News
SpaceX installs Mechazilla ‘claw’ on Starship launch tower
In the latest chapter of SpaceX’s Starship launch tower chronicle, the company has rolled a claw-like component to the pad and attached the device to the end of the tower’s newly installed Starship ‘quick disconnect’ arm.
A couple months ago, as SpaceX’s newest Starship prototype and first orbital-class vehicle first started to come together, it became clear that the company was implementing a significant design change starting with Ship 20 (S20). Contrary to five years of official Starship/BFR/ITS updates in which the ship (second stage) was expected to connect to pad ground systems (power, propellant, gases, communications) through the booster (first stage), Starship S20’s umbilical panel was instead conspicuously installed on the exterior of the ship’s hull.
Later on, in an interview and tweets, it became clear that the move away from longstanding ship-to-booster umbilical plans was part of CEO Elon Musk’s latest crusade: moving parts and complexity from Starship and Super Heavy to the launch pad at any cost. As a result, rather than adding a little extra weight to Super Heavy and likely reducing total payload to orbit by a percent or two for an extremely simple, protecting umbilical solution, SpaceX would instead have to implement a massive swinging arm that would reach out from Starship’s launch tower to connect it to pad systems.
While it’s hard to say if that decision and the major design changes it’s entailed will prove to be the right move, what is undeniable is how rapidly SpaceX turned on a dime to build and install the extremely complex mechanisms required. Assembly of what has come to be known as Starship’s tower quick-disconnect (QD) arm really only began in early July. Less than two months later, the finished base of that arm was lifted around halfway up the ~145m (~475 ft) tall launch tower and affixed to two sturdy hinges.
Three weeks later, a new and initially mysterious structure SpaceX began assembling around August 20th has also been finished and installed at the end of the QD arm – adding an actuating tip and apparent stabilization ‘claw’ to the already massive swinging structure. Aside from a bit of plumbing and wiring, the only thing that arm now appears to be missing is the actual quick-disconnect umbilical panel that will allow it to temporarily connect to Starships to deliver power, propellant, and connectivity.
That quick-disconnect mechanism will likely sit directly on top of the brand new claw and stand several meters tall to span the gap between the top of Super Heavy and Starship’s umbilical panel. Two large, actuating arms at the bottom of the arm’s tip will be able to grab Super Heavy, stabilizing the massive booster during Starship installation. Once firmly installed on top of the booster, the claw’s missing quick-disconnect mechanism will then move in to connect to Starship.
Of course, the quick disconnect arm is just one of – and the most minor of – three massive ‘Mechazilla’ arms destined for the launch tower. Just a few hundred feet to the west, SpaceX is hard at work fabricating and assembling two far larger tower ‘catch’ arms and the cradle-like frame they’ll eventually attach to. While they will also give SpaceX far more flexibility to stack and manipulate Super Heavy and Starship in high winds and less than optimal weather conditions, the ultimate purpose of those arms is to catch Super Heavy boosters (and, maybe one day, Starships). According to a new contributor to NASASpaceflight forums, those Mechazilla catch arms could be installed as early as “this weekend or next week.”
SpaceX confirmed today that it has officially signed its third massive compute deal, providing compute at its Colossus data center in Southaven, Tennessee.
Reflection AI will gain immediate access to NVIDIA GB300 chips at SpaceX’s Colossus 2 data center. In return, Reflection will pay SpaceX $150 million per month starting on July 1, with total payments reaching approximately $6.3 billion if the contract runs through its duration, which is until 2029. Either party can terminate the agreement with 90 days’ notice after the initial three-month period.
CNBC first reported the deal.
🚨 SpaceXAI has agreed to a new compute deal with Reflection AI.
Reflection gets access to NIVIDIA GB300s, and will pay $150M per month to SpaceXAI for the compute. pic.twitter.com/bNPare8U5u
— TESLARATI (@Teslarati) June 22, 2026
This latest partnership highlights SpaceX’s strategy of commercializing its massive Colossus supercomputing infrastructure, originally developed to power Elon Musk’s Grok AI models. The company has rapidly expanded its customer base in the AI sector following its February 2026 merger with xAI, a transaction that valued the combined entity at $1.25 trillion.
SpaceX has previously signed significant compute deals with other major players.
