Understanding the $2 Trillion SpaceX Story: How Elon Musk Always Stays Multiple Moves Ahead?
To infinity and beyond!
Author | Cao Siqi
Editor | Jing Yu
On June 12th, on the day SpaceX was officially listed, Elon Musk chose to go to the Starship base in Texas and remotely rang the opening bell of the NASDAQ with hundreds of employees.
On-site, he said in his familiar self-deprecating tone, "If someone had told me back then that I'd be here today, I'd probably think that person was high. Because even I thought the company would go bust at that time."
On this day, SpaceX officially listed on the NASDAQ with an issue price of $135, raising approximately $75 billion. The stock price soared at the opening and once exceeded $176 during the session, briefly pushing the market value above $2 trillion.
From being forced to start a business in 2002 because he couldn't buy a rocket to completing the largest IPO in human business history today, there are numerous counterintuitive and counter-consensus stories in this 24-year journey.
Although the company nominally builds rockets, the rocket business doesn't make money. Its most well-known achievement is rocket recovery, but the two other stories that support its valuation are Starlink and the newly included "space computing power" in the prospectus.
We've compiled 15 of the most representative small stories to help you gain a more comprehensive understanding of SpaceX.
1. The starting point of SpaceX came from a "public relations act" of a new business tycoon
In 2001, Elon Musk, who had just cashed out from PayPal, wanted to fund a public opinion project called "Mars Oasis" out of his own pocket: spend $10 - 20 million to send a mini greenhouse to Mars and take photos of green plants growing on the red soil to persuade Congress to increase the budget for NASA.
However, he was held back by the shipping cost. European rockets were too expensive, and after three trips to Moscow to buy decommissioned intercontinental missiles, he was dismissed as an amateur.
In many subsequent public speeches and media interviews, Musk said that after the setback, he believed that what was holding humanity back from reaching Mars was neither public will nor congressional budget, but the price of rockets themselves. So, "raising funds for NASA" turned into "making rockets affordable on his own."
In 2002, SpaceX was officially founded.
2. The company had been "failing" for the first six years
From 2002 to 2008, the first three launches of the Falcon 1 all failed.
In that era, all the know-how for building rockets was locked within the national space system. SpaceX couldn't buy blueprints, nor could it hire relevant personnel. Musk later joked in his self-narration that he became the company's chief engineer because "all the good people didn't want to come."
What's even more cruel is the physical nature of rockets: they can't be fully debugged on the ground, and the only way to learn is through launch, explosion, and then another try. The "three consecutive failures" were the tuition fees for a company to self-learn aerospace with live ammunition - only this tuition was priced in tens of millions of dollars, and Musk only had enough money for four tries.
3. The successful fourth launch kicked off the era of "commercial spaceflight"
On September 28th, 2008, the fourth launch of the Falcon 1 was successful - the world's first liquid-fueled rocket developed with private funds entered Earth's orbit.
Before this, "spaceflight" was by default a game for the state: the government spent the money, and the system did the work.
Three months later, NASA awarded a $1.6 billion International Space Station cargo contract (CRS) to this company that had just barely survived. This was the official birth of "commercial spaceflight" as an industry.
4. The new way of playing in commercial spaceflight
Traditional space procurement follows the "cost-plus" model: the contractor reports the actual cost, and the government adds a profit - the more you spend, the more you earn, and no one has the incentive to save money.
In the commercial cargo project (COTS/CRS), NASA gave SpaceX a fixed-price contract: a lump sum. Whatever you save is yours, and you bear the cost if it exceeds the budget. This seemingly dull procurement clause is the real institutional starting point for commercial spaceflight. It was the first time that "making rockets affordable" became a profitable business.
SpaceX's persistent focus on cost later was partly due to its nature and partly forced by this contract.
5. Reusable technology: getting the client to willingly pay for an "unreliable technology"
On December 21st, 2015, the first stage of the Falcon 9 rocket successfully landed back on the ground for the first time - exactly 13 years after the company was founded.
