Update: Shortly after publishing, SpaceX began a much more ambitious series of tests with the Starship launch tower’s two main arms, which are designed to lift and (one day) catch Starships and Super Heavy boosters.
After lifting the arm carriage about 15m (~50 ft), several times higher than January 3rd’s far more conservative kickoff, SpaceX fired up each arm’s main hydraulic actuator and opened them about as wide as they’re able to move. Unsurprisingly, the arms’ first powered lateral movement happened very slowly, obviously telegraphing caution but probably also hinting at the start of a calibration process needed to determine their full range of motion and associate those positions with certain sensor readings or telemetry to ensure they can be safely controlled. As of midnight CST, that testing has continued well into the night.
Regardless of the purpose, substantial powered movement is a major milestone for the tower’s main arms and all but guarantees that more extensive tests and simulations are soon to come.
SpaceX has moved Starbase’s rocket-catching “chopstick” arms for the first time since they were installed on the orbital Starship pad’s ‘launch tower’ two months ago.
After a shockingly brisk three-month period of assembly, the first arm installed in late August 2021 was a lone structure designed to swing in; grab and stabilize Super Heavy with its claw; fuel and power Starship; and quickly detach and swing away from the rocket during launch. A month and a half later, SpaceX begin installing a much larger pair of more complex arms in mid-October. Unlike the Starship quick-disconnect (QD) arm, the pair of arms that followed were almost nothing like anything built as part of another rocket launch complex.
Unlike other ‘arms’ related to other rocket launch facilities, the pair SpaceX began to install on Starbase’s launch tower were colossal, measuring more than 30m (100+ ft) long and 5-10m (15-30 ft) tall. Built out of heavy-duty steel pipe and affixed to an even sturdier pair of claw-like supports that grab onto the launch tower, the combined assembly likely weighs hundreds of tons. Aside from their sheer scale, Starbase’s main tower arms are also attached to a complex system of cables and an industrial-strength ‘drawworks’ commonly used on giant oil rigs and derricks.
They also feature huge actuators that allow the two arms to open and close, revealing a bit of their purpose. While the main reason they likely exist is to provide SpaceX with an all-weather alternative to cranes for lifting, manipulating, and precisely stacking Starships and Super Heavy boosters at the launch pad, the headline – ever since Musk revealed the idea – has always been plans to use those same arms to literally catch rockets out of mid-air.
To do so, they’ll need to be able to actuate and move extremely quickly and precisely up and down the Starship launch tower, matching the velocity and autonomously determining the position of landing Super Heavy boosters (and possibly Starships) to avoid major damage or the loss of entire vehicles. While arguably an unnecessary gamble and an attempt to micro-optimize the concept of operations of a rocket that’s yet to attempt a single orbital-class launch, SpaceX’s CEO is clearly committed to the idea and – whether or not the first iteration works – has fully delivered on the first complete lift-and-catch system.
On January 3rd, 2022, after removing a large amount of scaffolding in the days prior, SpaceX briefly and slightly moved the installed arms for the first time, using the drawworks to lift the entire arm-and-carriage assembly a few meters (~6 ft) up and down the tower. Once a few minor additional steps are taken, the chopsticks could be ready for much more extensive testing, beginning with basic lift, descent, and arm actuation tests to calibrate and then proof the first-of-its-kind mechanism. Later, SpaceX will likely simulate catching rockets in a wide range of scenarios. Somewhere before, during, or after that testing, SpaceX may perform another fit test with Starship S20 and Super Heavy B4 – but this time using the arms to lift and install the stages.
News
SpaceX is following in Tesla’s footsteps in a way nobody expected
In the span of just months in early 2026, SpaceX has transformed itself into one of the world’s most ambitious AI companies. The catalyst: its February acquisition of xAI.
When Elon Musk founded Tesla in 2003, it was a plucky electric car startup betting everything on lithium-ion batteries and a niche luxury Roadster.
Two decades later, Tesla is far more than a car company. Its valuation increasingly hinges on Full Self-Driving software, the Optimus humanoid robot, the Robotaxi program, and the Dojo supercomputer cluster purpose-built for AI training.
Musk has repeatedly described Tesla as an AI and robotics company that happens to sell vehicles. The cars, in this view, are merely the first scalable platform for real-world AI.
Now, SpaceX is tracing an eerily similar path, only faster and in a direction almost no one anticipated. Founded in 2002 to make spaceflight routine and eventually multiplanetary, SpaceX spent its first two decades perfecting reusable rockets, landing Falcon 9 boosters, and building the Starlink megaconstellation.
