Key Points:

1. SpaceX's IPO is not a traditional listing of a space company, but rather a super infrastructure pricing redefined by AI. The company raised $75 billion in funds at a valuation of $1.75 trillion, with revenue in 2025 only at $18.7 billion, corresponding to a Price-To-Sales Ratio of approximately 93 times. What truly supports the valuation is the new narrative composed of Starlink, AI computing power, xAI, and future orbital data centers.
2. Starlink is currently SpaceX's most robust cash flow engine, with revenues of $11.4 billion and an operating profit margin of 39% in 2025, relying on low-cost launches to form a global satellite internet barrier. Although the launch business incurs losses due to Starship development, it is the company's underlying competitive edge; if Starship achieves high-frequency reuse, the cost of space transportation may continue to collapse, opening up new markets such as satellites, orbital computing, and space manufacturing.
3. AI is the core of the valuation flexibility in this IPO. Anthropic's annual $15 billion computing power contract validates the commercial value of the Colossus Data Center; however, the AI business still incurs high investment and high losses, and the Orbital Data Center is more like a long-term option.
4. The RockFlow investment research team believes that SpaceX is an absolutely scarce asset, but $1.75 trillion has already pre-empted a large amount of future value. What investors are buying is not just the company's cash flow, but also Musk's narrative and execution capabilities.

SpaceX's IPO is by no means just a "space company going public".

According to the latest disclosed S-1 filing, Elon Musk is repackaging SpaceX as a super infrastructure company spanning space launch, satellite internet, AI computing power, AI applications, and future orbital Data centers

This IPO was completed as planned in mid-June, with a target valuation of $1.75 trillion and an intended fundraising of $75 billion . This will become the largest IPO in history and directly catapult SpaceX into the ranks of the world's highest-valued companies.

But the question also arises: How can a company with revenue of approximately $18.7 billion in 2025 support a valuation close to 93 times the Price-To-Sales Ratio?

The answer lies not in traditional space business, but in four key words: Starlink, AI computing power, Anthropic contract, and Orbital Data Center . In this article, the RockFlow investment research team will take you to explore the value and future of SpaceX from multiple dimensions.

SpaceX IPO: Space IPO or AI IPO?

SpaceX was founded in 2002 with the initial mission of reducing the cost of space transportation and ultimately making humanity a multi-planetary species. Over the past two decades, it has almost single-handedly reshaped the global commercial space industry.

But this time, the most notable aspect of the S-1 filing is SpaceX's complete reconstruction of its own narrative: it is shifting from a "space company" to an "AI infrastructure company".

In the S-1 filing, some keywords appear with extremely high frequency: Launched, 835 times, which means that the traditional core business remains the foundation, Starlink, 381 times, which means satellite internet remains a core asset, Starship, 222 times, is the key to the next-generation cost curve;

But at the same time, Calculation, 507 times, which means AI computing power has become the new mainstay, xAI, 400 times, which means the AI business has been deeply integrated, Grok, 243 times, this is the entry point for the AI application layer, Data center, 154 times, this is the focus of AI infrastructure

The meaning of these data is: SpaceX wants the open market to value it as an AI company rather than a traditional aerospace company.

This is not just empty talk. Because SpaceX has indeed generated actual revenue, actual customers, and actual capital expenditures in AI computing infrastructure.

Three major business segments: Starlink makes money, launch provides a foundation, and AI burns money but tells the future

According to the S-1 filing, SpaceX has now become a comprehensive technology infrastructure platform composed of three major segments:

1. Space: Launch and Starship
2. Starlink: Satellite Internet and Mobile Connectivity
3. AI: xAI, Grok, X, Colossus Data center and Future Orbital Computing

In 2025, SpaceX's consolidated revenue was approximately $18.7 billion , and adjusted EBITDA was approximately $6.6 billion .

The economic models of different businesses vary greatly, among which: As a cash flow engine, Starlink will generate $11.4 billion in revenue and incur $4.2 billion in capital expenditure in 2025; Launch business serves as a technological and cost Competitive Edge, with revenue in 2025 reaching $4.1 billion, and capital expenditure of $3.8 billion; AI business is more like a high-risk option, with $3.2 billion in revenue and $12.7 billion in capital expenditure in 2025

In other words: Starlink is the fundamental base today, AI is the valuation flexibility of tomorrow, and launch services are the underlying engineering capabilities that support both.

Starlink contributed approximately 61% of SpaceX's revenue in 2025, and is also the company's clearest source of profit.

By the end of March 2026, Starlink had covered 164 countries, regions, and markets, with approximately 9,600 low Earth orbit satellites, accounting for approximately 75% of the active and maneuverable satellites in low Earth orbit.

