The $3.2 Trillion Bet: Wall Street Sees SpaceX AI Revenue Exploding 100-Fold by 2030

 

 The $3.2 Trillion Bet: Wall Street Sees SpaceX AI Revenue Exploding 100-Fold by 2030

**Subtitle:** *From $32 billion to $3.2 trillion—Goldman and Evercore are making their case for a $1.8 trillion IPO. But with xAI founders gone, Colossus leased to a rival, and only two profitable quarters in history, is this the greatest growth story ever—or the most expensive hallucination?*

**Reading Time:** 8 Minutes | **Category:** Markets & Artificial Intelligence

## Introduction: The Number That Defies Comprehension

There is a number being whispered in the hallways of Goldman Sachs, Evercore ISI, and every major investment bank involved in the SpaceX IPO. It is a number so large that it strains credulity. It is a number that, if realized, would make SpaceX not just the largest company in the world, but one of the largest economic entities on the planet.

That number is **$3.2 trillion**.

According to a Goldman Sachs research note shared with potential IPO investors, the bank projects that SpaceX's artificial intelligence revenue will surge from approximately **$32 billion in 2025 to a staggering $3.22 trillion by 2030** —an increase of roughly **100 times** in just five years .

Let that sink in. $3.22 trillion in annual revenue from AI alone. To put that in perspective, that is more than the combined 2025 revenues of Amazon Web Services ($107 billion) and Nvidia ($115 billion) put together—by a factor of nearly 15 .

Total company revenue is projected to hit **$4.74 trillion** by 2030, with AI accounting for about 68% of the total . Starlink, the satellite internet business that is currently SpaceX's only profitable division, is projected to bring in $1.44 trillion. The launch business? A mere $83 billion .

Evercore ISI is even more aggressive. The firm projects AI revenue reaching **$3.31 trillion by 2030** and **$7.55 trillion by 2031**, with total company revenue surpassing **$1 trillion next year** .

These are not numbers. These are fever dreams. And yet, Wall Street is treating them with deadly seriousness as SpaceX prepares for what could be the largest IPO in history—targeting a $1.75 trillion valuation and a $75 billion raise .

In this deep-dive, we will break down the "orbital compute" thesis that underpins these projections, analyze the strange and troubled history of xAI that led to its dissolution and merger into SpaceX, and lay out the bull case and bear case for what might be the most consequential IPO of the decade.

> **The Bottom Line Up Front:** Wall Street's case for SpaceX rests on a single, monumental bet: that Elon Musk can build the world's largest AI infrastructure business, selling orbital computing power at a scale that dwarfs AWS, Azure, and Google Cloud combined. The bears say this is pure fantasy. The bulls say it is the future. The truth, as always, is somewhere in between—but the IPO price leaves no room for error.

## Part 1: The "Orbital Compute" Thesis – Why AI Belongs in Space

The core of the Goldman Sachs and Evercore projections is not grounded in the AI models that xAI built (Grok), nor in the rocket launches that made SpaceX famous. It is grounded in a concept that sounds like science fiction: **orbital compute**.

### The Terrestrial Bottleneck

Here is the problem that Elon Musk claims to have identified. Terrestrial data centers—the giant warehouses filled with Nvidia GPUs that power the AI revolution—are hitting fundamental limits . They require enormous amounts of electricity, vast quantities of water for cooling, and massive tracts of land. In many regions, the grid cannot handle the load. In others, environmental regulations block expansion.

Musk's solution is as audacious as it is simple: put the data centers in space.

**The SpaceX Advantage:** SpaceX is the only company on Earth with the launch cadence, orbital delivery cost structure, and constellation operations experience to make orbital compute a reality . Starlink already operates the largest satellite constellation in history. The same infrastructure that delivers internet to rural Iowa could, in theory, host AI inference workloads.

**The Colossus Precedent:** xAI's Colossus supercomputer in Memphis, Tennessee, currently houses over 220,000 Nvidia GPUs, including H100, H200, and next-generation GB200 accelerators . That facility is being leased to Anthropic for Claude's training and inference needs. But Musk has already announced plans for Colossus 2, which will be the world's first gigawatt-level AI training cluster . The thinking is that later iterations of Colossus will move to orbit.

