SpaceX AI: JPMorgan's $240 Target Hides a 20% Unlock and a 12.6B Quarterly Burn

The market is pricing a transformation, not a company. On August 25, JPMorgan issued a price target of $240 for SpaceX (SPCX), implying an 80% upside from the current trading price of $137.85. The stock has been struggling since its record-breaking IPO in June. The bank's thesis rests not on rocket launches, but on the AI division. Specifically, on the performance of Grok 4.6, the acquisition of AI coding tool Cursor, and a cross-selling strategy that has yet to prove itself. Bulls see a vertically integrated AI powerhouse. The data tells a different story. Follow the hash, not the hype. The context is a company caught between two identities. SpaceX is now a publicly-traded entity with a market cap near $1.7 trillion, yet its most valuable asset is arguably the Grok model developed by its AI division, a unit that lost $1.26 billion last quarter alone. JPMorgan's optimism is based on a single event: the release of Grok 4.6 on August 12, which added $500 billion to the company's market capitalization in a single day. This is the kind of volatility that makes analysts nervous, not confident. The bank also highlights the $4 billion ARR of Cursor, the AI coding assistant acquired earlier this year, noting that 75% of its revenue comes from enterprise clients. The logic is simple: these enterprise clients can be sold Grok licenses. The logic is also unverified. Let's dissect the core assumptions. JPMorgan claims Grok 4.6 sits on the "Pareto frontier" of AI models—no model is simultaneously cheaper and more intelligent. This is a powerful statement, but it is unsupported by any public benchmark data. No MMLU scores. No HumanEval results. No independent verification. The bank's claim is based on its own internal evaluation, which is not public. In my experience auditing AI protocols, a claim without a verifiable hash is a claim without substance. The second assumption is the data flywheel. Cursor's millions of coding sessions are being used for Grok's supplementary training, which the bank says has led to "significant improvements." This is a plausible strategy, but it raises questions about consent and data quality. Are Cursor users aware their proprietary code is being fed into a model that could one day compete with their own products? The legal and ethical implications are vast. The third pillar is the iteration cadence. JPMorgan expects a new model almost every month, with Grok 5 arriving before year-end. This is aggressive. OpenAI and Anthropic typically take six to twelve months between major releases. A monthly cycle suggests either incremental updates or a severe compromise on safety testing. Check the multisig. Always. The contrarian angle is that the bulls might be right about the fundamentals. Cursor's $4 billion ARR is real. The enterprise adoption rate is impressive. The potential for cross-selling Grok to these clients is logical. And the "Pareto frontier" claim, while unverified, could be true. If Grok 4.6 is genuinely superior on a cost-performance basis, it could disrupt the entire AI pricing model. This is what the market is betting on. But here's what the market is ignoring: the unlock. On September 9 and 10, nearly 370 million shares will be released, increasing the float by approximately 20%. This is a massive supply shock. With the stock already down from its highs, the pressure could be immense. The AI division is also burning through 86% of capital expenditures. This is not a profitable business; it is a highly speculative venture funded by public market capital. On-chain evidence never sleeps. The numbers are clear. JPMorgan's target implies a market cap of over $3 trillion, meaning the AI business must create $1.3 trillion in incremental value. That is an extraordinary bet on a model with no independent verification, a data strategy with unresolved consent issues, and an iteration schedule that sacrifices safety. The market is not pricing a company; it is pricing a narrative. The question is not whether Grok can code better than Copilot. The question is whether the narrative can survive contact with the balance sheet. The unlock is coming. The burn rate is unsustainable. And the AI division's revenue contribution remains undisclosed. When the hype fades, what remains is the ledger. And the ledger shows a company spending heavily to chase a future that has not yet been proven. Decentralized? No. Speculative? Yes. Verify. Don't just believe.