Hi Reader,
Below you'll find part three of Shruti's series on the AI IPO Paradox and SpaceX. We'd love to hear your feedback. Thanks for reading!
Warmly,
Sydney + Shruti
SpaceX's First Week. Here's What Actually Happened.
The pop, the Cursor acquisition, the OpenAI financials, and what this means for everything coming next
By Shruti Shah, General Partner, Symphonic Capital
Five days ago SpaceX went public at $135 a share.
By Tuesday it had hit $225. Briefly more valuable than Amazon. Briefly more valuable than Microsoft. The fourth most valuable company in the United States. Retail investors were borrowing money to buy it. Leveraged ETFs tied to SPCX launched and drew over a billion dollars in first-day trading volume. Elon Musk posted that SpaceX "might be able to reach approximately" $1 trillion in revenue by 2030.
By Thursday at close of trading it had pulled back to $185 — still up 37% from the IPO price, still the largest company by market cap to go public in history, still trading at a valuation that most fundamental analysis cannot justify.
I wrote two pieces before trading began about what was coming. I want to be honest about what week one actually tells us, and what it doesn't. Because underneath the headline numbers, the same structure I wrote about before the IPO has been running in plain sight all week.
The pop was real. The reason matters.
A 19% first-day pop is not unusual for a hyped IPO. What is unusual is the structure that created it.
SpaceX went public with roughly a 4% float. Only 4% of total shares were available to trade. When $250 billion in demand chases $75 billion in supply, and the stock opens to public trading with only a fraction of that supply available, the price mechanics have almost nothing to do with the underlying business. Thin supply plus massive demand plus index inclusion equals price action that looks like market validation but is not.
MSCI began adding SPCX to its indexes on June 13, the second day of trading. Nasdaq-100 inclusion is expected around July 7. Every passive index fund tracking those indexes is a forced buyer — automatic, price insensitive, regardless of what the AI segment losses look like. Your 401k may already own more of this than it did last week.
The pop was real. The stock went up. But understanding why it went up matters enormously for understanding what happens next. Because the same week the stock was rising on thin float and index mechanics, the machine was already running in a different direction.
The circular machine running in real time.
In March, three months before the IPO, Musk confirmed he had hired two of Cursor's product and engineering leads away from the company. Then in April, SpaceX signed a deal giving SpaceX the right to either acquire Cursor for $60 billion or pay $10 billion for their collaborative work together. SpaceX and Cursor have since been jointly training a new coding-focused AI model.
The sequence was: poach key employees, sign a work agreement and acquisition option, take the company public using that option as disclosed S-1 context, watch the stock pop 50%, then exercise the option using the inflated stock as currency. This was not a surprise post-IPO decision. It was a premeditated structure that used the IPO as the financing mechanism.
Cursor is a real business. It is likely making $3 to $4 billion in annualized revenue. But its market share of developers has declined from 41% in June 2025 to about 26% in May 2026, and it’s primarily concerned with selling into Enterprise. Anthropic now controls half the AI coding category. Grok 4.3, SpaceX's current coding model, placed 33rd on a coding benchmark, well below older models from OpenAI, Anthropic, and Google. SpaceX has not provided investors with details on Cursor's current customer list, momentum, or revenue trajectory. Public shareholders are being asked to accept 3.4% dilution for a business they cannot fully evaluate.
Thrive Capital, a venture fund that holds positions in both SpaceX and Cursor, has a combined stake that is now worth over $10 billion dollars. One firm on both sides of the transaction. The circular machine, visible in a single deal. And no shareholder vote was required. One person made this decision.
This same week, SpaceX added Roelof Botha to its board as an "independent" director and member of the audit committee. The whole point of an independent director on an audit committee is oversight, a voice that is not compromised by personal loyalty to the controlling shareholder. Botha is a longtime friend and ally of Musk's. Musk controls 82% of the votes. The controlled company exemption means SpaceX does not have to meet the same independence standards as other public companies. So the person added to oversee the financials on behalf of public shareholders is someone with a longstanding personal relationship with the man who controls the company. That is not illegal and it is disclosed. But it is worth understanding if you bought SPCX this week expecting standard public market governance protections.
The pop was mechanical. The Cursor acquisition was premeditated. The governance structure was designed. All of it is consistent with one thing: a structure built to advantage the people already inside.
What we do not know yet, and when we will find out
Nobody who built SpaceX has sold a single share yet. The IPO happened. The stock popped. But the insiders, the employees, early investors who did not sell on the secondary market, and executives, are locked up. They cannot sell yet. That is standard for IPOs: a lockup period prevents insiders from selling immediately after a company goes public, which protects the stock from being flooded with supply on day one.
SpaceX's lockup releases in structured waves between August and December. The first window opens around Q2 earnings, the first time SpaceX will report quarterly financials as a public company. At that point, 20% of eligible insider shares can be sold. Then another 7% every 35 days through October. Then the largest release, 28%, around Q3 earnings in November. By mid-December all restrictions will be lifted.