### The Terafab Ambition

On May 6, 2026—the same day xAI was formally dissolved into SpaceX—the company filed paperwork for a semiconductor fabrication facility called **Terafab** in Grimes County, Texas . The project carries an estimated cost of $550 billion to $1.19 trillion.

Yes, you read that correctly. Over a trillion dollars to build a chip factory.

If Terafab comes online as planned, SpaceX would become the first AI company to own its entire vertical stack: from chip design and fabrication to supercomputer assembly to inference hosting. This is the "AWS of Space" thesis—a vertically integrated compute utility that has no rivals because no one else can afford to build it .

**The Human Touch:** For the investor considering buying SpaceX shares at $135, the Terafab number is either the most compelling argument for the valuation or the most terrifying risk factor. A $1 trillion capital commitment requires a $1 trillion payoff. If orbital compute fails to materialize, that debt will sink the company.

### The "Colossus x Anthropic" Deal

The lease of Colossus 1 to Anthropic is the first real-world validation of the orbital compute thesis . Anthropic, which competes directly with OpenAI and Google, is paying SpaceX for exclusive access to the 220,000-GPU cluster.

But the deal has a twist that cuts against the "space compute" narrative. Why is Anthropic using a terrestrial cluster if the future is in orbit? The answer is that orbital compute is not ready yet. Colossus 1 is a proof of concept. The orbital data centers are a promise.

The deal also highlights the strange irony of xAI's demise. Musk founded xAI specifically to compete with OpenAI . The founders—all 11 of them, as we will explore—have since left. The company was dissolved. And its crown jewel asset, the Colossus supercomputer, is now being used to power xAI's *competitor*, Anthropic. The "war against OpenAI" has become a revenue stream for OpenAI's rival.

**The Creative Angle:** The orbital compute thesis is the most speculative element of the SpaceX investment case. It is also the most essential. Without it, the $3.22 trillion AI revenue projection collapses. With it, SpaceX becomes the most important infrastructure company of the 21st century. The binary outcome is either $0 or $10 trillion.

## Part 2: The Wall Street Forecasts – Goldman, Evercore, and the $1 Trillion Club

Let us look at the numbers in detail, because the scale of the projections is the entire story.

### The Goldman Sachs Model

Goldman's base case, shared with IPO investors, projects the following:

| Segment | 2025 Actual | 2030 Projected | Growth |

| :--- | :--- | :--- | :--- |

| **AI (xAI)** | $32 billion | $3,220 billion | **~100x** |

| **Starlink (Connectivity)** | $114 billion | $1,440 billion | **~12.6x** |

| **Launch Services (Rocket)** | $41 billion | $83 billion | ~2x |

| **Total Revenue** | $187 billion | $4,743 billion | **~25x** |

*Sources: *

Goldman projects that SpaceX's EBITDA will surge from $6.6 billion in 2025 to **$352 billion by 2030** . That would make SpaceX more profitable than most countries.

The model assumes that Starlink will continue to expand globally, capturing market share from terrestrial broadband providers in rural and remote areas. It assumes that the launch business will grow at a modest pace, driven by NASA Artemis contracts, Pentagon Starshield missions, and commercial satellite deployments.

But the AI number is the engine. Without it, the total company revenue in 2030 would be roughly $1.5 trillion—still massive, but less than a third of the projected total.

### The Evercore ISI Projections

Evercore is even more aggressive . The firm projects:

| Segment | 2025 Actual | 2030 Projected | 2031 Projected |

| :--- | :--- | :--- | :--- |

| **AI (xAI)** | $32 billion | $3,310 billion | $7,550 billion |

| **Starlink** | $114 billion | $1,470 billion | $1,770 billion |

| **Launch** | $41 billion | $83 billion | $86 billion |

| **Total** | $187 billion | $4,863 billion | $9,406 billion |

Evercore's model assumes that AI will account for **74% of SpaceX's revenue by 2031**, up from less than 20% in 2025. The launch business will shrink to just 1% of total revenue—a stunning reversal of priorities for a company named "Space Exploration Technologies."