More importantly, Starlink's operating leverage is being unleashed:

• Revenue in 2025 will increase by approximately 50% year-on-year
• Operating profit increased by approximately 120% year-on-year
• Adjusted EBITDA increased by approximately 86% year-on-year

The business model of satellite networks has a very important characteristic: extremely high upfront capital expenditure, but after the constellation is built, the marginal cost of new users is relatively low. In other words, as long as satellite capacity and ground terminal supply can keep up, Starlink's profit margin has room to continue rising.

SpaceX's launch business revenue in 2025 is approximately $4.1 billion, but the operating profit is negative, mainly because Starship R&D is still in a high-investment stage.

However, in terms of industry status, SpaceX's launch business is almost at a monopoly level:

• Cumulative completion of more than 650 launches
• By 2025, more than 80% of the global mass will be sent into orbit
• The Falcon 9 booster has achieved a maximum of 34 reflights
• The success rate of orbital launch missions exceeds 99%

What truly revolutionized the industry for SpaceX is turning "launch" from a single-use consumable into a reusable engineering system.

Historically, NASA's cost to send 1 kilogram of objects into space was approximately $18,500 ; Falcon 9 reduced the cost to approximately $2,700 per kilogram ; Falcon Heavy further reduced it to approximately $1,400 per kilogram .

If Starship achieves high-frequency and stable launches, SpaceX aims to reduce the cost to approximately $200 per kilogram .

When transportation costs drop by 99%, previously uneconomical businesses suddenly become viable, such as large-scale low Earth orbit satellite constellations, orbital data centers, space manufacturing, space tourism, lunar and Martian logistics, asteroid mining, and ground-to-ground ultra-high-speed transportation.

Starlink itself is the first large-scale commercial outcome of this cost reduction.

AI is the main axis of the valuation story: Anthropic's contract validates the value of computing power assets

What has attracted the most attention from Capital Markets in SpaceX's IPO this time is clearly AI computing power infrastructure.

Starting from May 2026, Anthropic agrees to pay $1.25 billion per month to SpaceX for using the computing power of its two Colossus AI Data centers located in Memphis. This is equivalent to an annualized revenue of $15 billion .

How exaggerated is this figure? SpaceX's total annual revenue in 2025 was $18.7 billion . That is to say, the annualized scale of just one AI computing power contract from Anthropic is close to 80% of SpaceX's total annual revenue in 2025.

The significance of this contract does not lie in Anthropic, but in the fact that it proves: SpaceX's AI infrastructure can be leased as an independent commercial asset without having to rely entirely on Grok's model capabilities.

In other words, SpaceX not only wants to compete with OpenAI, Anthropic, and Google Gemini in models, but also aims to be the "computing power landlord" behind these models.

This is an infrastructure business that is closer to cloud computing, energy, telecommunications, and data centers.

According to PitchBook's analysis, the construction cost of SpaceX's first two Colossus II clusters is approximately $2.7 million per megawatt, about four times the industry benchmark.

On the surface, this appears to be a high cost. However, when combined with Anthropic's monthly $1.25 billion computing power contract, the economics reverse: PitchBook's calculations show that, based on the disclosed costs, the payoff period could be less than 1 month; even if the cost doubles, the payoff period may only be about 2.2 months.

This explains why AI computing power infrastructure is given extremely high valuations in the current market. Against the backdrop of global shortages of GPU, data centers, power, and cooling resources, companies that can quickly bring computing power online are themselves scarce assets.

SpaceX's AI Cloud could be the "Musk version of AWS". The market has already given it a semi-joking yet accurate name:Elon Web Services.

There is real-world logic behind this statement. The core of AWS is the monetization of Amazon's internal infrastructure capabilities; the logic of SpaceX AI Cloud, on the other hand, is to open up the training and inference infrastructure required by xAI/Grok to other model companies and enterprise customers.

Potential advantages of SpaceX include:

• Extremely fast Data center construction speed
• Deep supply chain relationship with NVIDIA
• Basic load brought by xAI's own model requirements
• X platform provides distribution entry points for 550 million monthly active users
• Starlink provides global low-latency connectivity
• Future Orbit Data Center provides long-term narrative space

Of course, the problems are equally obvious:

• Anthropic is a direct competitor of Grok
• The contract allows either party to terminate with 90 days' notice in advance
• AI computing power rental prices may decline as supply expands
• Data center depreciation, energy, network, and chip iteration risks are very high
• AI business is still burning a significant amount of money at present

In the first quarter of 2026, SpaceX's revenue was $4.7 billion , but due to AI-related capital expenditures as high as $7.7 billion , the company incurred a loss of approximately $2.5 billion .

