Hi Reader,
Shruti wrote a new piece about the Space X IPO. We'd love your shares, thoughts, and feedback.
Best,
Sydney + Shruti
Everyone Is Reading the SpaceX S-1. Here's What They're Missing.
By Shruti Shah, General Partner, Symphonic Capital
I grew up in Memphis.
The neighborhoods near the COLOSSUS data center, the massive AI computing facility that SpaceX built there and named in its filing to go public, are neighborhoods I know. The people whose electricity bills went up when that facility came online are people I grew up with. The community advocates raising alarms about what it's doing to the aquifer that supplies drinking water to millions of people across the mid-South are people I respect.
None of that is in the 277 pages SpaceX filed with the Securities and Exchange Commission a few weeks ago.
I'm a venture investor. My fund backs early-stage AI startups in communities that traditional venture capital has largely overlooked — the same communities now hosting the infrastructure of the AI boom without capturing its returns. I watch that gap every day. The data centers arrive. The power bills go up. The investment dollars don't follow.
That's the lens I brought to reading the SpaceX S-1. And it's why certain things in that filing stuck out to me that aren't getting much attention elsewhere.
This isn't the SpaceX you think it is
When SpaceX starts trading on June 12 under the ticker SPCX, most people will think they're buying a rocket company, and maybe Starlink too, the satellite internet service that has crossed 10 million subscribers and changed how people in rural and remote areas can access the internet.
What they're actually buying is four companies bundled into one.
SpaceX acquired xAI — Elon Musk's artificial intelligence company — in February of this year, less than four months before filing for the IPO. xAI had previously acquired X, the company that owns Twitter, in March 2025. The financial statements filed with the SEC have been rewritten to include all of it.
So SPCX is: rockets, Starlink, xAI and its Grok AI model, and Twitter. One ticker. Four companies. The filing is explicit about this — the "AI segment" is defined as xAI, Grok, AI computing infrastructure, and X. It is one of the three main business lines of the company going public.
The part that's making money and the part that isn't
Here's what the filing actually shows. SpaceX is four businesses in one ticker. Only one of them is unambiguously profitable.
Starlink (Connectivity): $11.4 billion in revenue in 2025, growing nearly 50% year over year. The only clearly profitable segment. The engine that carries everything else.
Rockets and launches (Space segment): $4.1 billion in revenue in 2025. The existing Falcon 9 launch business is operationally strong — SpaceX controls more than 80% of all mass launched to orbit globally, at $67 million per launch with a 99%+ mission success rate. The segment actually turned a small $21 million profit in 2024 — before Starship R&D investment pushed it to a $657 million operating loss in 2025. Whether you count that as a loss or an investment in a future asset depends on your accounting preference. The underlying launch business works, but the Starship bet is expensive.
xAI and Grok (AI compute and frontier models): Part of the AI segment. Combined with X, the segment generated $3.2 billion in revenue in 2025 but lost $6.4 billion from operations. In just the first three months of 2026, that loss was another $2.5 billion.
X / Twitter: Bundled into the AI segment in the filing. Also losing money. Acquired by xAI in March 2025, four months before xAI itself was merged into SpaceX.
Starlink is the engine carrying everything else. The AI segment alone, xAI, Grok, and Twitter, is what pushed the company into the red. The company as a whole lost $4.94 billion net in 2025, after making $791 million in profit the year before, when it was SpaceX alone, before the xAI merger.
The price tag doesn't add up
SpaceX is targeting a valuation of $1.75 trillion. To put that in perspective: that would make it one of the most valuable publicly traded companies in the world on its first day of trading.
On standard accounting measures, Starlink is the only unambiguously profitable segment. At $1.75 trillion, investors are paying 153 times Starlink's annual revenue.
Here's why that matters. When Google went public in 2004, its valuation was roughly 8 times its annual revenue. Even during the height of the dot-com bubble — when investors were notoriously paying too much for technology companies — most companies didn't get valued at more than 20 to 30 times revenue.
153 times Starlink's revenue is a bet on a very specific future: that the AI losses turn into profits, that Starlink keeps growing, that the launch business eventually pays off, and that the political environment stays favorable to a company whose CEO is simultaneously running Tesla and SpaceX.
All of that might happen. But those are assumptions about the future embedded in today's price.
