Wall Street's infatuation with SpaceX has reached an almost frenzied level.

By issuing only 4.2% of its shares, it raised $75 billion—2.6 times the amount raised by Saudi Aramco, previously the largest IPO.

Within three days of going public, its market capitalization surpassed Amazon's, even though SpaceX's annual revenue is only 2.6% of Amazon's.

SpaceX has become the perfect debate topic: proponents fully exercise their imagination regarding its space stories, while opponents are left with question marks about its sky-high valuation.

The Aerospace 'Mixue Ice Cream & Tea'

What story has SpaceX told? We believe it's a tale of technological leadership fueling imagination.

Its business primarily consists of three segments: launches, Starlink, and AI. Through these, SpaceX has drawn a pie worth $28.5 trillion, equivalent to RMB 194 trillion.

How big is this pie? It's larger than China's current GDP.

This figure seems outrageous, but let's examine why.

The first business, launches, features the Falcon and Starship, with rational—even conservative—market estimates. SpaceX projects a $370 billion market for space-enabled solutions.

According to the prospectus, launch revenue will reach $4.1 billion in 2025, up 7.6% year-on-year. While this growth rate isn't outstanding, Q1 saw a decline from $870 million to $620 million, a 28.4% YoY drop.

However, launches are the foundation of SpaceX's ecosystem and storytelling. They offer several key advantages:

Low cost, initially driven by a vertically integrated manufacturing system. SpaceX independently develops and produces rocket engines, applying automotive casting techniques to aerospace engines. Materials have also been significantly optimized, with 30X stainless steel at $2,500/ton replacing carbon fiber at $30,000/ton. Early Falcon 9 models shared 80% of their components with the Heavy Falcon, reducing manufacturing and machining costs.

Reusability. SpaceX wasn't the first to achieve recoverability but was the first private company to successfully launch and recover orbital-class rockets. A decade ago, the Falcon 9 achieved first-stage recovery via 'landing legs,' making Musk a legend.

Recently, the Heavy Starship introduced the 'chopstick rocket capture' technique, continuously validating its reliability. The spacecraft has achieved soft-landing destruction, with full reusability on the horizon. Given time, this technology will likely match the Falcon 9's reliability, further reducing launch costs.

From manufacturing to launch, SpaceX has drastically reduced rocket costs. According to Sinolink Securities, a brand-new Falcon 9 costs about $50 million to produce, with the $30 million first stage being recoverable. This brings the marginal cost of internal launches down to nearly $15 million.

SpaceX's cost estimates show that from 1970 to 2000, the global average cost for launching a kilogram into low-Earth orbit was $18,500. The Falcon 9 reduced this to $2,700/kg, and the Heavy Falcon further lowered it to $1,400/kg—a 92% reduction. Full rocket reusability could cut costs by over 99%.

Not only cheap but also user-friendly. 'Landing legs' significantly shorten launch cycles, with recovered rockets ready for reflight in about 50 days. 'Chopstick capture' could reduce this to just 1 hour—remarkably efficient.

To date, SpaceX has completed over 650 launches, with the Falcon 9 achieving a 99% success rate.

Its carrying capacity is also formidable. The Falcon 9 can lift 23 tons, the Heavy Falcon 64 tons, and the third-generation Starship will soon reach 100 tons, surpassing traditional rocket concepts entirely.

Affordable, reliable, and high-capacity—calling SpaceX the aerospace 'Mixue Ice Cream & Tea' isn't an overstatement.

In the space narrative, Earth resembles an isolated island in the ocean, and SpaceX controls the cheapest transportation off the island. While not everyone will use it, many will. If humanity enters a space exploration era, SpaceX will likely dominate market share and hold 'pricing power' for an extended period.

Centered on its aerospace business, Musk continuously enhances Wall Street's recognition of space narratives.

Exploiting Capitalism

Starlink is Musk's first validated story.

With significantly reduced launch costs and high carrying capacity, SpaceX has taken over numerous launch missions. According to SpaceX, in 2025, it conducted 165 launches, accounting for 85% of U.S. and 51% of global launches; it deployed about 3,150 satellites, representing 85% of U.S. and 70% of global launches. By late 2025, SpaceX had 9,395 satellites in orbit, two-thirds of the global total.

A substantial portion of these supports Starlink.

As of late March, SpaceX had deployed over 9,600 Starlink satellites, serving 10.3 million users.

Starlink is profitable and SpaceX's cash cow. In 2025, it generated $11.387 billion in revenue, up 49.8% YoY, with an operating profit of $4.423 billion, a 120.4% YoY surge.

SpaceX pioneered this business, making space security increasingly relevant.

Let's examine Starlink's user base:

Individual and family users now span 164 countries and regions, particularly in rural, remote, maritime, and aerial areas where ground networks struggle. Monthly fees range from $40-$80—not cheap.

