big problem with its old headlights, NHTSA says