### The Morningstar Contrarian View

Not everyone is buying the hype. Morningstar, the independent research firm, has published a fair value estimate for SpaceX of just **$780 billion**—less than half the IPO target .

Morningstar's analysts argue that the AI projections are "implausible" and that the orbital compute thesis is "highly speculative with a material threat of value destruction" . They also point to the governance risks of Musk's super-voting control and the lack of any profitable track record outside of Starlink.

| Firm | 2030 AI Revenue Projection | Fair Value | Recommendation |

| :--- | :--- | :--- | :--- |

| **Goldman Sachs** | $3.22 trillion | Implied: $1.75T+ | Participate in IPO |

| **Evercore ISI** | $3.31 trillion | Implied: $1.8T+ | Participate in IPO |

| **Morningstar** | N/A (implicitly low) | $780 billion | **Avoid** |

| **ARK Invest** | N/A (implicitly high) | $2.5 trillion | Buy aggressively |

*Sources: *

### The Capex Tsunami

One more number worth noting: capital expenditures. Goldman projects that SpaceX's capex will rise from $207 billion in 2025 to **$360 billion in 2030** . Evercore is even more extreme, projecting capex of **$732 billion in 2031**, with $666 billion allocated to AI infrastructure .

These numbers are not sustainable for any company without massive, consistent profitability. The gap between the capex projections and the EBITDA projections is the risk factor that keeps the bears up at night.

**The Human Touch:** For the retail investor, the capex numbers are a warning. This is not a capital-light software business. This is the most capital-intensive business in history. If the revenue does not materialize, the debt will be crushing. The IPO is not a "safe bet." It is a venture capital bet dressed up in public market clothing.

## Part 3: The xAI Implosion – What Happened to Musk's AI Lab?

To understand the AI projections, you have to understand the troubled history of xAI. The company that is supposed to generate $3.22 trillion in revenue by 2030 does not technically exist anymore. It was dissolved on May 6, 2026, and merged into SpaceX .

### The Founder Exodus

xAI was founded in 2023 with 11 co-founders, drawn from DeepMind, OpenAI, Microsoft Research, and Tesla . The mission was to "understand the universe"—and more specifically, to compete with OpenAI, which Musk felt had betrayed its founding principles.

By May 2026, **all 11 founders had left** . The final departures came in March 2026, when pre-training lead Manuel Kroiss and Musk's long-time associate Ross Nordeen resigned.

The reasons for the exodus are not fully public, but industry insiders point to a familiar pattern with Musk-led ventures: brilliant talent is attracted, then repelled by chaotic management and an inability to focus on product over spectacle.

### The Product Gap

xAI's flagship product, Grok, has a small but loyal following. Apptopia data shows its U.S. mobile market share grew from 1.9% in January 2025 to 17.8% in January 2026 . Global web share is about 3.4%.

But Grok is not a serious competitor to Claude or ChatGPT in the enterprise market. Claude Code generated an estimated $25 billion in annualized revenue in 2025. ChatGPT Enterprise has millions of customers. Grok has no enterprise product at all .

The "personality" that makes Grok appealing to Musk's fans—snarky, irreverent, willing to "say what the others won't"—is precisely what makes it unpalatable to corporate clients. No CFO wants their internal AI to go off the rails with a political rant.

### The Colossus Lease

With the founders gone and the enterprise product non-existent, Musk faced a difficult choice. The Colossus supercomputer—built at enormous expense—was sitting idle. The training runs that had once consumed its cycles were no longer happening.

The solution was to lease the compute to Anthropic .

The deal gives Anthropic exclusive access to the 220,000-GPU cluster for Claude training and inference. In return, SpaceX receives a steady stream of revenue.

The irony is inescapable. xAI was founded to kill OpenAI. Instead, its most valuable asset is now feeding xAI's competitor .

**The Human Touch:** For the employee who joined xAI to work on frontier AI models, the dissolution of the company and the leasing of Colossus to Anthropic is a betrayal. They signed up to compete with OpenAI, not to build infrastructure for its rival. The "mission" was replaced by the "asset."

