The Ledger Doesn’t Lie: SpaceX’s Flat Bitcoin, $15.8B AI Push, and the $105B Unlock

Hype burns out; robustness remains in the ledger. That is not a slogan I repeat for social media; it is a filter I apply to every SEC filing that crosses my desk. On Aug. 4, SpaceX gave us the first quarterly report since its record IPO, and the headline numbers are easy to summarize: revenue up 92% year over year to $7.814 billion, adjusted EBITDA up 191% to $3.538 billion, net loss narrowing from $1.008 billion to $541 million. Wall Street applauded. Then it sold. The real story is not in the top line. The real story sits in the balance sheet, where billions of IPO proceeds have been transformed into money-market funds and government securities, while a position of 18,712 Bitcoin sits untouched. To understand what the market is pricing, and why the stock is under pressure, we have to audit both the portfolio and the narratives wrapped around it. I say audit because I have spent years reconciling what companies say with what the code and the contracts actually do. In blockchain, we trust the ledger. The first rule of reading corporate finance is exactly the same: trust the filing, not the headline. Let me set the stage for readers who have not followed the SpaceX-crypto saga from the beginning. The IPO was always unusual. It raised $85.675 billion in net proceeds, making it the largest record listing in history. It also made SpaceX one of the most visible public companies with direct Bitcoin exposure. When the S-1 landed in May, crypto investors immediately constructed two narratives. The first was that SpaceX shares would act as a Bitcoin proxy, giving equity investors indirect exposure to Elon Musk’s digital asset thesis. The second was that the IPO would inject fresh capital into X Payments and AI compute, turning SpaceX into a bridge between traditional finance and the crypto-native economy. Both narratives contained a kernel of truth, but the first quarterly report clarifies how small those kernels actually are. The company now holds cash, cash equivalents and marketable securities worth nearly $100 billion. Of that, $65.625 billion sits in money-market funds, more than triple the $21.339 billion reported at the end of 2025. Another $4.011 billion is in government securities classified as cash equivalents, and $6.487 billion is in marketable securities. Combined, money-market funds and government securities account for $76.123 billion. That is a balance sheet designed for stability, not for aggressive treasury engagement. Now look at the Bitcoin line. At June 30, SpaceX held 18,712 BTC with a cost basis of $661 million, exactly the same amount and cost basis reported at the end of 2025. The position remains $437 million above its cost, but the fair value fell from $1.637 billion to $1.098 billion during the first two quarters, producing a $539 million unrealized loss. As a percentage of cash and marketable securities, Bitcoin dropped from roughly 6.6% at year-end to about 1.1% after the IPO. That is not a treasury strategy change. It is a denominator effect. The company did not sell, according to the disclosure. It also did not buy. And while transaction-level activity is not disclosed, so purchases and sales could theoretically have offset each other, the simplest reading is that SpaceX simply let the position sit. I have seen this pattern before. In my DeFi Summer audit work, I spent hundreds of hours mapping voting centralization risks at Compound Finance, and the most important signal was often the absence of movement. A governance token that never changes hands can be either deeply committed or deeply indifferent. The same is true of a corporate Bitcoin balance. A company that parks $85 billion in money-market funds while leaving a small Bitcoin position frozen is not making a bull case. It is making a cold-wallet case. That is a message to the market, even if no press release says it. We audit the logic, for humans will always err. And the logic here is subtle. The Bitcoin position is now too small to matter financially, but too visible to ignore culturally. Selling it would generate a taxable event and a reputational storm, especially at a loss. Buying more would invite the same scrutiny that MicroStrategy attracts, and no CFO wants to explain a 6.6% allocation to a volatile asset during an AI capex supercycle. So the rational move is inaction. Inaction is often the most rational option in a world where both upside and downside are amplified by narrative. The market may see this as bullishly neutral or bearishly complacent, but I see it as evidence that the “Bitcoin treasury company” playbook is not universal. It is a niche strategy for companies with stable cash flows and a founder who wants to make a statement. SpaceX has chosen a different path. The second major story is the AI investment machine. SpaceX directed $15.828 billion into AI infrastructure during the second quarter, more than 21 times the $749 million spent a year earlier and roughly double the first quarter’s outlay. That represents 86% of the company’s total $18.369 billion capital expenditure. Across the first half, AI capital expenditure reached $23.551 billion, compared to just $3.316 billion in the prior-year period. This is not incremental spending. It is a transformation of the company’s cost structure. For context, Wall Street had estimated AI spending at $13.09 billion for the quarter, and SpaceX blew past that number while total capex came in slightly below the consensus of $18.58 billion. The message is unmistakable: almost every dollar of new investment is now flowing into compute. What is SpaceX getting for that money? The company reported AI revenue of $2.561 billion in the quarter, driven by computing agreements with Google, Anthropic, and internal products like Grok and X subscriptions. CFO Bret Johnsen said contracted compute deployments were producing payback periods of less than one year, faster than the payback on launch sites and satellite infrastructure. On its face, that is a compelling number. But I have learned to be careful with