Elon Musk launches TERAFAB: The $25B Tesla-SpaceXAI chip factory that will rewire the AI industry
It was an engineering and manufacturing powerhouse, not a software play. Yet, in the span of just months in early 2026, SpaceX has transformed itself into one of the world’s most ambitious AI companies. The catalyst: its February acquisition of xAI.
The xAI deal, announced on February 2, was structured as an all-stock transaction that valued the combined entity at roughly $1.25 trillion—SpaceX at $1 trillion and xAI at $250 billion. In a memo to employees, Musk framed the merger as the creation of “the most ambitious, vertically-integrated innovation engine on (and off) Earth.”
The new SpaceX now owns Grok, the large language model family that powers the chatbot of the same name, along with xAI’s massive training infrastructure. More importantly, it has a declared mission to move AI compute off-planet.
Earth-based data centers are hitting hard limits on power, cooling, and land. Musk’s solution is orbital data centers, or constellations of solar-powered satellites that act as supercomputers in the sky.
SpaceX has already asked regulators for permission to launch up to one million such satellites. Starship, the company’s fully reusable heavy-lift vehicle, is the only rocket capable of delivering the necessary mass at the required cadence.
Each orbital node would enjoy near-constant sunlight, vast radiator surfaces for passive cooling, and zero terrestrial real-estate costs. Musk has predicted that within two to three years, space-based AI inference and training could become cheaper than anything possible on the ground.
This is not a side project; it is the strategic centerpiece Musk has envisioned for SpaceX. Starlink already provides the global low-latency backbone; next-generation V3 satellites will carry onboard AI accelerators. Rockets deliver the hardware, while AI optimizes every aspect of launch, landing, and constellation management.
The feedback loop is self-reinforcing, too. Better AI makes better rockets, which launch more AI infrastructure.
Just yesterday, on April 21, SpaceX doubled down.
It secured an option to acquire Cursor—the fast-growing AI coding tool beloved by software engineers—for $60 billion later this year, or pay a $10 billion partnership fee if the full deal does not close.
Cursor’s models already help engineers write code at superhuman speed. Pairing that technology with SpaceX’s Colossus-scale training clusters (the same ones powering Grok) positions the company to dominate AI developer tools, much as Tesla dominates autonomous driving software.
Why SpaceX just made a $60 billion bet on AI coding ahead of historic IPO
The parallels with Tesla are striking. Both companies began in a single, capital-intensive sector: Tesla with EVs, SpaceX with launch vehicles. Both used early hardware success to fund AI at scale. Tesla’s Dojo supercomputers train neural nets on billions of miles of real-world driving data; SpaceX now trains on telemetry from thousands of orbital assets and re-entries.
Tesla’s FSD chip runs inference on cars; SpaceX’s future satellites will run inference in orbit.
Tesla’s Optimus robot will work in factories; SpaceX envisions lunar factories manufacturing more AI satellites, eventually using electromagnetic mass drivers to fling them into deep space.
Critics once dismissed Musk’s multi-company empire as unfocused. The 2026 moves reveal the opposite: deliberate convergence.
SpaceX is no longer merely a rocket company that sells internet from space. It is an AI company whose competitive moat is literal orbital infrastructure and the only vehicle that can service it at scale. The forthcoming IPO, expected later this year, will almost certainly be pitched not as a space play but as the purest bet on AI infrastructure the public market has ever seen.
Whether the orbital data-center vision survives regulatory scrutiny, astronomical concerns about light pollution, or the sheer engineering challenge remains to be seen.
Yet the strategic direction is unmistakable. Just as Tesla proved that software and AI could redefine the century-old automobile, SpaceX is proving that rockets are merely the delivery mechanism for the next great computing platform—one that floats above the clouds, powered by the sun, and limited only by the physics of orbit.
In that unexpected sense, history is repeating. Tesla stopped being “just a car company” years ago. SpaceX has now stopped being “just a rocket company.” Both are becoming something far larger: AI powerhouses with hardware moats so deep that competitors will need their own reusable megaconstellations to keep up.
The age of terrestrial AI is ending. The age of space-based AI is beginning—and SpaceX is building the launchpad.
Elon Musk
Why SpaceX just made a $60 billion bet on AI coding ahead of historic IPO
SpaceX has secured an option to acquire Cursor AI for $60 billion ahead of its historic IPO.
SpaceX announced today it has struck a deal with AI coding startup Cursor, securing the option to acquire the company outright for $60 billion later this year, while committing $10 billion for joint development work in the interim. The announcement described the partnership as building “the world’s best coding and knowledge work AI,” and comes just days after Cursor was separately reported to be raising $2 billion at a valuation above $50 billion.