This creates a dilemma: remote areas have limited purchasing power. Starlink offers a solution, but its high pricing may limit its customer base, reducing potential revenue.

Currently, SpaceX provides low-monthly-fee options for remote areas. Starlink's ARPU (Average Revenue Per User) has declined yearly, from $99 in 2023 to $81 in 2025, dropping sharply to $66 in Q1 this year, aiming to capture the long tail.

Enterprise, maritime, and aviation users receive dedicated network services. As corporate or government clients, they face higher fees, up to $500/month, making them the primary profit contributors.

Government and military users rely on SpaceX for encrypted communications and space traffic monitoring services.

SpaceX also envisions direct satellite-to-phone connectivity, requiring operator partnerships already in progress.

SpaceX estimates Starlink's market potential at $16 trillion (RMB 108 trillion), higher than Switzerland's GDP and ranking within the global top 20.

This estimate seems high but not entirely unreasonable.

For context, major operators like AT&T, T-Mobile, and Verizon have combined annual revenues of about $352.148 billion. Considering global operators, this market size is theoretically achievable.

However, market potential doesn't equate to captured share. With revenue in the billions, Starlink faces a long road to unlock a $16 trillion market.

Starlink's true value lies in securing orbital resources. According to media reports, Hu Haiying, chief commander of the Qianfan constellation satellite system, noted that Starlink's in-orbit satellites exceed 60% globally, with a planned total of 42,000, securing nearly 70% of prime low-Earth orbit positions. Space orbits and communication frequencies are non-renewable resources allocated on a first-come, first-served basis—eliminating competitors' options.

In this context, Starlink may establish a formidable moat in the global communications market, ensuring stable revenue and cash flow.

Through Starlink, Musk proved that absolute cost and efficiency advantages in satellite launches enable unique storytelling.

With Starlink established, AI becomes the true dream-weaving focus.

Wall Street's Enchantress

In its prospectus, SpaceX values the AI-related market at $26.46 trillion—equivalent to China's mainland GDP plus Germany's. Thus, most of SpaceX's 'pie' comes from AI.

Specifically: AI enterprise super-apps ($22.7 trillion); AI computing infrastructure ($2.4 trillion); AI consumer subscription markets ($760 billion); and AI digital advertising ($600 billion).

These align with agents, data centers, large models, and social platforms. The latter two relate to xAI and the X platform. For agents, the focus is on the AI programming tool Cursor.

On June 16, SpaceX announced acquiring Cursor's parent company, Anysphere, via an all-stock deal, valuing Cursor at $60 billion post-acquisition.

Cursor, a popular AI product in recent years, leads the Vibe Coding trend. Its core experience relies on Anthropic and OpenAI models. Under SpaceX, Cursor gains computing and large model support, enabling direct training of code large models.

However, Cursor, xAI, and social platforms remain Earth-bound, difficult to link with SpaceX's space narrative.

The true SpaceX connection lies in data centers.

Unlike OpenAI and Anthropic, which focus solely on large models, or NVIDIA, which merely 'sells water,' xAI does both.

Musk proved Tesla could sell 'water,' with Anthropic procuring $1.25 billion/month in SpaceX/xAI computing power. If contracts continue until May 2029, the total could exceed $40 billion.

Computing power is AI's most scarce resource—where there's supply, demand follows. The challenge lies in building that supply.

Currently, the U.S. lacks no chips but struggles with infrastructure like power and cooling water due to environmental concerns. Musk proposes space-based computing power.

In March, Musk launched the ambitious Terafab project. Jointly developed by Tesla, SpaceX, and xAI, it aims to build the 'largest chip factory ever,' producing 1 terawatt (TW) of AI computing chips annually—50 times global capacity—primarily for space deployment.

To achieve this, SpaceX plans to deploy up to 1 million satellites for orbital AI data centers.

Musk's visuals show these centers comprising massive satellites with solar arrays over 170 meters long, each providing 100 kW for onboard AI processors.

This concept ignited the space computing trend. According to Musk, deploying computing power in space offers advantages: 24/7 solar energy with higher radiation intensity, no clouds or rain, and no weather impact.

Additionally, it eliminates cooling water needs, using vacuum environments for radiative cooling—though this remains controversial. Regardless, the idea has gained traction, making space computing vital for valuation and market prospects.

In summary, low-cost, reliable satellite launches have built SpaceX's moat. Starlink proved that space narratives can generate revenue effortlessly. Leveraging AI's hype, SpaceX has inflated its valuation.

But does this 'pie' truly match reality?

