📊 SpaceX: From Space Launches to a Trillion-Dollar AI Market

An analysis of SpaceX’s business strategy offers a clear case study of how a technological platform can integrate satellite communications, media assets, and generative AI. For media executives, streaming platforms, and IT conglomerates, this analysis strips away the noise to reveal the true economics of AI infrastructure and outlines the blueprint for the global media market in the coming years.

Context: How It All Began

The name SpaceX is familiar to everyone, largely due to the figure of Elon Musk. The aerospace giant has been an industry fixture for a quarter of a century.

Curiously, this business might never have existed had Russian entities been more cooperative in the early days. At the start of his journey, the American entrepreneur personally flew to Moscow in an attempt to purchase ballistic missiles for his ambitious program to send humans to Mars. After protracted negotiations, the deal fell through; the buyer expected to acquire two units for $8 million, while the hosting party demanded $8 million for each. On his flight back from Russia, the startup’s founder decided to build his own launch vehicles.

The project managed to secure funding from the US defense budget to create its first prototypes. Following failures in its first three launches, the engineers had virtually run out of capital, but the fourth attempt proved successful, saving the enterprise from bankruptcy.

In the 2020s, the Starlink satellite constellation was added to the rocket-manufacturing business. This initiative grew out of the concept of the Iridium system—the first network for orbital internet distribution. Utilizing his heavy-lift rockets, Musk has deployed over 10,000 spacecraft into orbit, establishing a high-speed communications network worldwide.

Today, corporate leadership has decided to capitalize on the massive hype surrounding neural networks, adding an AI component to its technological portfolio. Below, we break down its structure in detail.

Key Figures:

• 650 space launches executed by the company throughout its history.
• 10.3 million subscribers make up the satellite internet customer base across 164 countries.
• 1,000 MW represents the nameplate power capacity of the conglomerate’s proprietary data centers.

Core Business Model Concepts

SpaceX’s operations are divided into three primary segments: Space, Connectivity, and AI. Total corporate revenue has reached $18,7 billion, with the network division making the largest contribution at $11.4 billion.

Contrary to widespread media narratives, the organization is not deeply unprofitable. Its operating cash flow is growing steadily, reaching $7 billion on revenue of $19 billion. The net loss reflected in its financial statements is driven purely by non-cash items: approximately $7 billion goes toward depreciation and amortization (D&A), and another $2 billion is allocated to employee stock option programs.

Capital expenditures (CAPEX) stood at $20 billion, slightly exceeding total annual revenue. Crucially, $12.7 billion of this total was directed toward AI development. This year, sector-specific investments are expected to rise to $15 billion, positioning the corporation on par with market leaders such as OpenAI and Anthropic.

The AI Business Segment Components:

• Social Network X (formerly Twitter): Formally, it has little connection to neural networks, yet it accounts for roughly 95% of all AI segment revenues ($3.1 billion out of $3.2 billion).
• Grok: A proprietary Large Language Model (LLM) that ranks among the world’s top 10 leaders in generation quality.
• Colossus: A massive computing infrastructure (data center) powered by 220,000 Nvidia GPUs.

Primary Commercial Drivers:

• The Anthropic Contract: SpaceX generates approximately $15 billion in annual revenue by leasing its computing capacity to the AI startup. The agreement runs through 2029, but includes a critical caveat: the client retains the right to terminate the contract without penalty by giving 3 months’ notice. Optimists believe the partnership will survive the full term amid the ongoing chip shortage, while pessimists expect Anthropic to exit as soon as it finishes building its own facilities.
• The Google Contract: A similar agreement has been struck with the search giant, which has committed to paying $11 billion over the next three years to lease 110,000 GPUs. Notably, Google owns a 6% stake in SpaceX, making it advantageous for the company to demonstrate its partner’s high profitability. This contract also includes a 3-month cancellation option.

Following its public listing on June 12, 2026, the conglomerate’s valuation reached approximately $2.5 trillion, placing it 6th among the world’s ten most valuable corporations, while Elon Musk himself became the first trillionaire in history.

Below is a detailed analysis of the segments detailed in the SpaceX prospectus.

Segment 1. Revenue Structure and Margins

Revenue dynamics across three key business lines: Space, Connectivity, and AI

The business segments are developing at asymmetric rates:

• Space (Rockets): A stable but relatively slow-growing sector.
• Connectivity (Telecom): The engine of the entire corporate group, demonstrating explosive growth and operating margins hovering around 40%.
• AI: A diversified segment whose revenue structure is heavily obscured in the documentation through the use of disparate metrics.

