SpaceX Is Profitable — And Building AI in Orbit

While most of the aerospace and tech world is still burning through investor capital with no clear path to profitability, SpaceX is doing something almost no one expected from a rocket company: making serious money — and quietly positioning itself to dominate the next era of AI computing.

📊 Key Figures

Metric	Value	Context
2025 Revenue	~$15.5B	Up significantly YoY
2025 Profit	~$8B	~52% profit margin
Starlink Revenue (2025)	~$10B	~65% of total revenue
Starlink Users (early 2026)	9M+	Global subscribers
2026 Revenue Projection	$22B–$24B	Starlink ~70% of total
SpaceX Valuation (Dec 2025)	~$800B	IPO target: ~$1.5T
Orbital Satellites Planned	Up to 1,000,000	FCC filing Jan 30, 2026
Terafab Facility Cost	$20B–$25B	Tesla/SpaceX/xAI JV, Austin TX

The Profitability Gap Is Real

The contrast Whole Mars Catalog is pointing to is stark. Numerous aerospace and EV competitors are still operating at deep losses, relying on capital raises and government subsidies to survive. SpaceX, by contrast, generated approximately $8 billion in profit on $15.5 billion in revenue in 2025 — an operating margin that most software companies would envy, let alone a company that builds and launches rockets.

The engine behind that profitability is Starlink. The satellite internet service contributed roughly $10 billion in revenue last year and now serves over 9 million subscribers globally. For 2026, analysts project total SpaceX revenue to climb to between $22 billion and $24 billion, with Starlink accounting for approximately 70% of that figure. That's not a side business anymore — it's the core of one of the most valuable private companies on Earth, with a valuation that reportedly reached $800 billion in December 2025.

The LLM Nobody Talked About

The more surprising element of the Whole Mars Catalog post is the mention of SpaceX developing its own major large language model. This connects directly to SpaceX's acquisition of xAI — Elon Musk's AI company — on February 2, 2026. That merger wasn't just a financial move. It was the foundation of a vertically integrated AI platform that combines rockets, satellite internet, direct-to-mobile communications, and now AI compute in a single stack.

The strategic logic: SpaceX already owns the delivery infrastructure (Falcon 9, Starship), the connectivity layer (Starlink), and now the AI intelligence layer (xAI/Grok). Building a proprietary LLM means SpaceX doesn't need to rely on OpenAI, Google, or Anthropic for the AI capabilities that will power its next-generation services — including whatever runs inside those orbital data centers.

Orbital AI: The Plan to Compute in Space

The space-based data center plan is where things get genuinely ambitious. On January 30, 2026, SpaceX filed an application with the FCC seeking approval to launch up to one million satellites specifically designed as orbital AI compute nodes. These aren't communication satellites — they're data centers in space.

The satellites — dubbed 'AI Sat Mini' — are designed to harness continuous solar power in orbit, bypassing the land constraints and power grid limitations that are throttling terrestrial AI data center expansion. SpaceX's custom 'D3' chip is being developed specifically for the thermal and radiation environment of space. Musk has stated that within two to three years, space will be the lowest-cost method for generating AI compute — a claim that sounds audacious until you run the math on what solar availability looks like at 1,000 km altitude versus building a data center in Texas during a heatwave.

Supporting all of this on the ground is Terafab — a planned $20–25 billion chip fabrication facility in Austin, Texas, structured as a joint venture between Tesla, SpaceX, and xAI. The facility is designed to produce one terawatt of computing power annually to feed the orbital constellation. Long-term, SpaceX has outlined plans to establish data centers and manufacturing facilities on the Moon within 10 years.

🔭 The BASENOR Take

📰 Deep Dive

What makes SpaceX's position genuinely unusual is the compounding nature of its advantages. Starlink's profitability funds launch cadence. Launch cadence funds orbital infrastructure. Orbital infrastructure enables space-based compute. Space-based compute powers the LLM. The LLM enhances every product in the ecosystem. Most companies trying to build AI infrastructure are buying their way in with outside capital. SpaceX is funding its AI ambitions from cash flow — which is exactly what Whole Mars Catalog is pointing to when they contrast SpaceX's trajectory against competitors burning billions with no end in sight.

The orbital data center plan is the most speculative piece of this, and it's worth being clear-eyed about the risks. A million-satellite FCC application is not the same as a million satellites in orbit. Regulatory approval, manufacturing scale, launch cadence, and the actual performance of radiation-hardened AI chips in space are all open questions. But the FCC filing is real, the xAI merger is closed, and the financial foundation to attempt all of this is genuinely in place — which is more than most competitors can say.

The IPO angle adds another dimension. A potential mid-June 2026 offering targeting a $1.5 trillion valuation — nearly double the $800 billion December 2025 figure — would be one of the largest public offerings in history. If SpaceX goes public while simultaneously holding FCC approval for an orbital AI constellation and a closed xAI merger, the narrative it brings to market is unlike anything investors have seen before. For the broader Musk ecosystem, including Tesla, a liquid, publicly valued SpaceX changes the strategic calculus on shared ventures like Terafab considerably. Follow our SpaceX coverage as this story develops.

---

Sources verified at publish time. Spotted an inaccuracy? Email editorial@basenor.com.