It granted Anthropic exclusive access to the full capacity of its Colossus 1 data center, which exceeds 300 megawatts and includes over 220,000 NVIDIA GPUs. Details from SpaceX’s IPO filings indicate Anthropic will pay $1.25 billion per month through May 2029, potentially generating around $45 billion over the term of the deal.
Additionally, Google agreed to pay SpaceX $920 million per month for compute capacity from October 2026 through June 2029. This 32-month period will provide Google access to roughly 110,000 NVIDIA GPUs, along with supporting processors and memory. Capacity ramps up through September at a reduced fee, with termination options after the first year.
SpaceXA also established arrangements for computing power with Cursor, an AI coding startup. SpaceX acquired them in a $60 billion all-stock deal.
These arrangements position SpaceX’s collective position as an AI infrastructure powerhouse with high-margin revenue potential. The Google deal alone could generate nearly $29.5 billion over its term, while the Reflection contract adds another $6.3 billion.
Combined with the Anthropic arrangement, SpaceX stands to realize tens of billions in revenue from compute leasing in the coming years, which diversifies beyond SpaceX’s traditional rocket launches and Starlink operation.
The deals underscore growing demand for advanced AI training and inference capacity amid chip shortages and surging model development needs. Reflection, valued at $25 billion and focused on “American open intelligence” with government and national security ties, cited recent restrictions on closed models as validation for open-source approaches.
For SpaceX, the partnerships transform capital-intensive data centers into flexible revenue sources while supporting its broader AI ambitions after the company has gone public.
It is safe to say SpaceX won’t be going for electric rockets anytime soon.
In a characteristically blunt reply on X, SpaceX frontman Elon Musk stated, “Unfortunately, electric rockets are impossible,” following reports that MSCI had assigned SpaceX its lowest possible ESG rating of CCC.
The assessment, issued just this past week, coinciding closely with SpaceX’s public market debut, placed the company on par with nations like Russia in sustainability scoring and cited significant risks in environmental, social, and governance areas.
MSCI flagged SpaceX’s exposure to rocket emissions and other operational impacts, alongside governance concerns such as concentrated control by Musk and limited shareholder protections. Musk’s terse comment directly addressed the environmental pillar, underscoring a core physical constraint that ESG frameworks often overlook when evaluating high-thrust industries.
Unfortunately, electric rockets are impossible
— Elon Musk (@elonmusk) June 21, 2026
Electric propulsion systems do exist and are widely used in space. Ion thrusters and Hall-effect thrusters accelerate ionized propellant, typically xenon or krypton, using electric fields, achieving very high specific impulse, often exceeding 3,000 seconds compared to roughly 300–450 seconds for chemical rockets.
This efficiency makes them ideal for satellite station-keeping, orbit raising, and deep-space missions where low thrust over long durations is sufficient. SpaceX’s own Starlink satellites employ electric propulsion for these purposes.
However, launching from Earth’s surface demands something entirely different: enormous thrust delivered rapidly to overcome gravity and atmospheric drag. A typical orbital-class booster must generate thrust far exceeding its weight, often in the millions of Newtons within seconds.
Chemical rockets achieve this through exothermic combustion of dense propellants, producing high-mass-flow, high-velocity exhaust. Electric systems, by contrast, expel very small amounts of mass at extremely high speeds. Generating equivalent thrust would require impractical onboard power levels, massive energy storage or generation systems, and prohibitive added mass, rendering the approach infeasible with current or near-term technology.
Musk has previously expressed a similar sentiment, noting a desire for electric orbital rockets while acknowledging the inescapable requirements of Newton’s third law and energy delivery. The distinction is clear: electric propulsion excels once a vehicle is already in space; it cannot replace the high-thrust chemical phase required to reach orbit from the ground.
The episode illustrates broader critiques of ESG ratings. Proponents argue they incentivize better risk management and long-term sustainability. Detractors, including Musk—who has previously called ESG a “scam”—contend that such metrics can penalize essential activities when no practical alternative exists, potentially discouraging innovation in sectors like space access.