1.75 Trillion Valuation: Buying the Company or Elon Musk Himself?

SpaceX plans to be valued at $1.75 trillion financing 75 billion USD . If calculated based on 18.70 billion USD revenue in 2025, the Price-To-Sales Ratio is about 93.6 times.

This is significantly higher than most current technology giants. From the perspective of the traditional Value Investment framework, this valuation is hard to be comfortable with.

However, SpaceX's pricing logic is a superposition of three levels:

1. Value of Starlink's Global Connectivity Infrastructure
2. Option Value of AI Computing Power Infrastructure
3. Musk Premium, also known as the "Elon Market Hypothesis"

The so - called "Elon Market Hypothesis" can be roughly understood as: in certain capital market cycles, the value of an asset depends not only on cash flow but also on its distance from Elon Musk.

This sentence may sound like a joke, but it has indeed recurred in the financing histories of Tesla, xAI, and SpaceX.

Martin Peers of The Information once proposed a relatively traditional segment valuation framework, suggesting that the value of SpaceX's operating business is approximately $678 billion:

Among them, Launch business valued at $300 billion, referencing multiples of high-growth launch companies such as Rocket Lab SpaceX AI is valued at $225 billion, calculated based on AI cloud revenue + the acquisition value of Cursor Starlink is valued at $150 billion, with the core assumption being that its revenue in 2026 will be approximately $15 billion, 10 times its sales X is valued at $3 billion, estimated based on advertising revenue and Snap's multiple

Compared with $1.75 trillion, this framework has a gap of approximately $1 trillion .

What is this gap? It can be understood as:

• The possibility of Starlink being undervalued
• Option value of Starship monetization
• Long-term growth premium of AI computing power
• Long-term Story of Orbital Data Center
• Ultra-long-term options for Mars colonization and space economy
• Premium on Musk's personal control and execution capabilities

Valuation guru Aswath Damodaran's preliminary valuation of SpaceX is approximately $1.22 trillion. His core assumptions include:

• The launch market continues to expand, with SpaceX maintaining its dominant position
• Starlink Becomes a Global Satellite Internet Giant
• xAI gains market share in consumer and some enterprise scenarios
• Other extended business provides forward option value
• The weighted average cost of capital is approximately 8%

Although this valuation is already very optimistic, it is still about one-third lower than the IPO target valuation.

This means that if an investor buys at $1.75 trillion , they need to believe that at least one of the following holds true:

• Starlink users have long exceeded current expectations
• AI cloud revenue quickly exceeded tens of billions of dollars
• The real monetization of the Orbital Data Center
• Starship significantly reduces launch costs and creates new markets
• SpaceX can enjoy the premium of globally scarce infrastructure over the long term
• The open market is willing to give Musk extremely high valuation multiples over the long term

1.75 trillion dollars is not buying SpaceX's present; it presupposes that people are willing to buy a very vast future.

Why is Starlink still likely to be undervalued?

Despite AI narratives stealing most of the spotlight, Starlink remains SpaceX's most critical asset in terms of financial quality.

Starlink already has more than 10 million users and provides services in 164 markets worldwide.

Its advantage lies in a set of extremely complete system-level capabilities:

• Own rocket launch capabilities
• Mass production capacity for satellites
• Low Earth Orbit (LEO) satellite constellation operation capabilities
• Mass production capacity of ground terminals
• Global network scheduling capabilities
• Diverse customer base including government, military, aviation, shipping, and consumers

This vertical integration makes it extremely difficult for competitors to replicate.

Amazon Kuiper, Globalstar, traditional telecommunications companies, and regional satellite companies may all participate in the competition, but it is difficult for them to simultaneously possess SpaceX's low-cost launch capabilities and satellite scale.

The next phase of Starlink's growth, in addition to home broadband, includes Direct-to-Cell , which is the direct connection of satellites to ordinary mobile phones.

If D2C is successful, Starlink's market space will expand from home broadband to:

• Smartphone blind spot coverage
• Automobile connection
• Aviation Internet
• Shipping Communication
• Agricultural Equipment
• Logistics Tracking
• Military Communications
• Emergency Communication
• IoT devices

This will propel Starlink from a "satellite broadband company" to a "global mobile connectivity infrastructure company".

PitchBook even predicts that the number of Starlink users could exceed 1 billion within the next 15 years.

This prediction is extremely aggressive, but it explains why some investors believe that Starlink and launch services alone could support SpaceX's high valuation.