## Part 4: The Bull Case – Why $3.22 Trillion Could Be Low

If the orbital compute thesis is correct, the Goldman projections might actually be *conservative*.

### The "AWS of Space" Moat

No other company can do what SpaceX is proposing. Amazon, Google, and Microsoft can build terrestrial data centers. They have the capital, the expertise, and the customer base. But they cannot launch payloads into orbit at SpaceX's cost structure. They cannot build the satellite constellation that would support orbital compute. They cannot operate the gigawatt-scale solar arrays that would power it.

If orbital compute works, SpaceX will have a natural monopoly. The moat is not just wide; it is astronomical.

### The Terafab Vertical Integration

The Terafab chip factory is the second piece of the moat . If SpaceX can design and manufacture its own AI accelerators, it will no longer be dependent on Nvidia. The margins on compute will be even higher.

And if Terafab can also produce chips for other customers, SpaceX could become a competitor to TSMC and Samsung—not just a customer.

### The Musk Premium

Finally, there is the Musk factor. As Morningstar notes, the valuation is "difficult to justify using conventional metrics" . That is because investors are not buying a launch company. They are buying a belief system.

Cathie Wood's ARK Invest projects a $2.5 trillion enterprise value for SpaceX by 2030 . That is in the same ballpark as the IPO target. But ARK's models assume that Musk's futuristic ambitions—Mars colonies, orbital manufacturing, asteroid mining—will begin generating revenue by the end of the decade.

The "Musk Premium" is the willingness of investors to pay for dreams, not just earnings. And in the current market, that premium is as high as it has ever been.

**The Human Touch:** For the true believer, the $3.22 trillion projection is not a fantasy. It is a floor. The orbital compute market, they argue, is not $3 trillion—it is $30 trillion. Every data center on Earth will eventually move to space. Every AI inference will run on a SpaceX satellite. The valuation is a bargain.

## Part 5: The Bear Case – The "Most Expensive Hallucination in History"

The other side of the argument is equally compelling.

### The "No Profit" Track Record

SpaceX has been profitable in exactly two quarters of its 24-year history: Q2 2023 and Q3 2023 . In 2025, the company lost $4.95 billion. In Q1 2026, it lost $4.27 billion .

Yes, the losses are driven by AI capital expenditures. But capital expenditures do not guarantee revenue. The 220,000 GPUs at Colossus are only generating revenue now because they are leased to Anthropic—not because xAI built a successful product.

### The "Grok is Dead" Reality

xAI no longer exists as an independent company. Its flagship product, Grok, has no enterprise presence and a minuscule market share. The founders are gone. The talent has scattered.

The $3.22 trillion AI revenue projection assumes that SpaceX will build a business that is currently non-existent. There is no path to that number that does not involve a radical change in strategy—or a radical change in the competitive landscape.

### The Technical Hurdles

Orbital compute sounds elegant. It also faces enormous technical hurdles :

- **Latency:** The speed of light is fast, but it is not infinite. Round-trip latency to a satellite in low Earth orbit is measured in milliseconds. For many AI applications, that is fine. For real-time inference, it is not.

- **Radiation:** Chips in space degrade faster than chips on Earth. The radiation environment is hostile. The lifetime of a GPU in orbit measured in years, not decades.

- **Cooling:** In the vacuum of space, there is no air to carry away heat. Radiators can work, but they are massive and expensive.

- **Power:** Solar panels can generate electricity, but at a fraction of the density of a terrestrial grid connection. A gigawatt-scale orbital data center would require square miles of solar panels.

These hurdles are not insurmountable. But they are expensive. And the timeline for solving them is measured in decades, not years.

### The Governance Risk

Finally, there is the Musk governance issue. Musk will control approximately 82-85% of the voting power after the IPO . Public shareholders will have no say in how the company is run.

If Musk decides that Mars colonization is more important than AI revenue, there is nothing shareholders can do. If Musk decides to move the company to Texas or dissolve the board entirely, there is nothing shareholders can do.

The Morningstar fair value estimate of $780 billion includes a "governance discount" to account for this risk .

