The code doesn't lie. But the narrative does. And when a $187 billion revenue rocket is projected to become a $33 trillion behemoth in 15 years, the code — or at least the fundamental laws of physics, economics, and engineering — is screaming a hard stop.
Over the past week, a fragmented report attributed to Morgan Stanley has been bouncing through Telegram groups and crypto Twitter. The headline: SpaceX will hit $33 trillion in annual revenue by 2040, driven by an "AI-powered orbital infrastructure." The source? A blockchain/Web3 aggregator, not the bank's official research terminal. Even before I open the log, that's a red flag hung on the periscope.
I've spent years debugging smart contracts during the 2017 ICO gold rush and dissecting the Terra LUNA de-pegging code in 2022. I learned one thing: when a story promises exponential returns without showing you the build artifacts, you're not reading analysis — you're reading a marketing deck. Let me treat this report the same way I treat a new DeFi protocol: extract the core claims, stress-test the assumptions, and look for the hidden re-entrancy vulnerabilities in the logic.
Context: What's Actually Being Said?
The report positions SpaceX as the future backbone of global AI compute. The idea: Starlink's satellite constellation will evolve from a broadband network into a distributed, orbital data center capable of handling AI inference — and eventually training — at scale. Morgan Stanley's analysts apparently slapped a 300 price target (presumably in some private market valuation) on the back of this vision, projecting revenue growth from $187 billion in 2025 to $319 billion by 2030 and eventually to $33 trillion by 2040.
Let me pause here. $33 trillion is roughly equal to the combined GDP of the United States and China today. This single company would generate more revenue than any nation-state on earth. The report's source claims this is based on "AI infrastructure" demand. But it never defines what that infrastructure looks like, how it's built, or who pays for it. That's not a financial model — that's a fever dream with a discount rate.
Core Analysis: The Three Technical Bombs That Deflate the Balloon
I'll break this down the way I break down a suspicious smart contract: variable by variable.
Bomb #1: The physics of orbital compute don't add up.
AI training is energy-intensive. A single H100 GPU draws 700 watts under load. A satellite the size of a Starlink V2 bus (roughly 1.5 tons) can support maybe 10 kilowatts of total power from its solar arrays. That's enough to run 14 H100s — assuming you can cool them in a vacuum and protect them from radiation without extra weight. To match the compute of a single modest data center (say, 100,000 H100s), you'd need 7,000 dedicated compute satellites. That's 10,000 tons of hardware to launch — at a time when SpaceX launches roughly 100 tons per month total. And that's just for inference. Training a frontier model like GPT-6 would require tens of thousands of these clusters, all interconnected by laser links that themselves consume power.
The code doesn't lie: the energy budget is the first hard constraint. Starlink satellites have a power-to-mass ratio that makes orbital AI compute an order of magnitude more expensive than ground-based alternatives. Even if SpaceX cuts launch costs to $100/kg, the marginal energy cost alone — converting sunlight to compute — is 10x that of terrestrial grid power. The report assumes a technology leap that simply doesn't exist in any public patent or engineering roadmap.
Bomb #2: The growth curve is absurd.
From $187 billion to $319 billion in 5 years implies a 14% CAGR — reasonable for a scaling tech company. But from $319 billion to $33 trillion in 10 years implies 60% CAGR, sustained for a decade. That's beyond any industrial revolution metric. Even Amazon during peak AWS growth never exceeded 40% for more than 4 years. The market cap of the entire global telecom industry today is about $2 trillion. The report wants SpaceX to own 16 times that in a single vertical.
I debugged bots that tried to front-run NFT mints. They failed because the gas market didn't scale linearly with demand. Similarly, the compute market doesn't scale without massive infrastructure bottlenecks. The report doesn't account for the fact that building orbital compute requires launching millions of custom ASICs — hardware that currently doesn't exist, would cost billions to develop, and would face years of radiation testing.
Bomb #3: The institutional flow doesn't match.
In 2024, I developed a tool to track on-chain movements from Galaxy Digital and Fidelity wallets to gauge institutional Bitcoin demand. The data was clear: institutions buy when there's a clear value proposition. For SpaceX's orbital AI, there is none. AWS, Google, and Microsoft are already building ground-based AI infrastructure at unprecedented scale. They own the customer relationships and the software stack. SpaceX would have to compete with the three most capitalized companies in the world — while simultaneously convincing them to become customers. That's not a business model; that's a hostage negotiation.
Liquidity is just trust with a timeout. And right now, trust in this narrative is all front-loaded. There's no on-chain evidence of institutional accumulation around this thesis. No large wallets moving into SpaceX secondary shares. No VC announcing a dedicated orbital AI fund. The buzz is coming from crypto Twitter, not from Goldman Sachs trading desks.
Contrarian Angle: What the Hype Misses
Retail traders see this report and think "buy the dip on SpaceX private shares" or "short the broader market because satellite compute will eat the world." Both are wrong. The real smart money play is the exact opposite of what the narrative suggests.
The more this story circulates, the more it exposes a deep anxiety among Wall Street analysts: they need a new growth narrative for SpaceX because the base business — Starship and Starlink — faces serious headwinds. Starship hasn't reached orbit with payload yet. Starlink's subscriber growth is slowing. The financial models need a new catalyst to justify the $150+ billion private valuation. AI is the perfect plug-and-play catalyst.
But smart contracts are cold, and margins are warm. The only verified signal here is that space-grade chip startups (like those building radiation-hardened FPGAs and ASICs) will see increased RFP activity from satellite operators desperate to add compute to their payloads. That's the real infrastructure play. Not SpaceX itself, but the pick-and-shovel suppliers: companies like Teledyne e2v, Cobham, and a handful of stealth startups.
Also, consider the regulatory thundercloud. A single private company running a global AI inference network raises data sovereignty flags in every country from France to India. The EU's AI Act and China's data localization laws would effectively block Starlink's AI features in the world's largest markets. The report assumes a frictionless regulatory environment that doesn't exist and may never.
Static analysis misses the human variable. And the human variable here is geopolitics.
Takeaway: Three Verifiable Signals to Watch
Forget the $33 trillion. That number is a Rorschach test for your biases. Instead, watch these three concrete data points over the next 12 months: 1. Did SpaceX file a patent for an orbital AI compute module? If not, the entire thesis is vapor. 2. Are there real RFQs (request for quotes) for space-qualified AI accelerators from satellite primes? If the supply chain doesn't start moving, the demand isn't real. 3. Has any major cloud provider — AWS, Azure, GCP — signed a compute hosting agreement with Starlink? If not, they're either building their own or they don't believe orbital compute will compete on price.
I've seen this movie before. In 2017, I audited a token that promised to decentralize cloud storage. The whitepaper was beautiful. The code was an empty contract. The founders walked with $20 million.
The only difference here is the scale of the audience and the size of the check. But the same principle applies: trace the funds, ignore the noise. And when a report promises 33 trillion, look for the exit liquidity. It's always closer than the moon.