The Tesla argument — and why this is different
The most obvious counterargument to everything above is Tesla.
Elon Musk ran Tesla at a loss for nearly a decade after it went public in 2010 at a valuation of $1.7 billion. The company lost money consistently through at least 2015. Most analysts were skeptical. Most early investors who sold early left extraordinary returns on the table. A $10,000 investment at Tesla's IPO is worth approximately $3.8 million today.
So the bull case for SPCX is straightforward: bet on the man, not the current financials. If you believed in Musk in 2010 and held, you were right. Why wouldn't the same logic apply now?
It's a fair argument. But there are three meaningful differences worth holding in mind.
First, the scale. Tesla went public at $1.7 billion. SpaceX is going public at $1.75 trillion — roughly a thousand times larger. Tesla had room to grow into almost any valuation from that starting point. SpaceX has to justify a number that prices in extraordinary execution across rockets, satellite internet, AI, and social media simultaneously, from day one.
Second, the structure of the losses. Tesla was a single company losing money in service of building one product: the car. Every dollar lost was going toward the thing that eventually made money. SpaceX is more complicated: Starlink is clearly profitable. The rocket business is operationally strong on existing Falcon 9 but posts an accounting loss of $657 million because of Starship development investment. The AI segment — xAI, Grok, and Twitter — lost $6.4 billion. So Starlink is carrying the AI losses, with the rocket business somewhere in between depending on how you account for Starship. The financial foundation is narrower than it looks from the outside.
Third, the governance. In 2010, Tesla shareholders had real votes. They couldn't easily override Musk, but they had meaningful say in governance. At SpaceX, public shareholders have no practical ability to influence any decision. Betting on Elon as a manager is one thing. Betting on Elon with zero governance recourse — where even a decision to merge an entirely different company into SpaceX four months before the IPO faces no shareholder check — is a different kind of bet.
The Tesla comparison is the strongest argument for buying SPCX and could be worth taking seriously. So is the difference in what you're actually buying.
The Amazon argument — and why it's more complicated than it looks
There is a second compelling counterargument worth addressing: Amazon.
For years Amazon operated at razor-thin margins or losses while AWS quietly became the profit engine that now subsidizes everything else. Starlink looks remarkably similar, a high-margin infrastructure business generating billions in profit while carrying a portfolio of other bets. The parallel is clean and the people making the bull case for SPCX are right to reach for it.
The difference is what those other bets actually are.
Amazon's retail losses were a deliberate reinvestment in a business with clear unit economics. Every dollar spent on logistics and fulfillment was building toward a profitable scale in a market Amazon was creating and controlling. The path from loss to profit was legible — more customers, more volume, more leverage on fixed costs.
SpaceX's AI segment is burning more than $6 billion a year competing directly against OpenAI, Anthropic, and Google. All of them are also burning billions. All of them have larger installed bases in AI specifically. Grok is a real product with real users — but it is entering a war it didn't start, in a market defined by competitors with deeper AI-specific expertise and longer runways in the domain.
Amazon was investing in a business it controlled. SpaceX is entering a competition it didn't originate, against opponents who are better resourced in that specific fight.
That doesn't mean Grok fails. It means the Amazon comparison flatters the AI bet more than the competitive dynamics fully justify. Starlink is genuinely AWS-like. The question is whether xAI is retail, a business that eventually works at scale, or something else.
You don't get a vote
This is the part that's clearest in the filing and least discussed in the coverage.
SpaceX is going public with two different types of shares. The shares being sold to the public, the ones ordinary investors can buy, each carry one vote. The shares that Elon Musk holds carry ten votes each. And his class of shares gets to elect a majority of the company's board of directors, no matter what the public shareholders want.
The filing is explicit: SpaceX will qualify as a "controlled company" under stock exchange rules, which means it is exempt from certain requirements that other public companies have to follow around independent oversight and governance.
What this means in practice: the decision to merge xAI into SpaceX four months before this IPO, creating the losses now baked into the financials, was Musk's to make alone. Future decisions about the company's direction will be his too. Public shareholders, whether they own 100 shares or 100 million shares, have no meaningful say in any of it.
When you buy SPCX, you are buying a financial stake in a company that one person controls entirely. It's disclosed in the filing but worth knowing before June 12.
Your retirement account is buying this whether you want it to or not
Here is the part of this story that goes beyond people who actively choose to invest in the IPO.
