To Every Fund Manager Who Put SPCX In Their Model Portfolio This Week
To Every Fund Manager Who Put SPCX In Their Model Portfolio This Week
[Memo format. No sender. No recipient. You know who you are.]
Let's establish what actually happened.
SpaceX raised $75 billion on June 12 in the largest IPO in history. Priced at $135. The stock hit $225.64 on June 16 — a 67% gain in four trading sessions, on a company losing $4.28 billion per quarter, with a float of approximately 4% of total shares outstanding. Retail investors poured $369.8 million into SPCX in those first three sessions. Four times the pace of Nvidia at a comparable stage. The vibes were immaculate. The analysis was not.
On the same day the stock peaked — same day — SpaceX announced it was acquiring Cursor for $60 billion in all-stock consideration. Freshly printed public equity, still warm from the Nasdaq printer, handed over to a private company whose market share in AI coding tools had declined from 41% to 26% over the preceding twelve months. By Thursday the stock was at $184.98. The five-day volume-weighted average price sat at $181.71. The average buyer who chased SPCX into the open market was, by that point, approximately breaking even. The $620 billion incinerated in two days was not a rounding error. It was a verdict.
The Cursor deal is what you should be looking at. Not the price action. The deal.
Here is the version of events that SpaceX would like you to believe. xAI — now SpaceXAI, having merged with SpaceX in February — needed an enterprise AI product with real developer adoption. Cursor has over a million paying users. More than half the Fortune 500 uses it. It hit $2 billion in annualized recurring revenue faster than any B2B software company in history, including Slack, Zoom, and Snowflake. The Colossus supercomputer infrastructure gives Cursor the compute backbone to train models that can outcompete Anthropic's Claude Code and OpenAI's Codex. This is vertical integration. This is the complete closed-loop AI stack: compute, data, applications. Jensen Huang called Cursor one of the six core enterprises driving the digital workforce revolution. This is a $28 trillion TAM. The math works out.
That is the pitch deck version.
Now here is what the S-1 said, buried in the risk factors and financial statements nobody reads at the IPO roadshow. SpaceX reported a net loss of $4.9 billion for full-year 2025. The cause was xAI, which burned $6.36 billion in operating losses on $12.7 billion in capital expenditure. Starlink — the actual business that generates actual money — produced $4.4 billion in operating profit, which xAI's GPU ambitions simply ate. In Q1 2026, the consolidated net loss was $4.28 billion. One quarter. All eleven of xAI's original co-founders had left the company before it merged with SpaceX. Musk himself said in March that xAI "was not built right first time around." The entity powering $26 trillion of the $28.5 trillion TAM projection in the IPO prospectus had no original founding team and was described by its own controlling shareholder as a failed first attempt.
And into this, SpaceX dropped $60 billion of freshly issued public stock on a coding tool with declining market share.
Cursor's shrinkage from 41% to 26% market share happened for a specific reason that deserves to be said plainly: Anthropic launched Claude Code, it was better, and developers switched. Cursor had been model-agnostic by design — it ran Claude, GPT, and its own Composer models simultaneously, and that flexibility was precisely why developers chose it over locked ecosystems. The xAI acquisition destroys that value proposition. Once the deal closes in Q3, SpaceX has an obvious financial incentive to route every API call through Grok instead of Claude or GPT. The million developers who chose Cursor because it wasn't tied to any one model will notice. Some will leave. The ones who matter — the enterprise customers generating the revenue — will run a vendor risk assessment and update their procurement policies accordingly.
Cursor CEO Michael Truell described the acquisition as being "excited to partner with the SpaceX team to scale up Composer." Not quite the energy of someone who thinks his product just got supercharged.
There is a structural story underneath the headlines that is more interesting than whether SPCX bounces on Monday. SpaceX went public as an AI company. Not a rocket company. The launch business and Starlink are the revenue, but $26 trillion of the advertised TAM was attributed to AI — orbital data centers, enterprise applications, the full Musk vision. The rocket business is the collateral. The AI story is the valuation. At $2.37 trillion, SpaceX is the seventh-largest company in America by market cap and about the 200th largest by revenue. That gap has to close somehow. Cursor was supposed to be the mechanism.
The problem is that $60 billion for a product in market-share decline, paid in stock that was itself priced on the assumption of AI success not yet demonstrated, is not a gap-closing move. It is a commitment deepening. SpaceX is now more exposed to the xAI narrative, not less. If xAI and the Cursor integration deliver — competitive coding models, enterprise AI workflows, some fraction of that $22.7 trillion "enterprise applications" opportunity — the stock recovers and this all looks visionary in retrospect. If it doesn't, SpaceX paid sixty billion dollars it didn't have in cash for a product it needed for credibility rather than revenue, using currency it created by telling investors an AI story it hadn't yet proved.
The December 2026 lockup expiry is the appointment everyone is avoiding looking at. Right now, 4% of SpaceX shares are tradeable. The remaining 96% unlocks in December. The thin float that drove SPCX from $135 to $225 in three sessions is the same thin float that makes every sell order between now and November carry outsized price impact. When insiders can exit — employees, early investors, the pre-IPO holders who've been sitting on paper gains for years — the supply dynamic reverses with the same mechanical brutality it operated on the way up. Morningstar put fair value at $780 billion before any of this. That implies roughly 67% downside from the IPO valuation, and they hadn't even modeled the Cursor dilution yet.
The Nasdaq 100 rebalance around July 6, which some retail investors are treating as a floor, is worth roughly 1.6 to 2.2% of temporary price impact, per the more careful analysis. Not a floor. A speed bump.
Retail investors who received IPO allocations at $135 through Robinhood and Fidelity still have gains, though most received a handful of shares against requests for hundreds. The guy who asked for a thousand and got seventeen sold at $160 and exited. The guy who asked for seventy-five and received eleven bought four more in the open market and is now holding fifteen shares at a blended cost above $185. He thinks he's in a stock. He's in a story. And the story just got a lot more complicated.
None of this makes SpaceX a bad company. Starlink is a genuinely extraordinary business. The launch economics are real. The moat in heavy-lift orbital capability is arguably the strongest in the market. The case for owning SpaceX at $780 billion is coherent.
The case for owning SpaceX at $2.37 trillion rests entirely on whether a restructured AI division with no original founding team, led by a CEO simultaneously running Tesla, X, and a government efficiency project, can deliver on a $26 trillion AI opportunity — starting with a coding tool whose market share has been declining for twelve months — before the lockup expires and the 96% hits the tape.
That is a lot of belief to price in at $185.