News
SpaceX CEO Elon Musk explains how Starships will return from orbit
In the near future, SpaceX wants to begin putting its first two full-scale Starship prototypes through a series of increasingly challenging test flights, eventually culminating in their first Super Heavy-supported orbital launch attempts.
SpaceX CEO Elon Musk took to Twitter over the last 48 or so hours to answer a number of questions about how exactly Starship is meant to make it through orbital reentries – by far the most strenuous period for the ship and without a doubt the single most challenging engineering problem SpaceX must tackle.

Discussed yesterday on Teslarati, SpaceX technicians began the process of attaching numerous Tesla Model S/X battery packs to a subcomponent that will eventually be installed inside Starship Mk1’s nose, offering a storage capacity of up to 400 kWh. The need for all that power (Crew Dragon relies on a few-kWh battery) is directly related to Starship Mk1’s methods of reentry and recovery, recently described in detail by Elon Musk.
As noted above, ~400 kWh of batteries are needed to power the electric motors that will actuate Starship’s massive control surfaces – two large aft wings and two forward canards/fins. According to Musk, Starship’s “stability is controlled by (very) rapid movement of rear & fwd fins during entry & landing”, meaning that the spacecraft will need to constantly tweak its control surfaces to remain in stable flight.

By far the biggest challenge SpaceX faces is ensuring that Starship can survive numerous orbital-velocity reentries with little to no wear and tear, a necessity for Starship to be cost-effective. In Low Earth Orbit (LEO), Starship will be traveling no less than 7.8 km/s (Mach 23, 17,500 mph) at the start of atmospheric reentry. In simple terms, the process of slowing from orbital velocity to landing on Earth involves turning the vast majority of that kinetic energy into heat. As Musk noted yesterday, this reality is just shy of unavoidable but there is some flexibility in terms of how quickly one wants to convert that energy into heat.
The fastest route to Earth would involve diving straight into the atmosphere, dramatically increasing peak heating on a spacecraft’s surface to the point that extremely exotic heat shields and thermal protections systems become an absolute necessity. SpaceX wants to find a middle ground with Starship in which the spacecraft uses its aerodynamic control surfaces and body to generate lift, slowly and carefully lowering itself into Earth’s atmosphere over a period of 15+ minutes. Musk notes that this dramatically lessens peak heating at the cost of increasing the overall amount of energy Starship has to dissipate, a bit like cooking something in the oven at 300 degrees for 30 minutes instead of 600 degrees for 10 minutes.
To an extent, Starship’s reentry profile is actually quite similar to NASA’s now-retired Space Shuttle, which took approximately 30 minutes to go from its reentry burn to touchdown. Per the above infographic, it looks like Starship will take approximately 20 minutes from orbit to touchdown, owing to a dramatically different approach once it reaches slower speeds. Originally described by Musk in September 2018 and again in recent weeks, Starship will essentially stall itself until its forward velocity is nearly zero, after which the giant spacecraft will fall belly-down towards the Earth, using its wings and fins to maneuver like a skydiver. The Space Shuttle landed on a runway like a (cement-encased) glider.
This unusual approach allows SpaceX to sidestep the need for huge wings, preventing Starship from wasting far more mass on aerodynamic surfaces it will rarely need. The Space Shuttle is famous for its massive, tile-covered delta wing and the leading-edge shielding that partially contributed to the Columbia disaster. However, it’s a little-known fact that the wing’s size and shape were almost entirely attributable to US Air Force demands for cross-range performance, meaning that the military wanted Shuttles to be able to travel 1000+ miles during reentry and flight. This dramatically constrained the Shuttle’s design and was never once used for its intended purpose.