SpaceX is seeking early entry into the Nasdaq-100 — one of the major stock market indexes that tracks large technology companies. When a company joins a major index, every fund that tracks that index has to automatically buy shares in it. This includes the index funds that sit inside millions of Americans' 401ks, pensions, and college savings accounts.
This buying is automatic. It doesn't involve anyone deciding whether SpaceX is a good investment. It happens because of how index funds work. They're designed to track an index, not to pick stocks.
This means that if SpaceX is admitted to the Nasdaq-100, a significant portion of American retirement savings will automatically gain exposure to whatever happens to the stock next — including the risk that a company valued at 153 times the revenue of its only profitable business eventually gets repriced closer to what the underlying businesses actually support.
The people affected by this won't necessarily know it happened. They'll just notice, eventually, what it did to their balance.
What this means for Memphis — and everywhere else like it
Let me come back to where I started.
The SpaceX filing cites a report with this title: "The 175 GW Crisis: America's Power Grid Cannot Keep Up with AI Data Centers." This is a real concern, documented in the S-1, about whether the electrical grid can support the scale of AI computing infrastructure being built right now.
In Memphis, that isn't abstract. COLOSSUS consumes roughly a gigawatt of power. Memphis Light Gas and Water, the utility that serves the city, has to absorb that load alongside the residents and businesses it already serves. The people most affected are often lower-income households who have less ability to absorb higher electricity costs.
Data centers of this scale also consume significant amounts of water for cooling. The Memphis Sand aquifer, one of the largest freshwater aquifers in North America, has been a source of concern among local environmental advocates for months. These concerns are not hypothetical. They are ongoing and documented. They are not in the S-1.
Memphis was promised jobs when these facilities were announced. In practice, gigawatt-scale AI computing infrastructure employs relatively few people, and the roles it does create tend to be highly specialized, not the kind of broad workforce opportunities that community leaders imagined when they negotiated the tax incentives that brought the facilities there.
This is the pattern I see from where I sit. The communities that have been overlooked by mainstream investment don't just miss the upside. They host the infrastructure, pay the utility bills, and carry the environmental costs, while the financial returns flow to shareholders elsewhere.
Memphis is not a beneficiary of this buildout. It is the host.
What this all adds up to
Three things I'm watching on SpaceX specifically:
First day trading. The shares available to the public on June 12 will be a small slice of the total company, roughly 4 percent based on the reported raise. Thin supply with high demand can push a stock up fast on day one. That early price action doesn't necessarily reflect long-term value.
The AI revenue question. The losses in the AI segment are documented. What isn't yet clear is whether Grok, the Grok API, and enterprise AI products can generate enough revenue to justify the AI segment's share of the $1.75 trillion valuation. That's the question the next several quarters of earnings will answer.
Index inclusion timing. If SpaceX is fast-tracked into a major index within weeks of trading, the automatic buying from index funds will create additional upward pressure on the stock regardless of the underlying financials. That automatic buying and the fundamental business are two different things.
But those are questions about SpaceX specifically. The bigger question, the one I've been sitting with since I started reading this filing, is about the shape of this economy.
The AI boom is producing extraordinary concentrations of capital, infrastructure, and power. The data centers are going into communities like Memphis — communities that offered land, power, and tax incentives and were told they would benefit. The index mechanics are routing ordinary Americans' retirement savings into these companies without asking. The circular financing that I've written about elsewhere is propping up public market valuations that may not survive contact with disclosed financials.
Meanwhile, the companies quietly building real products for real people in those same overlooked communities — companies earning trust rather than assuming it, building data assets through genuine service rather than scraping them — are largely invisible to the capital flows I've just described. They don't show up in the S-1. They don't get the index inclusion. They don't benefit from the circular machine.
I believe those companies are where the durable value in AI actually lives. Not because of a thesis on a slide deck, because the data they've built can't be replicated, the customers they've earned can't be poached by a better-funded competitor with more Azure credits, and the communities they serve have been waiting for this kind of investment for a long time.
The SpaceX S-1 is 277 pages. The impact on the Memphis community isn’t in any them. That gap, between where the costs of the AI boom land and where the returns go, is the most important economic story of this moment. And it is hiding in plain sight.
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. Her piece on the broader AI IPO wave — The AI IPO Paradox — is linked here.