Here is why the timing matters. Each release window is timed to an earnings disclosure. The people closest to the business, the ones with the most information about whether the AI segment losses are narrowing, whether Cursor is gaining or losing ground against Anthropic, whether the Grok revenue trajectory justifies the valuation, get their first chance to sell at the exact moment the market is digesting new financial information for the first time. If insiders sell aggressively when that first window opens in August, that tells you something. If they hold, that tells you something different.
The same week the SpaceX lockup clock started, audited OpenAI financial documents were obtained by independent journalist Ed Zitron and reviewed by the Financial Times. The timing matters because OpenAI is one of the next companies heading to public markets — and the numbers tell the same story.
OpenAI generated $13.07 billion in revenue in 2025. R&D alone was $19.18 billion, including $10.59 billion paid directly to Microsoft for compute. Cost of revenue was $7.5 billion. Sales and marketing was $5.73 billion. Total operating loss was $20.92 billion. OpenAI loses roughly $1.60 for every dollar of revenue it generates.
That $10.59 billion paid to Microsoft is the circular machine documented in a single accounting line. OpenAI raises capital from investors at an $852 billion valuation. It pays that capital to Microsoft as compute costs. Microsoft's revenue rises. Microsoft holds a 27% stake in OpenAI. The circle is in the audited financials, just as it was in the SpaceX S-1.
SpaceX went public at $1.77 trillion with $4.9 billion in net losses and got a 37% pop. OpenAI is heading toward an IPO at $852 billion with a $20.92 billion operating loss and no path to profitability until 2030. When it files its S-1 publicly, these numbers will face scrutiny for the first time — analysts modeling them, short sellers stress-testing them, quarterly disclosure requirements following every subsequent quarter.
And Anthropic soon too. Google has invested over $40 billion into Anthropic. Anthropic has committed $200 billion to Google Cloud, a deal representing more than 40% of Google's entire revenue backlog. Anthropic now controls roughly half the AI coding market, the exact category SpaceX just paid $60 billion to enter through Cursor. When Anthropic files its S-1, the circular relationship with Google becomes public record in the same way OpenAI's relationship with Microsoft just became visible.
I am not predicting what the market does with any of this. Markets can stay disconnected from fundamentals for a long time. Tesla proved that for years. But the mechanism is now in motion and the documents are on the record.
The risk nobody is pricing
The circular machine is not just a story about individual companies being overvalued. It is a story about systemic concentration. A handful of private AI companies are the primary revenue source for a handful of public companies, Microsoft, Google, Oracle, CoreWeave — whose stock prices are embedded in indexes that ordinary Americans hold in their retirement accounts.
Oracle already showed us the shape of what happens when investors start pricing circular risk. It is down 50% from its September high, before a single S-1 was filed. Multiply that across Microsoft, Google, CoreWeave, and SoftBank, and you have a significant repricing event in large cap tech.
I am not predicting a market crash. The S&P 500 is broader than AI-adjacent tech and the mechanism I am describing is a repricing, not necessarily a collapse. But the concentration of risk in this corner of the market is being significantly underestimated. The AI narrative has been carrying valuations the fundamentals cannot yet support. At some point the narrative has to meet the numbers. When it does, the impact will not stay in a boardroom. It will show up in retirement accounts and in the communities already bearing the costs of the infrastructure that made this all possible.
I benefit from this. I still think about everyone else.
I want to be honest about something before I close.
I benefit from this. I work in venture capital. The AI IPO wave, the SpaceX pop, the OpenAI filing, the narrative re-legitimization of the asset class, makes our job at Symphonic a lot easier. LPs who have been waiting for liquidity will get some of it. New capital will enter the venture market. Funds like mine will find it easier to raise. Startups we back may find it easier to raise at better valuations because the AI narrative is stronger. I am not outside this machine looking in. I am inside it, and I am benefiting and will continue to benefit from it.
And I still think about what it means for everyone else.
The Memphis residents who got higher electricity bills, a federal Clean Air Act lawsuit, and jobs that largely did not materialize. The ordinary investors whose 401ks will own SPCX through index mechanics whether they chose to or not. The enterprise customers starting to balk at AI pricing and demanding a measurable return on investment. The founders building real products in overlooked markets who have been invisible to the capital flows I have been describing.
The AI boom will create real wealth. It already has. Musk's shares are worth over $1 trillion. Thrive Capital's combined SpaceX and Cursor stake is worth over $10 billion. The retail investors who bought at $135 are up 37% after one week.
And, The NAACP lawsuit against xAI's Southaven data center is still pending. COLOSSUS data center is still drawing power from the Memphis grid. The communities hosting this infrastructure are still the hosts.
The real test of who this IPO was built for starts in August when the first lockup window opens. I will be watching.
Shruti Shah is General Partner at Symphonic Capital, an early-stage venture fund investing in AI as infrastructure for underserved communities across health, wealth, and climate resilience. This is Part Three of an ongoing series:
Part One — The AI IPO Paradox
Part Two — Everyone Is Reading the SpaceX S-1. Here's What They're Missing