However, Starlink is not without risks:

• Growth in developed markets may gradually slow down
• ARPU in low-income regions may decline
• D2C technology experience still needs to be verified
• Intense competition for spectrum resources
• Complex regulatory and national security reviews in various countries
• Satellite updates and maintenance require continuous capital expenditure
• Starship delays could limit the deployment of next-generation satellites

Starlink's story is compelling, but it is not a risk-free perpetual motion machine.

Orbit Data Center: The most science-fiction and perhaps the most critical long-term option

The most imaginative part of SpaceX S-1 is the Orbital AI Computing Satellite .

According to the document, SpaceX plans to start deploying orbital AI computing satellites as early as 2028. These satellites operate in a sun-synchronous orbit, leveraging the more stable and higher-intensity solar energy in space to process high-energy-consuming AI workloads and transmit real-time intelligence back to Earth via the Starlink network.

It may sound like science fiction, but the logic behind it is not absurd.

AI Data centers are facing an increasing number of real-world bottlenecks:

• Insufficient power supply
• Difficulties in land approval
• Cooling resources are tight
• Residents oppose
• Environmental protection pressure
• Grid connection queuing
• The chip is in place, but the data center is not yet completed

According to relevant information, multiple states in the US are already discussing restrictions or bans on the construction of new large-scale onshore Data centers.

Against this backdrop, the Space Data Center has at least several attractive points, at least in theory:

• can utilize more stable solar energy
• Reduce pressure on land and water resources
• Naturally synergistic with the Starlink network
• can be sent into orbit at low cost via Starship
• Suitable for some non-real-time or batch processing AI workloads

However, investors cannot overlook engineering challenges:

• Impact of Space Radiation on Chips
• Heat dissipation is extremely difficult
• High maintenance and upgrade costs
• Calculation of device lifespan is uncertain
• Data transmission bandwidth and latency limitations
• Launch failure risk
• Orbital Debris Risk
• Unit Economics has not been verified

The Orbital Data Center is currently more like a long - term option than a mature business that can be immediately discounted.

But this type of option is precisely the part that SpaceX is best at telling and easiest for the market to price.

Buying SpaceX is essentially buying an individual option on Elon Musk

SpaceX's corporate governance structure is very clear:Elon Musk has absolute control.He holds approximately 42% equity stake, but controls approximately 85% of the voting rights through Class B super-voting stock.

This means that after going public, SpaceX is not a public company in the ordinary sense. Ordinary shareowners have economic interests, but it is difficult for them to influence the company's strategy.

Investors need to accept:

• Musk can lead major capital allocation
• The company may frequently adjust its narrative focus
• There may be complex transactions among ecosystems such as SpaceX, xAI, X, and Tesla
• High-risk projects may consume cash over the long term
• Corporate governance discount and founder premium can coexist simultaneously

This is exactly where SpaceX differs from ordinary blue-chip companies.

Investing in SpaceX means buying into a super engineering platform driven by a super founder. This may bring astonishing returns, but it may also bring significant volatility.

Musk's long-term incentive plan includes incentives of up to 1 billion restricted stock units of SpaceX , with triggering paths including:

• Market capitalization reached a series of milestones, peaking at $7.5 trillion
• Establish a permanent human colony on Mars with at least 1 million residents

This once again shows that SpaceX's corporate goals differ from those of traditional listed companies. It is built around Musk's long-term mission: to reduce the cost of accessing space, expand the boundaries of human civilization, and embed AI infrastructure within it.

It is very grand and also very "unstable".

IPO Timing: SpaceX is Seizing the AI Capital Window

The timing of this IPO is crucial.

In May 2026, the global AI capital markets remained in a state of high excitement:

• NVIDIA's revenue in the latest quarter reached $82 billion
• Year-on-year growth of approximately 85%
• Gross profit margin is approximately 75%
• Operating profit is approximately $54 billion
• Large technology companies' AI capital expenditure continues to rise
• AI hardware companies such as Cerebras still enjoy high multiples after going public
• OpenAI and Anthropic have also been reported to have IPO plans

The direction that the market is currently most willing to pay a high valuation for remains AI. And SpaceX's submission of Form S-1 during this window period and its transformation into an AI-driven entity is a highly efficient capital market move.

According to Thomas Tunguz's estimate, if SpaceX, Anthropic, and OpenAI all go public at the standard proportion of outstanding shares, the three companies may need to raise up to hundreds of billions of dollars from the open market.

This means: Whoever goes public first may be more likely to siphon off liquidity. SpaceX's push for an IPO ahead of OpenAI and Anthropic has obvious strategic significance.

