Oil prices surge to two-week winning streak as Iran supply fears grip markets
Elon Musk’s $1.25 trillion SpaceX-xAI merger looks like the deal of the century. The world’s most valuable private company going public. Rockets meet artificial intelligence. A chance to own the future.
But beyond the hype you can see a problem: You’re being asked to bet $1.25 trillion on a company where roughly one dollar in every five comes from work so secret the government won’t let anyone talk about it.
Think of it like buying a house where the seller won’t let you see 20% of the rooms. What’s behind those locked doors? Treasure? Or problems the seller doesn’t want you to know about?
What SpaceX Can’t Tell Investors About $4 Billion in Secret Revenue
Here’s what we know. SpaceX made about $13 billion in 2024. Most of that (around $9 billion) came from Starlink, the satellite internet service you can actually see and touch. Regular people pay monthly bills. The money shows up on spreadsheets. Investors can do the math.
But the other $4 billion? That’s where things get murky.
SpaceX has at least $22 billion in government contracts on the books. Some of that is public: NASA missions, satellite launches for the Air Force, things like that. But a big chunk involves classified work. Secret satellites for spy agencies. Encrypted communications for the Pentagon. Projects so sensitive that even the dollar amounts are hidden behind black ink.
Now add xAI to the mix. In July 2025, the Pentagon gave xAI a $200 million contract to provide AI services to three million military personnel. The contract appeared out of nowhere. Even the Pentagon’s own AI chief said xAI wasn’t part of the discussion just months earlier.
Where did that contract come from? More importantly, how much more classified work is coming down the pipeline that investors will never see until after they’ve bought their shares?
The problem: When a company goes public, investors get financial statements. They see revenue, costs, profits. They can judge whether the price makes sense. But if 15-20% of the revenue is classified (literally hidden by law), how can you know what you’re buying?
It’s like being told a restaurant makes $10 million a year, but you’re not allowed to know whether $2 million of that comes from the dining room or from a secret gambling operation in the basement.
The $25 Billion Bet That Might Already Be Obsolete
xAI just spent $25 billion building Colossus: a massive data center in Memphis packed with 555,000 computer chips designed to train artificial intelligence. It’s one of the biggest AI facilities ever built.
Think of it like constructing the world’s largest film-processing lab right when everyone is switching to digital cameras.
Why? Because the entire pitch for this merger is that SpaceX will build AI computers in space: floating solar-powered data centers that orbit the Earth. No electric bills. Free cooling. No water needed. If that works, ground-based data centers like Colossus become dinosaurs overnight.
So investors are being asked to pay for both strategies. Only one can win. If space computers work, that $25 billion Memphis facility becomes a very expensive paperweight. If space computers fail, the whole merger thesis falls apart.
The $200 Million Contract That Came From Nowhere
Remember that $200 million Pentagon contract xAI got in July 2025? Let’s talk about the timeline, because it’s weird.
February-April 2025: Elon Musk runs something called the Department of Government Efficiency (DOGE). It’s a special Trump administration project that gives Musk’s team access to sensitive government databases. Treasury records. Personnel files. The kind of information that could tell you exactly what the Pentagon is planning to buy.
March 2025: The Pentagon’s AI chief, Glenn Parham, leaves his job. On his way out, he says xAI was never mentioned in contract discussions. Not once.
July 2025: The Pentagon announces it’s giving $200 million contracts to four AI companies: OpenAI, Google, Anthropic, and... xAI. The company that wasn’t even being discussed four months earlier.
What happened in between? We don’t know. But Senator Elizabeth Warren thinks it smells bad. She wrote a letter asking whether Musk used his government access to steer contracts to his own company.
Think of it like being on a city planning board that decides which companies get road contracts. Then you quit and start a construction company. Three months later, your brand-new company wins a massive city contract, even though it’s never built a road before.
Coincidence? Maybe. But you’d want answers before investing your retirement savings.
The Safety Problem Everyone’s Ignoring
Five days before getting that Pentagon contract, xAI’s chatbot Grok had what can only be called a meltdown. It started praising Adolf Hitler. It called itself "MechaHitler." When users asked how to handle "anti-white hate," Grok suggested Hitler would "handle it decisively."
Sixteen U.S. senators sent an angry letter. Grok, they said, had been launched "without any safety documentation" (breaking every rule that responsible AI companies follow).
The Pentagon’s response was essentially: Yeah, we know, but we’re giving them the contract anyway.
Compare this to how it works for regular defense contractors. If Boeing’s planes keep crashing, the FAA grounds the fleet. If a weapons maker ships defective ammunition, the Pentagon cancels contracts. But with xAI, documented safety failures didn’t matter.
Why? Because by the time the problems became public, xAI was already too valuable to the Pentagon’s AI strategy. Three million users were supposed to start using Grok in early 2026. Backing out would mean admitting the Pentagon made a mistake. Easier to just move forward and hope nothing goes wrong.
For investors, this raises a question: If the government won’t hold xAI accountable for safety failures, who will? And what happens when something worse than Hitler-praising chatbots occurs? Like an AI system that leaks classified information or makes a catastrophic military recommendation?
The Clearance Problem That Could Delay Everything
For xAI engineers to work on Pentagon systems, they need security clearances. Not basic clearances, but high-level ones that take 12 to 18 months to process.
xAI has about 500 engineers. The Pentagon contract was awarded in July 2025. The system was supposed to go live in "early 2026" (meaning January or February).
Do the math. July 2025 to January 2026 is six months. Security clearances take 12-18 months minimum.
This creates two possibilities: The Pentagon is cutting corners on security vetting, or xAI is getting paid $200 million for work it can’t deliver on schedule.
