SpaceX (SPCX): the world's most ambitious company just went public

Miguel Braun · June 2026 · Special Investment Report

SpaceX · Nasdaq: SPCX

A business that combines rockets, satellite internet, artificial intelligence and data centers in space — all under one roof.

The price and the key numbers

Revenue projection

1. Rockets and reusability: the master key

To understand SpaceX you first have to understand the difference between its two rockets, because they cause a lot of confusion:

The confusion is natural: today's Starlink satellites fit perfectly on Falcon 9 and are already in orbit. Starship failing its tests doesn't affect today's business — but it is the requirement to unlock the future bets (larger satellites, orbital data centers, Mars).

The reusability revolution isn't just technical — it's economic:

2. Starlink: the cash machine with enormous upside

Starlink is SpaceX's satellite internet network and the only profitable segment today. With more than 7,000 satellites in low orbit and 10.3 million subscribers across 164 countries, it grows 100% a year in users and 50% in revenue. EBITDA margin of 63% — extraordinary for a business with so much physical infrastructure. This cash funds everything else: the rockets, the AI, the data centers.

The real upside: less than 1% of the market captured

The total global telecom market is ~$1.5 trillion. Reaching $50B in 2028 implies just 0.3% of that market. The value proposition is hard to beat: better speed, lower latency and a cost that keeps falling — especially in areas with no terrestrial infrastructure: ships, planes, rural regions, developing countries.

The catalyst ahead: direct-to-cell

The next version enables a direct connection to the smartphone — no antenna, no router. This removes the need for terrestrial infrastructure to get coverage. For traditional telecom operators, Starlink stops being a niche competitor and becomes inevitable global infrastructure. Bank models project that this feature alone could take revenue from $10B to $50B between 2025 and 2028.

3. Artificial intelligence: three bets in one

Here it helps to separate two things that often get confused. Grok (SpaceX's AI model) existed before the IPO. What's new is that SpaceX became an infrastructure vendor — like the difference between a restaurant and a wholesale ingredient distributor. One makes the final product; the other supplies the raw material to everyone who makes products. Both businesses now coexist under the same roof.

3A. The AWS of artificial intelligence (today — terrestrial data centers, already operational)

SpaceX built enormous GPU clusters to train its own models. Just as Amazon discovered it had spare server capacity and created AWS — today Amazon's most profitable business, with a third of the global cloud market — SpaceX is doing the same: renting out the compute it has to spare. In a few weeks it signed contracts with Anthropic (creators of Claude, the main competitor to ChatGPT) and Google, becoming the world's 4th hyperscaler. How did it pull this off so fast? The edge is construction speed: it stood up a 100,000-GPU cluster in 19 days. What takes anyone else 3-4 years, Elon does in under a month — designing the data centers from scratch with a rocket-engineering mindset. That speed also translates into negotiating power: every supplier (turbines, transformers, land with power) prefers to work with whoever builds fastest and has the strongest payment credibility.

Signed contracts: Anthropic ~$22-23B/GW/yr · Google ~$50B/GW/yr · Of every $10 SpaceX spends today, $7.70 goes to AI.

3B. Data centers in space: the advantage no one else can build (future — requires Starship)

This is the bet that makes SpaceX unique. On Earth, 1 gigawatt of compute capacity costs ~$25B in infrastructure alone (land, cooling, power, construction) — and that cost is inflationary because the U.S. power grid is already saturated in many regions. It's not a money problem: it's a physics problem. In space, the same gigawatt would cost ~$5B, because: free cooling (heat dissipates into the vacuum), solar power 5x more efficient (no clouds, near-continuous sunlight) and no land or building permits. The chips process in orbit and send results by laser to the Starlink antennas. Google, Microsoft and Amazon can't replicate this because they don't have the rocket. It's the only absolute competitive advantage of the 21st century in infrastructure.

Estimated cost: $5B/GW in orbit vs. $25B/GW on the ground — a 5x reduction in half of a data center's total cost.

3C. Terafab: building its own chips to depend on no one (long term)

Today SpaceX depends almost 100% on Nvidia for the GPUs it needs. That's an enormous risk: if Nvidia raises prices, has a waiting list or prioritizes another customer, the whole data center business stalls. Terafab is the answer — its own chip factory, announced on March 21, 2026 as a joint venture between Tesla, SpaceX and xAI, with an estimated investment of between $55B and $119B in Texas. It will produce two types of chips: inference chips for Tesla's cars and Optimus robots, and D3 chips designed specifically for SpaceX's orbital satellites. 80% of the output would go to orbital satellites and only 20% to terrestrial applications. Intel joined as the main manufacturing partner in April 2026, contributing its 18A process, one of the most advanced available. The logic is the same one he applied to rockets: instead of buying engines from Rocketdyne, he built them himself from scratch and ended up better and cheaper than everyone. If Terafab works, it does the same with semiconductors — closing the loop of complete vertical independence: own rockets, own satellites, own chips, own AI models.

Investment: $55B–$119B · First chips: late 2027 · Volume production: 2028 · Target: 1 terawatt of compute per year.

4. Cursor and the financial engineering of the IPO

On June 16, 2026 — days after the IPO — SpaceX closed the purchase of Cursor (Anysphere) for $60B in stock. Cursor is the leading AI coding tool: $2.6B in annualized revenue, present in 67% of Fortune 500 companies.

Why Cursor is worth $60B: the learning loop

SpaceX didn't buy a product. It bought a virtuous data cycle. Every time a developer uses Cursor, it generates signals about how real code is written — which suggestions get accepted, where people get stuck, what patterns companies use. That data feeds back into xAI's models, which improve, attracting more users, who generate more data. A loop impossible to replicate from scratch because it requires years of real interactions with millions of professional developers.

5. Unlock schedule and risks

6. The three premises for the thesis to work

For the IPO valuation to make sense you have to believe all three things at the same time. Today only one is proven:

Conclusion: a bet on the future

SpaceX is not simply a rocket company that went public. It is the only company in the world building three 21st-century infrastructures simultaneously: the most ambitious internet access network on the planet, the supply chain of space, and the computing cloud of the future — with an advantage no competitor can replicate: access to space.

Business history has few precedents of a company that in 30 days went from not existing in the AI market to becoming the world's 4th hyperscaler, surpassing Microsoft in valuation, buying the most valuable AI coding company on the market, and announcing it could reach a trillion dollars in revenue by 2030. All of this while operating the most-used launch system on the planet and the fastest-growing satellite internet network in the world.

The question is not whether SpaceX is the best business in the world today — clearly not: it loses $4B a quarter. The question is whether, five years from now when we look back, this IPO will have been the starting point of the most valuable company in history. That answer depends on three things: that Starlink keeps growing, that the AI contracts multiply, and that Starship flies. Two of the three are in development. Only one is proven. That is exactly what's thrilling about the market.

Disclaimer

This report is informational and prepared exclusively for advised clients of Fox Capital. It does not constitute personalized financial advice, nor an offer or recommendation to buy or sell securities. Projections are third-party analyst consensus estimates and do not guarantee future results. Investing involves risks including the possible loss of capital. Past returns do not guarantee future returns.