payback periods. In my economics training, a payback period tells you when an investment recovers its direct cost, not whether the project is economically profitable after R&D, depreciation, and working capital. Here, the AI segment still recorded a $1.257 billion operating loss, accompanied by $1.885 billion in depreciation and amortization and $2.178 billion in research and development spending. The segment is consuming cash, even as the contract pipeline grows. The company has signed $14.1 billion in cloud-services agreements and added another $6.7 billion in contracts after the quarter ended. That gives a total contracted backlog of $20.8 billion, an impressive number for any enterprise business. But compare that backlog to the quarterly investment run-rate. If SpaceX continues spending around $15.8 billion per quarter on AI, annual capex would be roughly $63 billion. With $20.8 billion in contracted revenue, the gap is large. The company is betting that the contract backlog will expand rapidly as demand for computing power keeps accelerating. That bet may be correct, but it is not yet proven. A CFO can say that a specific compute contract has a payback of less than a year, but that is a contract-level metric, not a segment-level one. The segment pays for the data center, the cooling, the research teams, and the undersea cables. Those costs are not allocated to any single contract, and they are the reason the segment loses money today. This is where the key tension of the SpaceX post-IPO era emerges. The market must somehow price a company with two very different financial personalities. On one side, SpaceX is a conservative treasury manager, holding government securities and money-market funds. On the other side, it is an aggressive AI infrastructure investor, spending money at a rate that dwarfs most hyper-scalers. Those two personalities do not necessarily conflict, but they create a denominator problem. When cash and marketable securities balloon to $100 billion, the Bitcoin position shrinks in relative importance. When AI capex absorbs 86% of total investment, the equity story becomes about execution, not about the balance sheet. The market wanted a Bitcoin proxy and an AI champion in one package. What it got, instead, is a company that is neither fish nor fowl, at least for now. The third story is the one that is currently moving the price: the lockup expiry and the crowded short trade. On Aug. 6, insiders become eligible to sell approximately 900 million shares worth about $105 billion at current prices. Tom Dunleavy, Head of Venture at Varys Capital, described this as one of the largest lockup expirations in market history. The shares already exist, but the expiration could sharply expand the float if employees and early investors choose to realize gains. And this is only the first stage. Another block becomes eligible after the third-quarter earnings report, with additional restrictions expiring on Dec. 8. By then, about 40% of SpaceX’s outstanding shares could be freely tradable. Elon Musk’s own stake remains locked until June 2027, but founder lockups do not prevent the rest of the cap table from moving. Short sellers have already positioned for further weakness. S3 Partners estimated that 95% of SPCX shares available to borrow were out on loan, while short interest reached 34% of the public float. That is an extraordinarily crowded trade. It also means there is very little lending supply left for new shorts. In a moment of positive news, this could produce a violent squeeze. But the market structure is more complicated than a simple equity short. Expectations of volatility have also spilled into crypto derivatives. CoinGlass data reviewed by CryptoSlate showed SPCX futures volume and open interest reaching their highest levels since the contracts launched. In the last 24 hours, trading volume climbed to about $6.85 billion, while open interest approached $720 million. Rising open interest does not tell you whether traders are predominantly bullish or bearish, because every futures contract contains both a long and a short position. What it tells you is that uncertainty is being monetized. People are paying for the right to be wrong, and the price of being wrong is going up. I seek the signal amidst the noise of the crowd. The signal here is not that short sellers hate SpaceX; the signal is that the options and futures markets have become a venue for two-sided hedging. AI investors want protection against a $100 billion unlock. Short sellers want protection against a sudden supply squeeze. Market makers are happy to sell that protection, but they too need to hedge. The result is a self-referential volatility spiral that can push the underlying stock in either direction. Observed short interest of 34% of the float is high, but it is not a vote on the merit of the company. It is a vote on the near-term supply-demand balance. And near-term supply-demand is dominated by known events: the Aug. 6 unlock, the earnings date, and the Dec. 8 restrictions. The common narrative is that the lockup expiration is the sole cause of the selloff. I think that is too convenient. The pressure is actually a combination of two waves meeting in the open water. The first wave is the supply overhang, which is real and measurable. The second wave is the AI spending ambiguity. SpaceX beat revenue expectations, but the market is not convinced that $15.8 billion per quarter of AI capex will translate into durable profits. The CFO’s payback period comment is seductive, but payback is not return. When a company’s equity valuation depends on future contracted revenue, a slight delay in contract signings can justify a large multiple compression. The downside is asymmetric. Every quarter where AI revenue does not accelerate will be met with the question: where is the compounding? The contrarian view is that SpaceX’s decision to keep Bitcoin flat and park cash in money-market funds is actually the most sophisticated move available. It is a form of optionality. The company is preserving dry powder at a time of extreme capital intensity. If AI