Before this, SpaceX's obsession with rocket recovery had gone through a long period of trials and failures: in the first two flights of the Falcon 9 in 2010, they tried to recover the first stage with a parachute - but the rocket disintegrated during re-entry into the atmosphere before the parachute could open. Starting in 2013, they switched to a reverse thrust scheme. In the following two years, they made nearly ten attempts: some crashed hard on the sea surface, some exploded and toppled on the barge deck, and none returned intact.
However, almost none of these tests were self-funded special tests. Instead, they were piggybacked on the client's paid launches - the same rocket both carried out the mission and conducted the experiment. The client's payload was sent into orbit in the first half, and the deal was done. According to industry practice, the first stage of the rocket after delivering the payload was supposed to be discarded into the sea. SpaceX used it to practice landing on the way.
Musk's "calculation" was: if it exploded, it was just junk; if it succeeded, it would rewrite the history of spaceflight. So, in fact, SpaceX used NASA's orders as a scholarship to complete the "reusable" degree for free. Today, the mission success rate of the Falcon 9 is about 99.4%. In 2025, the Falcon 9 series was launched 165 times, and the booster recovery failed only 3 times.
Falcon 9 rocket during a commercial launch | Image source: SpaceX
6. Today's SpaceX: Starlink makes money to support AI
The prospectus shows that in 2025, SpaceX's total revenue was $18.7 billion, with a net loss of $4.9 billion.
However, when broken down by segments, the story is completely different: the connectivity business where Starlink belongs contributed about $4.4 billion in operating profit in a year, making it the only profitable segment of the company; the space business where rockets are involved had a small loss of about $660 million - mainly because about $3 billion was invested in the development of the Starship.
The real big hole was xAI, which was consolidated into the accounts: it had an operating loss of about $6.4 billion in a year, swallowing up all of Starlink's profits and still not being enough.
In other words, if we only look at the "old SpaceX" (rockets + Starlink), it is already a profitable company. It is the AI it bought for the next story that made it "unprofitable" again.
7. Starlink is the "internal client" that Musk pre - arranged for reusable rockets
In January 2015, Musk publicly announced the Starlink project. It is a "broadband network in the sky" composed of thousands of low - orbit small satellites by SpaceX, selling Internet services to ground users - especially in areas such as the sea, wilderness, and remote areas where fiber - optic cables and base stations can't reach.
In December of the same year, the Falcon 9 landed successfully for the first time. That is to say, before the "affordable rocket" was proven, the "client for affordable rockets" had already been internally approved.
This is not a coincidence but two halves of the same arithmetic problem: the global rocket launch market is only worth $5 - 6 billion a year, and there has not been much change in the past decade. So, the affordable launch capacity can't meet the demand in this market; conversely, to build a network of thousands of satellites globally, the accounts simply don't work out without affordable launch capacity.
8. Before the Starship succeeds, its "buyer's market" has already changed once
A similar pre - investment story happened to the next - generation heavy - lift rocket, the Starship.
In 2014, SpaceX laid the foundation for the Starship base in Boca Chica, Texas - in that year, the Falcon 9 hadn't even had a successful recovery. While the previous generation hadn't been realized, the next generation had already started construction.
What's more noteworthy is the change of the client: the initial narrative of the Starship was about "humans" - Mars colonization and space travel, which Musk had talked about for many years; after the concept of space computing power emerged, the top client of the Starship quietly became the "data center."
The logic remains the same: the Falcon 9 has a payload capacity of about 20 tons in low - Earth orbit, and its client is Starlink; the Starship is planned to have a payload capacity of 100 - 150 tons (planned value in low - Earth orbit). Tourists can't consume such a large payload capacity, but the equipment needed for a space data center might.
For each increase in the size of the rocket, Musk has to "create" a larger commercial client for them.
The first launch of Starship V3, carrying a total of 33 Raptor V3 boosters | Image source: SpaceX
9. "Catching the rocket with chopsticks"
On October 13th, 2024, during the fifth test flight of the Starship, two robotic arms on the launch tower caught the slowly descending booster in mid - air, which went viral on the Internet.