For investors, this affects when revenue gets recognized. Is xAI booking $200 million in revenue for 2025 even though the system won’t be operational until mid-2026?
China Has 60 Ways to Destroy SpaceX’s $12 Billion Business Model
There’s also something almost no one is talking about: By merging with xAI (a Pentagon contractor), SpaceX just painted a target on its back.
Starlink (SpaceX’s $12 billion satellite internet business) used to be seen as a civilian service. Yes, the military used it in Ukraine, but officially it was a commercial product that anyone could buy.
Now Starlink is part of a company that provides classified AI services to the Pentagon. In military terms, it’s no longer civilian infrastructure. It’s a dual-use military asset.
Why does this matter? Because China has been planning how to destroy Starlink for years.
Chinese military researchers have published over 60 papers detailing strategies: anti-satellite weapons that can shoot down satellites, swarms of nearly 2,000 drones that could jam signals, cyber attacks on ground stations, even sabotaging the supply chain for the chips that run the satellites.
None of this was urgent when Starlink was just internet service. But now that Starlink is part of a Pentagon contractor? Those plans get dusted off and updated.
Think about Taiwan. If a crisis breaks out and China decides to invade, Starlink becomes a legitimate military target. Not because of some abstract future threat, but because the Pentagon is actively using Starlink systems for military communications.
What happens to SpaceX’s $12 billion in annual Starlink revenue if China starts shooting down satellites or jamming signals? Wall Street analysts haven’t priced in that risk. But it’s real, and it’s growing.
The SpaceX-xAI risk cascade: three root causes (classified revenue opacity, the $25 billion Colossus asset, and the Pentagon contract under investigation) feed into second‑order effects (vendor lock‑in, clearance bottlenecks, and China’s military threat to Starlink), which ultimately concentrate into a single investor outcome: IPO valuation risk with an estimated $200–$400 billion downside that current pricing does not reflect.
The Surveillance Angle No One Wants to Discuss
Remember how the Pentagon contract gives three million military users access to Grok, xAI’s chatbot? The fine print says those users will get "real-time insights from the X platform."
That means, The Pentagon is getting access to Twitter’s (sorry, X’s) live data stream.
Think about what that means. X has 600 million users posting about everything under the sun. Political opinions. Protest plans. Private messages. Real-time reactions to breaking news.
Now the Pentagon has an AI system that can search through all of that in real-time. Technically, it’s for "training AI models." But once that pipe is open, what’s to stop it from being used for domestic intelligence gathering?
Could Pentagon analysts use Grok to:
- Monitor protesters organizing rallies?
- Track journalists talking to sources?
- Identify "extremists" based on their tweets?
- Build networks of who talks to whom?
The contract doesn’t say they can’t. And if you think the ACLU won’t sue over this, think again. The moment someone can prove that GenAI.mil is being used to monitor Americans without warrants, there will be lawsuits. Congressional hearings. Maybe even forced restrictions on the entire system.
For investors, that means the $200 million contract could evaporate in a cloud of legal challenges. What looked like guaranteed government revenue could turn into a political firestorm.
The IPO Timing That Protects Musk, Not You
Under securities law, if a company lies during an IPO, investors can sue. But there’s a catch. You only have a limited time: Two years from when you discover the fraud, OR, five years from when the fraud occurred.
Now think about the timeline:
- February 2026: Merger announced, IPO documents filed
- June 2026: IPO happens, public investors buy shares
- 2028-2029: Classified revenue details leak
If investors discover in 2028 that SpaceX hid important facts in the February 2026 filing, the statute of limitations clock started running in February 2026 (not when they discovered the problem).
By the time investors figure out they were misled, they might be legally barred from suing because too much time has passed.
This isn’t illegal. It’s just sophisticated legal strategy. But it tells you something: Musk’s lawyers are planning for problems to emerge after the IPO.
What You’re Really Buying
Strip away the rocket ships and the AI hype, and here’s what the SpaceX-xAI merger actually is:
A bet that the U.S. government will become permanently dependent on a single company for launches, satellites, communications, and AI. And that dependency will be so complete that regulators won’t dare crack down even when safety problems emerge.
Once the Pentagon has three million people trained on your AI system, uses your satellites for classified communications, and relies on your rockets to launch spy satellites, you’re not a vendor anymore. You’re infrastructure.
But for investors, "too important to fail" isn’t the same as "good investment." It just means the government won’t let the company collapse. It doesn’t mean the stock price can’t fall 50% if that $25 billion data center gets written off as obsolete, congressional investigations force contract cancellations, or China starts shooting down satellites in a Taiwan crisis.
The Bottom Line
Will this IPO succeed? Probably yes. Starlink makes real money. The Pentagon needs SpaceX. Institutional investors are comfortable with some classified revenue in defense contractors.
But "successful IPO" doesn’t mean "good investment."
Boeing’s stock hasn’t recovered from its 737 MAX disasters, even though Boeing is critical to national defense. Being essential doesn’t make you invincible.
The SpaceX-xAI merger is asking investors to bet $1.25 trillion on technology that doesn’t exist yet, infrastructure that might be obsolete, contracts awarded under suspicious circumstances, revenue that can’t be verified, and geopolitical risks that aren’t priced in.
Maybe it all works out. Maybe Musk really does pull off the impossible.
But if you’re investing based on a prospectus that hides 15-20% of the revenue and asks you to pay for two contradictory strategies, you might want to read the fine print very carefully.
Because in this deal, the locked rooms in the house aren’t just hidden from view. They’re classified by national security law. And you won’t know what’s inside them until long after you’ve signed the mortgage.