**The Human Touch:** For the institutional investor, the governance risk is the dealbreaker. They are being asked to pay $135 per share for a company where their vote is essentially worthless. That is not investing. That is patronage.

## Frequently Asked Questions (FAQ)

**Q: What is the Goldman Sachs AI revenue projection for SpaceX?**

A: Goldman Sachs projects that SpaceX's AI revenue will grow from approximately $32 billion in 2025 to **$3.22 trillion in 2030** —an increase of about 100 times .

**Q: When will SpaceX go public?**

A: SpaceX is targeting an IPO on the Nasdaq under the ticker **SPCX** as early as June 12, 2026. The company is offering 555.5 million shares at $135 each, aiming to raise $75 billion at a valuation of approximately $1.75 trillion .

**Q: What is "orbital compute"?**

A: Orbital compute is the concept of hosting AI data centers on satellites rather than on Earth. Proponents argue that space offers unlimited solar power, natural cooling, and no land-use restrictions. SpaceX is uniquely positioned to build this infrastructure because of its low-cost launch capabilities and Starlink constellation .

**Q: Why did xAI dissolve?**

A: xAI was formally dissolved on May 6, 2026, and merged into SpaceX. All 11 founding members left the company over the preceding months. xAI's flagship product, Grok, failed to gain enterprise traction, and its most valuable asset—the Colossus supercomputer—has been leased to competitor Anthropic .

**Q: Is SpaceX profitable?**

A: SpaceX has been profitable in only two quarters of its history (Q2 and Q3 2023). The company lost $4.95 billion in 2025 and $4.27 billion in the first quarter of 2026, driven primarily by AI-related capital expenditures .

**Q: What is Morningstar's valuation of SpaceX?**

A: Morningstar has published a fair value estimate of **$780 billion** for SpaceX, less than half of the IPO target valuation. The firm argues that the AI projections are "implausible" and that the orbital compute thesis is "highly speculative with a material threat of value destruction" .

**Q: Should I buy SpaceX stock at the IPO?**

A: (Disclaimer: Not financial advice.) The bull case rests on a belief that orbital compute will become a $3 trillion+ market by 2030 and that SpaceX will capture the majority of it. The bear case points to the lack of profitability, the exodus of xAI's founders, the governance risks of Musk's voting control, and the enormous capital expenditures required. The gap between the two views is so wide that it may be best to watch the first few days of trading from the sidelines .

## Conclusion: The Most Important Stock of the Decade

We started this article with a number: $3.22 trillion. That is the AI revenue that Goldman Sachs believes SpaceX can generate by 2030.

We end with a question: *Is that number visionary or delusional?*

The answer will determine the fate of the largest IPO in history. If the orbital compute thesis is correct, $135 per share will look like a bargain in five years. If it is not, the stock could follow the path of many hype-driven IPOs: a surge on the first day, followed by a long, slow decline into penny-stock territory.

**For the Believer:**

This is the chance to own a piece of the infrastructure that will power the next industrial revolution. The orbital compute thesis is audacious, but so was landing a rocket on a drone ship. SpaceX has a habit of doing the impossible.

**For the Skeptic:**

The numbers do not add up. The company has no profitable track record. The AI division has collapsed. The founders have left. The valuation is based on a concept that does not exist yet. This is a gamble, not an investment.

**For the Curious:**

Watch the first week of trading. The volatility will be extreme. The IPO is a test of whether the public markets are willing to pay for Musk's vision—or whether the "retail frenzy" has finally reached its limit.

**The Bottom Line:**

Wall Street sees SpaceX AI revenue exploding 100-fold by 2030. The projections are either the most ambitious in financial history or the most expensive hallucination. The IPO price of $135 per share leaves no room for error.

The rocket is on the pad. The countdown has begun. On June 12, we will find out if the world is ready for liftoff.

---

**#SpaceXIPO #ElonMusk #AIInfrastructure #OrbitalCompute #Investing #SpaceX #IPO2026**

---

*Disclaimer: This article is for informational purposes only. It does not constitute financial advice. IPO price targets are subject to change; past performance does not guarantee future results. Always consult a licensed professional before making investment decisions.*