Analysis of operating profitability and cost structures by business segment

Despite press reports concerning financial difficulties, a detailed audit validates the viability of the model: the rocket division yields positive cash flow once D&A and stock options are added back. The telecommunications block demonstrates an excellent profit margin of 40%. Meanwhile, the AI segment’s negative margin (minus 200%) represents a completely standard trajectory for an investment phase in the tech industry.

Within the structure of total CAPEX, the share of neural network technologies is rising rapidly: while they accounted for 61% ($12.7 billion) last year, that figure reached 76% ($7.7 billion out of $10.1 billion) in the first quarter of this year.

Forecast of Total Addressable Market (TAM) dominated by AI solutions

Estimates putting the Total Addressable Market (TAM) at $28 trillion—with AI accounting for $23 trillion—appear to be more of a marketing play for investors. This figure factors in every conceivable economic impact of AI integration across the US economy. However, since Musk is developing both a proprietary LLM and the underlying computing infrastructure, the authors of the prospectus are formally entitled to utilize these metrics.

Segment 2. AI Division Financial Performance

According to the filings, the new AI platform encompasses the Grok model, consumer and enterprise solutions, the X (Twitter) platform, and computing clusters. The primary objective of the prospectus is to obfuscate the fact that 95% of revenues here are still generated by the social network via advertising and subscriptions.

Comparative financial analysis of the AI segment for the 2023 and 2024 periods

During this timeframe, the AI division was essentially in its infancy, and R&D expenditures neared zero. Declining ad revenues from the microblogging platform were partially offset by initial infrastructure revenues.

Dynamics of revenue and investment expenditure in AI infrastructure at the turn of 2024 and 2025

Quarterly financial results of the AI division for the first quarter of 2026

At the beginning of this year, AI revenue demonstrated modest growth of 12%. However, the situation shifted dramatically upward in Q2 due to massive inflows from capacity leasing to third-party companies.

Out of SpaceX’s total workforce of 22,000 employees, approximately 3,000 to 4,000 specialists are dedicated to the AI division.

Segment 3. Infrastructural Foundation (Colossus and Terafab)

The conglomerate’s main competitive advantage in the tech race is the phenomenal speed at which it constructs data centers.

The high-tech COLOSSUS II computing complex in Memphis, Tennessee

The company built the Colossus and Colossus II computing clusters with a combined capacity of 1 GW. The Colossus II site was launched in just 91 days, whereas the industry average for similar facilities with a 100 MW capacity stands at around 750 days. SpaceX highlights this engineering record to future shareholders a total of seven times throughout the document.

This gargantuan complex, packed with 220,000 Nvidia cards, is leased to Anthropic for $1.25 billion per month. To evaluate the efficiency of these sites, the company uses a “nameplate compute draw” metric based on the number of accelerators and their factory specifications.

Comparative chart of the nameplate compute draw across the company’s infrastructure assets

In parallel, the Terafab project—a large-scale production facility for proprietary AI chips—is being developed in partnership with Intel and Tesla. External estimates suggest capital allocation for this project could reach up to $119 billion.

Segment 4. Evolution of the Grok Model

To justify its trillion-dollar capitalization, the founder positions Grok not merely as another LLM, but as a “truth-seeking model.”

Grok AI model placement within the global dynamic leaderboard by Artificial Analysis

In global rankings, Grok firmly holds the 7th position, while ranking 2nd in terms of processing speed, making it one of the most cost-effective neural networks on the market (generation costs roughly 10 times less than Anthropic’s Claude).

Beyond text, the company is actively developing the Grok Voice engine and the Imagine multimedia system, which generates billions of images and videos monthly. The ecosystem’s total account base exceeds 1.3 billion, with approximately 117 million users regularly interacting with built-in AI features. Additionally, SpaceX has acquired an option to purchase the popular developer tool Cursor for $60 billion.

Segment 5. A Look into the Future

In the final section of the prospectus, leadership outlines truly futuristic plans: the creation of orbital data centers.

Concept deployment of space-based orbital data centers

The stated strategic objective is to launch up to 100 gigawatts of computing capacity into orbit annually, powering a new generation of AI infrastructure directly from space.