In addition, Nasdaq has relaxed the rules for including new stocks in the index to attract SpaceX to go public. Typically, newly listed companies need to wait 90 days before being eligible for inclusion in the index, but the "fast-track inclusion" mechanism allows popular IPOs to enter the index 15 days after listing.

This will generate demand for passive funds to buy. But it also means that while index funds buy SpaceX, they may be forced to sell other constituent stocks.

SpaceX's IPO is not just a single company event, but may also become a market liquidity redistribution event.

How should SpaceX set its prices?

The RockFlow Investment Research Team believes that SpaceX cannot be simply valued using a single P/S or P/E ratio. A more reasonable approach is to break it down into four layers of assets:

Among them, Starlink is a current cash flow asset, with the core of its valuation being user growth, ARPU, profit margin, and FCF, and it has medium risk; Launch and Starship are engineering competitive edge assets, with the core of valuation being cost curve, market share, and the opening of new scenarios, presenting medium to high risks; Colossus and AI Cloud are AI infrastructure assets, with the core of valuation being computing power leasing, GPU utilization, and customer contracts, and they carry high risks; Orbital Data Center, Mars, and Space Manufacturing are ultra-long-term option assets, with the core of valuation being future market space and probability of technological breakthrough, and extremely high risk.

Under the optimistic scenario:

• The number of Starlink users continues to grow at a high speed
• D2C Becomes Global Communication Standard
• Starship Achieves High-Frequency, Low-Cost Launches
• AI cloud revenue has exceeded tens of billions of dollars
• Anthropic contracts bring in more large customers
• The Orbital Data Center will be verified as feasible around 2030

In this case, a valuation of $1.75 trillion may not be unacceptable, and could even be just the starting point for a higher market capitalization in the future.

Under the neutral scenario:

• Starlink Continues to Grow, but ARPU Declines
• Launch business remains leading, but Starship monetization is slower than expected
• AI Cloud has revenue, but gross margin and price are under pressure
• Grok struggles to catch up with OpenAI, Anthropic, and Gemini
• The Orbital Data Center remains in the long-term concept stage
• Market cools on high multiples for AI

In this case, SpaceX is closer to the reasonable value range of $1.0 trillion - $1.3 trillion .

Under the pessimistic scenario:

• Excess supply of AI computing power has led to a decline in rental prices
• Anthropic contract termination or scale reduction
• Starlink Growth Slows
• Starship progress delayed
• Capital expenditure continues to devour cash flow
• Corporate governance issues lead to valuation discounts
• AI Capital Markets Cycle Reversal

Under this circumstance, the $1.75 trillion valuation may face significant risk of retracement.

Conclusion: SpaceX is a scarce asset, but its IPO price has already pre-empted a significant amount of future value

RockFlow's investment research team believes that SpaceX is definitely one of the most unique technology companies globally.

It has:

• Globally leading reusable rocket system
• Near-monopoly commercial launch capabilities
• The world's largest low-orbit satellite internet network
• The high-growth and profitable Starlink business
• Engineering capabilities for rapid construction of AI Data centers
• User, capital, and narrative resources of the Musk ecosystem
• Long-term options such as orbital calculations, Martian logistics, and space manufacturing

These asset portfolios have few comparable companies in the global open market.

But scarcity does not mean that any price is reasonable.

The expectations implied by the $1.75 trillion valuation are extremely high. It not only requires Starlink to become a global communication giant, but also demands rapid monetization of the AI computing power business, and expects the market to continue paying a high premium for the super platform controlled by Musk.

From a fundamental perspective, Starlink is the most solid asset; From the perspective of valuation elasticity, AI computing power is the largest variable; From a long-term perspective of imagination, orbital data centers and starships have broken through the ceiling; From a risk perspective, corporate governance, capital expenditure, and AI cycle fluctuations are the three major uncertainties.

Our judgment on SpaceX's IPO can be summarized in three sentences:

1. SpaceX is a physical infrastructure platform in the AI era.
2. Starlink provides the cash flow base, AI computing power provides valuation flexibility, and Starship provides long-term options.
3. A valuation of $1.75 trillion is not unimaginable, but it is already highly dependent on future realization, leaving limited safety margins for ordinary investors.

For long-term investors, SpaceX may be one of the most important technology assets in the next decade.

But for IPO buyers, the key question is: At a valuation of $1.75 trillion, how much has the market already paid in advance for this greatness?

The RockFlow investment research team believes that SpaceX's IPO will be one of the most important asset pricing events in the global Capital Markets in 2026. It will not only test the open market's risk appetite for AI infrastructure but also whether investors are willing to buy into Musk's long-term narrative again.

This time, the rocket launch is not just about satellites, but also the entire AI bull market's imagination for the future.