demand softens, SpaceX has nearly $100 billion in liquid assets to sustain the crash. If demand accelerates, that money can be deployed in a matter of weeks. The Bitcoin position, meanwhile, is a tiny call option on future institutional adoption. It costs nothing to hold if you ignore the unrealized loss, and it provides a cultural bridge to the crypto community without jeopardizing the treasury. That is not cowardice. It is risk management. Faith in people is costly; faith in math is free. A $100 billion treasury backed by government securities is math. A $1 billion Bitcoin position is hope. The two can coexist, as long as you know which one is which. Open source is a covenant, not just a license. That line has guided my entire career, from analyzing Bitcoin whitepapers to auditing Compound’s governance. It also applies to corporate disclosures. A company that files transparently, reports unrealized losses, and discloses its Bitcoin cost basis is honoring a kind of covenant with the market. The market may punish the company for the lockup, or for the AI spending, but it should not punish it for clarity. The alternative is a company that hides its positions and plays games with fair value. That is the kind of company I spent many years warning readers about during the ICO boom. The ICO boom taught me a simple lesson: when the narrative is loud, the structure is weak. During that period, I reviewed over 40 whitepapers and found predatory tokenomics in about 30% of them. The common thread was an over-reliance on future buyers. The ecosystem would work, supposedly, when the next wave of users arrived. In SpaceX’s case, the future buyer is not a retail token holder; it is the cloud customer who signs a five-year compute contract. The model is less speculative, but the timing is still everything. If SpaceX can convert its AI backlog into recurring revenue before the market loses patience, then the current selloff will look like a gift. If not, the balance sheet will become a target. Analysts will say that SpaceX is wasting capital in a race it cannot win. That debate is the real source of the volatility, and no amount of short covering will resolve it. What should a crypto investor take away from this first quarterly report? I would point to three insights that are not in the press release. First, the Bitcoin position is now a piece of archaeology. It describes a past relationship, not a future strategy. Any trading strategy based on SpaceX as a Bitcoin proxy is going to be disappointed, because the ratio is simply too small. Second, the AI infrastructure spending is the hidden bull case for commodity electricity and data-center construction, but it is also the hidden risk for every smaller AI-mining company that cannot match SpaceX’s procurement power. The market is about to see a huge player enter the compute market, and that player has a $100 billion balance sheet. Third, the lockup expiration is not a single event. It is a phased release, and the market will remain in a state of uncertainty for at least the next four months. Traders should focus more on the December unlock date than on the August one. I have never believed that code is the only law that does not sleep. Actually, I have written those words more than once, and they are true, but they are incomplete. Code does not sleep, but neither do balance sheets. The same ledger that proves a transaction is the ledger that reveals a company’s true priorities. SpaceX’s ledger shows a company that is simultaneously cautious and aggressive. It is cautious about Bitcoin, careful to maintain an unchanged position in a historic bull market. It is aggressive about AI, willing to spend $15.8 billion in a single quarter. The two attitudes are not contradictory. They are complementary. The caution funds the aggression. The Bitcoin position is walking money; the AI infrastructure is doing money. If the AI bet pays off, the Bitcoin position will be a historical footnote, not the main event. The selloff after an earnings beat is never a pleasant sight. But in the sideways, uncertain market we are currently living through, the reaction is almost a mathematical necessity. A stock that doubles in a year can easily give back 20% when liquidity suddenly expands and the short interest reaches 34%. The facts of the business have not changed in the last 24 hours. The company still has almost $100 billion in cash, a growing AI backlog, and a Bitcoin position that has survived a $539 million unrealized loss without a single execution. What changed is the available supply of shares, and supply changes matter in the short term. They do not change the long-term path of the company, unless management loses its nerve. Let me end with a forward-looking observation rather than a summary. The next two quarters will tell us far more about SpaceX than this one did. Watch the quarterly AI revenue line. Watch the contracted backlog. Watch whether the Bitcoin position remains frozen at 18,712 BTC. If the position remains frozen, the market will understand that SpaceX has moved on from the crypto treasury narrative. If it changes, that will be the only signal that matters. The stock price may wobble, the futures market may spike, and the short sellers may celebrate, but the ledger will stay open. And the ledger does not lie. The only question is whether we are willing to read it slowly enough to hear what it says. In the end, I keep coming back to the first principle I learned as an economist: incentives create behavior. SpaceX’s incentives now point in two directions. The AI division wants to devour capital, and the treasury wants to preserve it. The Bitcoin position sits in the middle, too small to anchor the balance sheet but too iconic to abandon. That is not a flaw. It is a hedge against its own history. Hype burns out; robustness remains in the ledger. The ledger of SpaceX now shows a company that is no longer a startup and not yet a hyper-scaler. It is a transition in progress, and the market hates transitions. But transitions are where the signal is richest, if you are willing to ignore the noise.