Navigating the storm to find the steady current.
The Hook: A Trillion-Dollar Fiction
In early April, Morgan Stanley released a blockbuster investment analysis on SpaceX. The headline numbers were staggering: 2025 revenue of $18.7 billion projected to skyrocket to $319 billion by 2030 and a jaw-dropping $33 trillion by 2040. The driver? An AI satellite constellation dubbed “Starmind.” The target price: $300 per share. But the stock had already retraced from a post-IPO high of $225 to $125. To any market veteran, this smells less like an investment thesis and more like a carefully constructed narrative — one designed to sell shares, not to reflect reality.
Reading the code that writes the culture.
Context: The Narrative Machine
We’ve seen this playbook in crypto many times. A company with a solid if unglamorous core business — in SpaceX’s case, launch services and Starlink connectivity — requires a sexy new story to justify a multi-trillion-dollar valuation. Enter Starmind: an “orbital AI data center” powered by Starship, promising to revolutionize compute by moving it to low Earth orbit. The narrative is irresistible: AI meets space, transcendence of physical limits. Morgan Stanley’s analyst Adam Jonas sold it as the next frontier of exponential growth.
But let’s examine the mechanics. The report’s total addressable market (TAM) for Starmind is $28.5 trillion, of which $26.5 trillion is AI-related. That is essentially the entire projected AI market decades out. Yet the report offers zero evidence that space-based compute can capture even a fraction of that — no customer contracts, no pricing model, no unit economics. This is narrative engineering at its finest, exactly the same trick used during the ICO boom of 2017, where whitepapers would claim TAMs of all global banking fees.
Core: Deconstructing the Narrative Mechanism
As someone who audited over 50 ICO whitepapers in 2017, I learned to spot the telltale signs of narrative-driven valuation. The Morgan Stanley report has every single one. Let’s take the core claim: Starmind will enable “orbital AI computing” for low-latency global inference. Sounds plausible? Not when you dig into the physics.
Technical Absence: The Empty Promise
The report offers zero technical detail on how Starmind is supposed to work. What chips will these satellites use? Nvidia H100s? They require 700W per GPU — far beyond the power generation capability of any current satellite, even with huge solar arrays. How do you dissipate heat in vacuum? Radiative cooling demands massive surface area, increasing weight and launch cost exponentially. The communication bottleneck? Laser links within the constellation may be fast, but downlinking results to Earth remains narrow — a few gigabits per second at best, compared to the terabytes per second of a modern data center. These are not minor engineering problems; they are existential barriers.
From my experience in cybersecurity, I know that any system that promises to solve physical limitations through narrative alone is a red flag. The Starmind concept is technically premature by at least a decade, if not more. The report never mentions these constraints.
Economic Absurdity: The $33 Trillion Question
The $33 trillion revenue figure for 2040 is pure fiction. To put it in perspective: the entire global GDP in 2023 was about $105 trillion. For one company to capture $33 trillion in revenue — roughly 30% of global economic output — is economically nonsensical. This is not a forecast; it’s fairy dust meant to distract from the lack of a viable business model. In crypto, we saw similar numbers from Terra (LUNA) when it projected itself as absorbing all global payments. We know how that ended.
Even if Starmind launched tomorrow and miraculously worked, the cost per FLOP in orbit would be orders of magnitude higher than on Earth. The launch cost alone for a single AI satellite (even with Starship) would be in the millions. Ground-based data centers enjoy Moore’s Law, cheap electricity, and massive cooling economies of scale. There is no plausible cost advantage for space compute.
Sentiment Analysis: The Market Is Not Buying
The stock price tells the story. From a post-IPO peak of $225 down to $125, the market rejected the initial narrative. Morgan Stanley’s $300 target is a call for the narrative to be re-embraced, but the price action suggests skepticism. Using on-chain data and sentiment tools (as I do for crypto narratives), we can see that social media mention volumes for “SpaceX AI” spiked during the report release but quickly decayed. The hype cycle is peaking, not accelerating.
Contrarian Angle: The Real Opportunity Is in the Core Business
Here’s the counter-intuitive insight that most analysts miss: the most valuable part of SpaceX may be exactly the parts the narrative wants to downplay — its launch monopoly and Starlink’s growing subscriber base. Starlink alone is projected to generate $12-15 billion in revenue by 2025, with a path to $30 billion by 2030. The launch business, with Starship reducing cost per kg to orbit by an order of magnitude, could capture a huge share of the global launch market. A sum-of-the-parts valuation for these two businesses alone yields $1500-2000 billion, or roughly $100-150 per share — close to the current price. That’s a reasonable, honest valuation.
The Starmind narrative is actually a liability. It distracts management from executing on Starlink profitability and launch reliability. It invites regulatory scrutiny (militarization concerns, spectrum allocation) and signals to investors that the company is chasing hype rather than delivering on its core mission.
Moreover, the contrarian play is to recognize that the hype itself creates a short-term trading opportunity. If the market is overly exuberant about Starmind, one could short SpaceX’s stock or derivatives (if available) when the narrative peaks. The collapse of similar narratives (Pelosi AI trades, SPAC space companies) suggests a mean reversion is likely.
Takeaway: What to Watch For
As a narrative hunter, I know that every hype cycle leaves a trail of milestones. The real test for Starmind is not the $33 trillion dream but the next 12-24 months of engineering reality. Watch for these signals: - The next Starship launch: does it carry any AI compute payload? If not, the timeline is years away. - SpaceX hiring: do they post job openings for space-grade ASIC designers or orbital data center software engineers? If not, the project is still a Powerpoint. - Customer announcements: is there a single anchor tenant for Starmind? A government contract or a partnership with a cloud provider?
If none of these appear by mid-2027, treat the narrative as dead. The code writes the culture, but only real engineering can sustain it.