Previously, the Falcon 9 proved that rockets can be "recovered" and "flown again" - but each time it came back, it had to be salvaged at sea and sent back to the factory for refurbishment, with a cycle of weeks. In essence, it was still "repair and reuse." What the Starship aims for is something else: like an airplane, land, check, refuel, and take off again.
Landing legs are dead weight and take up payload capacity; landing far away requires transportation. Having the booster return directly to the embrace of the launch tower means that the place where it lands is also the place where it takes off again - the intermediate links are minimized, and the turnover goal changes from "weeks" to "hours."
The so - called "catching the rocket with chopsticks" actually points out the ultimate form of rockets in SpaceX's view: from being recoverable to "operating like a flight."
10. We may not need a "domestic Starlink," but we definitely need "domestic launch capacity"
The "Chinese version of Starlink" is a popular narrative, but there is a fact that is often overlooked: Starlink solves the problem of "out of reach of ground base stations" - in areas such as the sea, wilderness, and sparsely populated places; while China has the world's most comprehensive ground communication network, so the experience of Starlink - like services in China is naturally limited.
The real issue lies elsewhere: satellites have more than just communication uses - remote sensing, navigation, and future space computing power all require sending a large amount of stuff into space cheaply and frequently.
In other words, China may not need to replicate the "product" of Starlink, but it can't bypass the "launch capacity" behind it. For China's commercial space industry, "whether to have a network in the sky" is not the most core question; "whether to have the ability to build the network" is.
11. Breaking the "never IPO" pledge
SpaceX used to be the most determined "never IPO" company in Silicon Valley. Musk's public reason was that the short - termism of the capital market was incompatible with long - term goals like Mars colonization.
The turning point came in the fourth quarter of last year: the user growth and per - user revenue of Starlink had reached a ceiling, and the capital expenditure for the new story of "space computing power" was so large that only the public market could afford it.
The prospectus revealed that in the first quarter of 2026 alone, the capital expenditure of the AI business exceeded the sum of the space and connectivity segments.
So, going public is not an end - of - the - journey celebration but a financing move for the next big bet.
12. Space computing power is a "consensus," but the details are still unknown
Although the concept of space computing power is new, it has quickly become a super - consensus in the technology industry in the past six months, and almost no one publicly opposes it anymore.
However, when digging deeper, there seems to be no common answer to all the technical details:
What does a space data center look like? There is no public product definition. What kind of data does it process and where does the data come from? No one knows either.
Using the classic three elements of the AI industry as an example - the algorithms are advancing rapidly on the ground, but both "data" and "computing power deployment" are still blank in the space context. Is it for pre - training or inference? The requirements for power supply, heat dissipation, and networking are completely different for the two, and the corresponding satellite designs are also completely different. For a direction priced by trillions of dollars, the product form hasn't even been finalized.
Of course, from another perspective, this just means that there are still a lot of empty seats at the table.
13. Silicon Valley has put real money into "space computing power"
The support of major Silicon Valley tech giants for space computing power is not just lip - service.
Musk merged SpaceX and xAI through a corporate restructuring involving trillions of dollars in assets. The prospectus states that the in - orbit data center will be deployed as early as 2028.
Google launched "Project Suncatcher": published technical papers, planned to launch two prototype satellites equipped with self - developed TPUs, and is in talks with SpaceX about a launch contract.
Jeff Bezos' Blue Origin submitted an application for the "Sunrise Project" of 51,600 data center satellites to the U.S. Federal Communications Commission in March 2026.
Former Google CEO Eric Schmidt acquired the rocket company Relativity Space in 2025, with the public goal of sending a data center into orbit.
Starcloud, in which NVIDIA participated in the investment, sent an H100 chip into orbit in November 2025 and completed in - orbit model training.
Buying companies, merging assets, applying for licenses, and launching satellites - the infrastructure competition has begun.
