SpaceX IPO & the Gigasat Bet: What X Is Saying About the Orbital Data Center Play (2026)
By Surya Pratap
June 17, 2026
14 min read
AI & Technology
June 17, 2026
14 min read
On June 12, 2026, SpaceX opened for trading on Nasdaq under the ticker SPCX at $135 per share — and closed the day at $161, a +19% first-day pop that cemented it as the largest IPO in stock market history. The $1.75 trillion valuation eclipsed Saudi Aramco's 2019 listing and generated the kind of noise on X (Twitter) that most companies never see in a lifetime.
But the IPO was only half the story. Days before listing, SpaceX unveiled the AI1 satellite — an orbital AI data center — and announced the Gigasat factory in Bastrop, Texas, with a plan to manufacture over 1,000 of these space-based supercomputers per year. Total investment: up to $119 billion. This article breaks down what actually happened, what X is debating, and how to think about the investment thesis.
SpaceX confidentially filed with the SEC on April 1, 2026, assembling a consortium of 21 major investment banks to underwrite a $75 billion raise — the largest capital raise in IPO history. The original target was $1.5 trillion, but investor demand pushed the final price to a $1.75 trillion valuation.
One structural choice set this IPO apart from every other large listing: Elon Musk reserved roughly 30% of shares for retail investors, versus the typical 5–10% in institutional-dominated deals. The reasoning was explicit — Musk wanted the people who already use Starlink and follow SpaceX on X to have meaningful access at the IPO price rather than buying in at a secondary market premium. X lit up with reactions the moment this was confirmed.
The company's SEC filing confirmed what was already known in private markets: SpaceX has accumulated $41.3 billion in total losses since 2002. Its only profitable segment is Starlink, which generates 65–80% of company revenue. Starlink's contribution likely accounts for 40–55% of the total IPO valuation even though it is one segment of a multi-division company. Every other SpaceX division — Falcon 9, Starship, Raptor engines — runs at a loss or breaks even.
In February 2026, SpaceX acquired xAI — the parent company of the Grok AI models, the X social network, and xAI's ground-based data center infrastructure. This was not a bolt-on deal. It collapsed four previously separate Musk entities under a single public company umbrella:
This consolidation is the strategic context for the orbital data center plan. SpaceX is not simply a rocket company going public — it is positioning as a vertically integrated AI infrastructure company that controls the rocket, the launch window, the satellite, the network, the AI model, and the social platform where users interact with that model. The term analysts on X keep reaching for is “the most ambitious vertical integration play in tech history.”
SpaceX unveiled the AI1 satellite days before the IPO — a timing that was clearly deliberate. AI1 is a solar-powered supercomputer designed for low-Earth orbit, built to run AI workloads from space. Its key specifications:
| Specification | AI1 Orbital Data Center |
|---|---|
| Solar wingspan | 70 metres (wider than a Boeing 747) |
| Peak compute power | 150 kW |
| Average compute power | 120 kW |
| Ground equivalent | ~1 Nvidia GB300 rack |
| Orbit type | Low-Earth Orbit (LEO) |
| First prototypes | Early 2027 |
The key advantages SpaceX is betting on are cooling (space is naturally cold and heat dissipates freely into vacuum) and power generation (solar panels in LEO operate with no atmosphere, cloud cover, or day/night cycles the way ground installations do). Critics on X and in analyst notes are quick to point out that the economics remain unproven — the cost-per-teraflop in orbit versus a well-run ground data center has not yet been publicly validated. But SpaceX has a track record of being laughed at for cost projections that later proved correct with Falcon 9 reusability.
The vehicle for SpaceX's orbital data center ambition is the Gigasat factory, announced alongside the AI1 satellite. Located on approximately 1,000 acres in Bastrop County, Texas, at full build-out it will encompass up to 11 million square feet of manufacturing space. To put that in context: Amazon's largest fulfillment center is roughly 4 million square feet. This would be almost three of those, dedicated to building satellites.
The production target: over 1,000 AI1 satellites per year by late 2027. If you run the math, that is effectively manufacturing and launching a 70-metre solar-winged supercomputer every eight hours. The end goal is 1 gigawatt of orbital AI compute capacity annually — a figure that would represent a meaningful fraction of current global AI training capacity if the satellites reach target performance.
What makes the Gigasat plan distinctive is its vertical integration depth. SpaceX intends to manufacture the following on-site at Bastrop:
The investment numbers are staggering: $55 billion for phase one and up to $119 billion total. For comparison, the entire budget of NASA in 2025 was roughly $25 billion. SpaceX is proposing to spend nearly five times that on a single manufacturing campus over the life of this program.