SpaceX thankfully doesn’t have its own “US Air Force” stand-in making highly consequential demands (aside from Elon Musk ?). Instead, Starship will continue the SpaceX tradition of vertical landing, falling straight down – a bit like a skydiver (or a brick) – on its belly and flipping itself over with fins and thrusters for a propulsive vertical landing. In this way, Starship doesn’t have to be a brick forced to fly, like the Shuttle was – it just needs to be able to stably fall and quickly flip itself from a horizontal to vertical orientation.
Additionally, Starship is built almost entirely out of steel, whereas the Shuttle relied on an aluminum alloy and needed thermal protection over every square inch of its hull. Steel melts at nearly twice the temperature of the Shuttle’s alloy, meaning that Starship will (hopefully) be able to get away with nothing more than ceramic tiles on its windward half, saving mass, money, and time. Once Starship completes its first 20 km (12.5 mi) flight test(s), currently scheduled no earlier than mid-October, SpaceX will likely turn its focus on verifying Starship’s performance at hypersonic speeds, ultimately culminating in its first orbital-velocity reentries.
Check out Teslarati’s Marketplace! We offer Tesla accessories, including for the Tesla Cybertruck and Tesla Model 3.
News
SpaceXAI signs agreement with Anthropic for massive AI supercomputer access
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.
News
Tesla unveils mysterious prototype at Giga Texas: Is the Model Y L coming to America?
The Model Y L has been available in China for some time, but Americans are wondering when it will potentially come to the United States, offering a larger version of the best-selling vehicle in the world, as the Model X is officially phased out.
Tesla unveiled a mysterious prototype, covered up between a Model Y and a Cybertruck at Gigafactory Texas, perhaps giving yet another hint that the Model Y L is coming to America.
The Model Y L has been available in China for some time, but Americans are wondering when it will potentially come to the United States, offering a larger version of the best-selling vehicle in the world, as the Model X is officially phased out.
Giga Texas observer and drone operator Joe Tegtmeyer captured an image of the vehicle on May 6, showing a fully-covered prototype parked alongside a standard Model Y and a Cybertruck.
This mystery Tesla is covered at Gigafactory Texas
What do you think it is? https://t.co/l5WVKLi9yM pic.twitter.com/CcOybDkCkn
— TESLARATI (@Teslarati) May 6, 2026
From top-down and angled views, the prototype appears nearly identical in scale to the Model Y but reveals noticeably distinct rear proportions—an elongated rear door that stretches farther over the wheel arch and rear glass that flows uninterrupted to the spoiler lip.
The side-by-side placement provides an immediate size reference. The mystery vehicle sits comfortably between the compact Model Y and the massive Cybertruck, suggesting it occupies a practical middle ground for families seeking more interior room without jumping to a full-size pickup.
Enthusiasts quickly took to social media with guesses ranging from an extended-wheelbase Model Y to a potential station-wagon variant.
The sight of this prototype follows an earlier look at another shrouded body-in-white resting in a wooden shipping crate at the Giga Texas plant in late March.
That prototype appeared to display an elongated silhouette. Some analysis seems to show nearly exact dimensions as to what is reported for the Model Y L in the Chinese market, approximately 4.98 meters long with a 3.04-meter wheelbase, roughly seven inches longer overall than the U.S.-spec Model Y. The rear-door extension and glass-to-spoiler design were identical to the current sighting:
The Model Y L has already proven popular in China, where it launched in six- and seven-seat configurations and quickly ranked among the top-selling mid-to-large SUVs. Owners enjoy roughly 10 percent more cargo space and enhanced family versatility.
Tesla has remained silent on U.S. plans other than CEO Elon Musk saying it could come in late 2026, but localizing production at Giga Texas would make strategic sense.
With the Model X phase-out and steady Model Y output already humming along expanded lines, a longer-wheelbase variant could add tens of thousands of annual deliveries without major retooling.
The latest sighting arrives amid Tesla’s broader push to refresh its lineup. Whether this prototype represents the long-rumored Model Y L, a subtle Juniper-style update, or something entirely new remains unconfirmed.
Yet the consistent visual cues—precise dimensional match, distinctive rear styling, and strategic placement at Giga Texas—point strongly toward an extended Model Y designed for American families who want extra space without sacrificing the Model Y’s efficiency and affordability.Tesla watchers will be monitoring future drone flights closely.
If the prototype is indeed the Model Y L, it could mark a significant expansion of the company’s best-selling vehicle and deliver the extra room many U.S. buyers have been requesting for years. For now, the blue tarp keeps its secrets—but the clues are getting harder to hide.
News
Tesla Roadster gets an update, but not the one fans were looking for
Tesla has quietly filed a new trademark application for its next-generation Roadster, giving enthusiasts their first official glimpse of fresh branding for the long-teased electric supercar.
Tesla has been slow to show its hand regarding the massive project that is the Roadster, but it is now coming forth with a new update.
However, it is probably not the one fans were looking for.
Tesla has quietly filed a new trademark application for its next-generation Roadster, giving enthusiasts their first official glimpse of fresh branding for the long-teased electric supercar.
The February 3 filing includes an inverted triangular badge with the word “ROADSTER” centered above four vertical lines that, according to the application, represent “speed, propulsion, heat, or wind.”
A sleek, angular wordmark and a minimalist curved-line silhouette hinting at the car’s aerodynamic shape round out the trio of marks.
I found something cool. Tesla has filed a new trademark application for its next-generation Roadster. It could be the new Roadster logo/badge.
The filing says the lines depict speed, propulsion, heat or wind.
(I took the liberty of making the logo red. Trademark filings are… pic.twitter.com/W9JSDwTRL7
— Sawyer Merritt (@SawyerMerritt) May 6, 2026
For a program that began with Elon Musk’s 2017 reveal, this is tangible forward motion. The original Roadster proved EVs could be thrilling; the next generation aims higher, with promises of sub-two-second 0-60 mph acceleration and, in its most extreme configuration, optional SpaceX cold-gas thrusters for rocket-like thrust.
The new trademarks suggest Tesla is now locking down the visual identity that will accompany those headline specs, as well as a small hint that maybe we’re finally getting close. However, the company has not revealed any progress on the vehicle itself or its specs to the public.
It continues to tease with developments like this one.
That said, the update lands with a familiar bittersweet note. Fans have waited nearly a decade since the initial unveiling. Production was once eyed for 2020, then 2021, then later still. In the intervening years, Tesla has delivered the Model Y, Cybertruck, Semi, and major autonomy advances while scaling its energy business.
The Roadster has taken a back seat, and the delays have been genuinely disappointing. Many longtime supporters have grown frustrated watching renderings and hearsay while other marques roll out ever-faster electric sports cars.
Yet, the Roadster program itself still sparks genuine excitement. It represents the purest expression of Tesla’s “accelerate the world’s transition to sustainable energy” mission—pushing performance boundaries to prove EVs can outperform anything with an engine.
The new branding, modest as it is, keeps that promise alive. It tells owners and prospective buyers that Tesla hasn’t forgotten the car that started it all.
No one would blame fans for wanting more than a logo right now. But in an industry where many concepts never leave the drawing board, the fact that Tesla continues to invest in and protect the Roadster’s identity is reason for measured optimism.
The wait has tested patience, but when the next-generation Roadster finally arrives, the new badge may well adorn one of the most exciting cars ever built. For those who have followed the journey this far, that payoff still feels worth it.