Elon Musk dubs the S&P 500 ESG as “outrageous scam” after Tesla gets booted from index
SpaceX has sought to mitigate launch-related impacts through reusability: Falcon 9 boosters have flown more than 30 times in some cases, dramatically lowering the manufacturing and emissions burden per kilogram delivered to orbit. Starship’s design further emphasizes rapid reusability and methane propellant, which can theoretically be produced via sustainable pathways.
Ultimately, Musk’s remark serves as a reminder that certain engineering realities persist regardless of scoring systems. As humanity expands its presence in space for communications, science, and exploration, balancing genuine environmental progress with technological necessity remains a central challenge.
ESG frameworks may evolve, but the fundamental limits of electric launch propulsion are unlikely to change soon.
Tesla just trademarked ‘MEGAPOD’ with the United States Patent and Trademark Office (USPTO), its latest move in what seems to be a hint that the company is incredibly focused on its AI efforts and storage needs as compute increases.
The application carries serial number 99893717 and lists the applicant as Tesla, Inc., located at 1 Tesla Road, Austin, Texas 78725.
The filing remains in ‘live pending’ status, and it is a new application waiting for assignment to an examining attorney. It has not yet been published or registered.
Tesla just trademarked MEGAPOD
Summary:
“Modular data center hardware systems for artificial intelligence computing, comprised of computer servers, computer hardware for artificial intelligence processing, computer networking hardware, electrical power distribution units, and… pic.twitter.com/3l85DsKadl— Robin (@xdNiBoR) June 19, 2026
According to the official goods and services description in the application, Tesla describes ‘MEGAPOD’ as:
“Modular data center hardware systems for artificial intelligence computing, comprised of computer servers, computer hardware for artificial intelligence processing, computer networking hardware, electrical power distribution units, and cooling systems, sold as a unit; self-contained modular computing hardware systems for artificial intelligence workloads; integrated computer hardware platforms for artificial intelligence computing, namely, enclosures containing computer hardware, power distribution hardware, and cooling hardware, sold as a unit; downloadable software for monitoring, managing, optimizing, and regulating modular artificial intelligence computing hardware systems.”
This description specifies complete, self-contained modular units that integrate servers and specialized AI processing hardware with networking components, power distribution, and cooling systems. It also includes associated downloadable software for oversight and optimization of these systems. The language emphasizes hardware sold “as a unit” and enclosures that combine the necessary elements for AI computing workloads.
Tesla has an established history of developing and commercializing modular hardware systems. Its Megapack product line, for example, consists of utility-scale battery energy storage systems designed as containerized units for grid applications. The MEGAPOD filing follows a similar pattern of protecting a name for modular, integrated hardware platforms, this time focused on artificial intelligence computing infrastructure.
This could be an early move, especially as Tesla did not have trademark rights to the word ‘Cybercab,’ the name of its self-driving, ride-hailing-focused vehicle.
Trademark applications of this type allow companies to secure priority rights to a name for defined categories of goods and services. The USPTO examines applications for compliance with legal requirements, including distinctiveness and absence of conflicts with prior marks. If the application proceeds successfully through examination, publication, and any opposition period, it could result in a federal trademark registration providing nationwide protection. This is what Tesla’s obvious intention is with ‘MEGAPOD.’
Public reports and analysis suggest MEGAPOD could represent modular, container-style AI computing pods designed for easy deployment. These would bundle servers, AI accelerators, power systems, and cooling into self-contained units suitable for distributed AI workloads. This approach aligns with Tesla’s announced AI compute strategy.
In March 2026, Elon Musk outlined plans for “Digital Optimus” (also referred to as Macrohard), a joint Tesla-xAI project for AI agents capable of handling complex digital tasks. The plans include running these agents on Tesla’s AI4 hardware in parked vehicles as well as dedicated compute units installed at Supercharger stations, which collectively offer substantial unused electrical capacity.
What is Digital Optimus? The new Tesla and xAI project explained
A modular hardware platform like the one described in the ‘MEGAPOD’ filing would support scalable, rapid deployment of such distributed compute resources. It could complement Tesla’s other AI infrastructure efforts, including the Dojo supercomputer used for training models and the development of AI systems for autonomous driving and robotics, by enabling edge or regional AI inference without reliance on traditional centralized data centers.
SpaceX confirms third massive compute deal at Colossus data center
Elon Musk responds to SpaceX’s ESG rating and says its rockets won’t go electric