From $18.7 Billion to $1 Trillion

On social media, Musk told a financial commentator, 'I'd be surprised if revenue doesn't exceed $1 trillion by 2030.'""Assuming Musk's statement holds, at 2025's revenue level, SpaceX needs a 121.7% CAGR over five years—far above its 34.05% CAGR the past two years. Q1 revenue grew only 15.42%.""No one denies Musk's or SpaceX's greatness, but objectively, much remains speculative. Musk isn't infallible—SolarCity, Hyperloop, etc.—but his successes with Tesla, Starship, Falcon, and Starlink overshadow these failures.

Dissect SpaceX's current main businesses.

There is indeed an intergenerational advantage in the launch business. However, this does not mean there is no possibility of catching up.

The main competitors, in addition to Blue Origin, include Chinese companies. The former is already the world's second company to master orbital-class rocket recovery technology.

The maiden flight of China's heavy-lift launch vehicle, the Long March 9, is planned for 2028, with a vehicle body that surpasses the Starship V3. In the commercial space sector, models like the Zhuque-2 and Zhuque-3 are also catching up. Among them, the Zhuque-3 rocket attempts to integrate the architecture of the Falcon 9 with Starship technology (stainless steel vehicle body + liquid oxygen methane), although its maiden flight and recovery failed last December.

Musk assessed that it would take 'more than five years to achieve Falcon 9 reliability.' However, in five years, SpaceX's lead will be within sight.

In terms of launches, SpaceX is known for its high launch efficiency and low launch costs, but it is not the only launch provider. Chinese aerospace is already renowned for its low costs, stability, and reliability. As long as there is market demand to catalyze growth, even if Chinese aerospace cannot compress costs as drastically as SpaceX in the short term, it can still build its own launch capabilities.

Take Starlink as an example. Although Musk's Starlink is being constructed very rapidly, the deployment of the 'Chinese Starlink' will not be slow either. The reason it is slower than Starlink is that China's base station infrastructure is very well-developed, and the demand for satellite networks is not as urgent.

After Starlink introduced the 'Starlink Direct to Cell' feature, the traditional operator ecosystem is facing changes, and the business prospects are becoming clearer. The 'Chinese Starlink' Qianfan constellation has already begun intensive launches, with a Phase 1 target of 1,296 satellites to be completed by 2027; Phase 2 will add approximately 10,000 satellites, with over 10,000 satellites networked by 2030; and the final phase will exceed 15,000 satellites, integrating into the 6G ecosystem.

Therefore, if the market space is validated, commercial competition at the Starlink level may have just begun. Starlink aims to seize market share from traditional operators and will also face competition from across the ocean. This market will not allow SpaceX to easily monopolize it.

AI represents an even more long-term competition. Regarding large models, xAI is hardly ranked among the top in the United States, lagging behind competitors such as Anthropic, OpenAI, and Google.

In the Terafab project, whether the chip manufacturing factory can be competitive remains highly uncertain. This may be the most challenging business on the planet. It requires not only capital but also deep process technology accumulated through years of large-scale production. Even if the factory is successfully built, controlling yield rates will be a significant challenge.

Morgan Stanley believes that such a project could cost over $20 billion and take several years to complete. Integrating logic, memory, and advanced packaging technologies runs counter to the decades-long trend of industry specialization.

In addition, Terafab faces issues of supplier concentration and talent bottlenecks. Advanced extreme ultraviolet lithography systems rely on a few suppliers, and there is a gap in semiconductor engineering talent reserves, fab construction experience, and supply chain maturity compared to TSMC and Samsung. Therefore, for Terafab, SpaceX has to collaborate with Samsung, Intel, and others.

As for the space data center, on the one hand, this project still needs to be validated, and on the other hand, SpaceX is not the only one optimistic about it.

Space computing power has also been recognized by Amazon's Bezos. Bezos predicts that gigawatt-scale facilities will be launched into space within 10 to 20 years. He firmly believes that space data centers will 'defeat ground-based facilities in terms of cost.' Considering that Amazon Web Services holds over 30% of the global cloud computing market share, Amazon could deeply integrate space data centers with AI cloud services in the future, building an orbital version of AWS, which would be more competitive and have clearer business prospects than SpaceX.

SpaceX may be the most unique entity on the planet to date. It is almost impossible to find any comparable company or business model. With rocket launches as its foundation, it provides a narrative of space exploration, aiming for the stars and the sea. Its image of leading humanity into a new era of space exploration gives SpaceX a certain inherent 'nobility.'

However, returning to rationality, the scenarios Musk envisions are incredibly grand but must ultimately come down to the 'mundane' aspects of daily life. The reality for SpaceX is that its revenue is only in the tens of billions. From launches to Starlink to AI, it faces numerous competitors and consumes vast amounts of capital. The current fervent AI boom is the best opportunity for SpaceX to replenish its resources.