X was the live commentary layer for both the IPO and the Gigasat reveal, which is fitting given that X itself is now a SpaceX asset. The conversation breaks into three distinct camps:
The loudest voices on IPO day were bullish. More than $233 million in SpaceX perpetual contracts changed hands in 24 hours before the IPO, with open interest above $263M — sustained speculative demand that suggested the $135 price would not hold the ceiling. Traders on prediction market platform Polymarket gave it better-than-even odds of closing above a $2 trillion market cap on day one. The most-shared take: “This is the most important IPO of our generation.” The bull case rests on Starlink's growth trajectory, the Starship cost-per-kilogram curve, and the bet that orbital AI compute will eventually be cheaper than ground compute at scale.
A quieter but persistent thread on X pointed at the $41.3 billion in accumulated losses and the fact that every SpaceX business unit except Starlink runs at a loss. The skeptics' argument: this is a Starlink IPO with rockets and space dreams bolted on. At $1.75 trillion, the market is pricing in decades of optionality that may or may not materialize. The AI1 satellite in particular drew pushback — the cost-per-TFLOP in orbit has never been publicly tested at scale, and ground data center economics have been improving rapidly with each generation of Nvidia silicon.
A third group on X — largely infrastructure and deep-tech investors — framed the Gigasat announcement as the more important story than the IPO itself. Their read: what SpaceX is building in Bastrop is a new category of capital asset — manufacturing capacity for space-based compute. If the orbital data center thesis is correct, the moat is not the satellite design (which can be copied) but the ability to build and launch 1,000 of them per year. That is a manufacturing and logistics advantage that takes a decade to replicate, not a software advantage that competitors can close in 18 months.
Starlink is a real, growing, profitable business with a defensible distribution advantage (satellite coverage in areas terrestrial ISPs cannot reach). If you strip out all of SpaceX's other ambitions and just value Starlink on a discounted cash flow basis, the question becomes whether Starlink alone justifies a meaningful fraction of the $1.75T valuation. At 65–80% of revenue, it very possibly does — at a premium SaaS multiple. This is the most conservative, fundamentals-based entry point.
SpaceX's core proprietary asset is Starship: a fully reusable heavy-lift vehicle that is targeting a cost-per-kilogram to orbit below $100 — an order of magnitude cheaper than any competing rocket. If Starship reaches its target cost curve, the entire business case for orbital data centers, point-to-point cargo delivery, and Mars colonization changes radically. This frame values SpaceX as a bet on Starship execution risk, not on current cash flows.
The boldest frame treats SPCX as an infrastructure bet on the next platform shift. Just as AWS captured the cloud transition by being first to industrialize it, SpaceX is attempting to industrialize orbital compute before any competitor has the launch cadence to match. If you believe AI workloads will eventually run at planetary scale and that space-based compute will be part of that infrastructure, SpaceX is the only company that could plausibly own the full stack — rocket, satellite, network, and model. The risk in this frame is timeline: the $119B Gigasat investment plays out over a decade, and a lot can change in frontier AI in 10 years.
Even if you have no interest in buying SPCX, the SpaceX strategy is worth studying as a masterclass in platform thinking. The company identified a structural bottleneck in AI (compute capacity and cost) and built backwards from it: a cheaper launch vehicle to reduce cost-per-kilogram, a manufacturing plant to scale unit production, a satellite design that removes the two biggest data center cost components (land/power in high-demand markets), and a distribution network (Starlink) that is already generating the cash to fund the next layer.
For founders building on top of AI infrastructure, there are two near-term signals to watch:
The SpaceX IPO is not just the largest public offering in history — it is a statement about where the company believes the next 20 years of technology infrastructure will be built. The Gigasat factory and AI1 satellite are not side projects; they are the core thesis for why SpaceX is worth $1.75 trillion rather than just what Starlink alone would command.
The bull case is compelling: a company that controls the rocket, the factory, the satellite, the network, and the AI model — and is the only entity on Earth capable of launching 1,000 orbital data centers per year — deserves a significant infrastructure premium. The bear case is real: the unit economics of orbital compute are unproven, the timeline is a decade, and the ground compute industry will not be standing still.
What is not debatable is that SpaceX has changed the conversation about where AI infrastructure goes next. The question is no longer whether data centers will eventually exist in orbit. It is whether SpaceX will be the company that builds them — and whether the window between now and 2027 proves the economics out.
We help founders go from idea to launched MVP in 4–8 weeks — designing for the AI infrastructure shift, not against it. 15+ MVPs shipped. $2.4M+ raised by our founders.
Book a Free Discovery Call