The move makes strategic sense given where each company currently stands. Cursor currently pays retail prices to Anthropic and OpenAI to the same companies competing directly against it with Claude Code and Codex. That means every dollar of revenue Cursor earns partially funds its own competition. With SpaceX bringing computational infrastructure to the Cursor platform, that could reduce Cursor’s dependence on OpenAI and Anthropic’s Claude AI as its providers. Access to SpaceX’s Colossus supercomputer, with compute equivalent to one million Nvidia H100 chips, gives Cursor the infrastructure to run and train its own models at a scale it could never afford independently. That one change restructures the entire unit economics of the business.
Elon Musk teases crazy outlook for xAI against its competitors
Cursor’s $2 billion in annualized revenue and enterprise reach across more than half of Fortune 500 companies gives SpaceX something its xAI subsidiary currently lacks, which is a proven, fast-growing software business with real enterprise distribution.
For Cursor, SpaceX’s $10 billion in joint development funding is transformational. Cursor raised $3.3 billion across all of 2025 to reach that $2 billion in revenue. A single $10 billion commitment from SpaceX, even as a development payment rather than an acquisition, dwarfs everything Cursor has raised in its entire existence. That capital accelerates product development, enterprise sales infrastructure, and proprietary model training simultaneously.
The timing is deliberate. SpaceX filed confidentially with the SEC on April 1, 2026, targeting a June listing at a $1.75 trillion valuation, in what would be the largest public offering in history. The company is expected to begin its roadshow the week of June 8, with Bank of America, Goldman Sachs, JPMorgan, and Morgan Stanley serving as underwriters. Adding Cursor to the portfolio before that roadshow gives IPO investors a concrete enterprise software revenue story to price in, alongside rockets and satellite internet.
The deal also addresses a weakness that became visible after February’s xAI merger. Several xAI co-founders departed following that acquisition, and SpaceX had already hired two Cursor engineers, signaling where its AI talent strategy was heading. Cursor, for its part, faces a pricing disadvantage competing against Anthropic’s Claude Code.
Whether SpaceX exercises the full acquisition option before its IPO or after remains the open question. Either way, this deal reshapes what investors will be buying into when SpaceX goes public.
Elon Musk
How much of SpaceX will Elon Musk own after IPO will surprise you
SpaceX’s IPO filing confirms Musk will maintain his voting power to make key decisions for the company.
Elon Musk will retain dominant voting control of SpaceX after it goes public, according to the company’s IPO prospectus that was filed with the SEC. The filing reveals a dual-class equity structure giving Class B shareholders 10 votes each, concentrating power with Musk and a handful of other insiders, while Class A shares sold to public investors carry one vote.
Musk holds approximately 42% of SpaceX’s equity and controls roughly 79% of its votes through super-voting shares. He will simultaneously serve as CEO, CTO, and chairman of the nine-member board after the listing. Beyond that, the filing includes provisions that may limit shareholders’ influence over board elections and legal actions, forcing disputes into arbitration and restricting where they can be brought.
The case for Musk holding this level of control is grounded in SpaceX’s actual history. The company’s most important bets, from reusable rockets to a global satellite internet constellation, were decisions that ran against conventional aerospace thinking and would likely have faced resistance from a board accountable to investor gains. Fully reusable rockets were considered economically irrational by established industry players for years. Starlink, which now generates over $4 billion in annual operating profit, was widely dismissed as financially unviable when it was proposed. The argument for concentrated founder control seems straightforward, and the decisions that built SpaceX into what it is today required someone willing to ignore consensus and absorb years of losses.
SpaceX files confidentially for IPO that will rewrite the record books
For context, Musk’s position is significantly more dominant than Zuckerberg’s at Meta. The comparison with Tesla is also worth noting. When Tesla did its IPO in 2010, it did not issue dual-class shares. Musk has only recently pushed for enhanced voting protection, proposing at least 25% control at Tesla in 2024 after selling shares to fund his Twitter acquisition left him with around 13%.
SpaceX has clearly learned from that experience and structured the IPO differently by planning to allocate up to 30% of shares to retail investors, roughly three times the typical norm for a large offering. The roadshow is expected to begin the week of June 8, with a Nasdaq listing rumored to be a $1.75 trillion valuation and a $75 billion raise.
Tesla confirmed HW3 can’t do Unsupervised FSD but there’s more to the story
Tesla isn’t joking about building Optimus at an industrial scale: Here we go
