How the Pie Gets Cut: What Really Happens When a Company Goes Public

On the morning of June 12, 2026, Space Exploration Technologies Corp. sold 555,555,555 shares of stock to the public at one hundred thirty-five dollars apiece. The company collected roughly $74.4 billion. It was the largest initial public offering in the history of the world.

Buried on page 71 of the prospectus, in a table most buyers never opened, was this:

Read those two rows against each other. Everyone who owned SpaceX before that morning — Musk, the venture funds, the sovereign wealth funds, the employees who took stock instead of salary — had put roughly $81 billion into the company across twenty-four years, and owned 95.8 percent of it. The public put in $75 billion in a single morning and received 4.2 percent.

Both numbers are correct. Neither is a scandal. But the distance between them is the thing we want to explain in this letter, because that distance is where nearly every misunderstanding about IPOs lives.

We are going to build the machinery to understand that table from the ground up. We will start with a pizza shop, pass through Shark Tank, and end inside the largest offering ever conducted — covering along the way how dilution actually works, what "non-dilutive" money really is, who the underwriters are and what they get paid, what a lock-up schedule looks like from the inside, how a mega-IPO reaches into your index fund without asking, how SPACs, spin-offs, and ADRs differ from the ordinary route, and how to read the stock-based compensation disclosures well enough to forecast the dilution that is still coming.

Everything we cite from SpaceX is a public document you can pull up yourself in about ninety seconds, and teaching you where to look is the real purpose of this piece. The next time a company you actually care about files to go public, we want you to be able to do this without us.

Start small: a pizza shop with one hundred shares

Before the billions, a shop. Maria owns a pizzeria. When she incorporated it, the company issued one hundred shares of stock, and she owns all one hundred. The number one hundred is arbitrary — it could have been ten or ten million — what matters is only what fraction of the total you hold:

Maria owns 100 of 100 shares, so she owns 100 percent. "Shares outstanding" simply means every share the company has issued that somebody holds. Keep an eye on that denominator; this entire letter is about things that change it.

Now the shop's books. It owns an oven worth $30,000, furniture and fixtures worth $15,000, and $5,000 of cash: $50,000 of assets. It also owes the bank $20,000 on the loan that bought the oven. What the owners actually have is what would be left if the assets were sold and the debts paid:

Here, $50,000 − $20,000 = $30,000. That figure — also called book value, or net worth — is the accountant's answer to "what is the owners' stake worth on paper." Spread it across the shares and you get the per-share version:

One more idea and the toolbox is complete. Suppose a regular customer offers Maria $500 for one of her shares. Why would anyone pay $500 for $300 of book value? Because the shop earns $20,000 a year, and a share is a claim on all the future years, not on the furniture. The price a buyer will actually pay is the market value, and it usually differs from book value — for a good business, it is usually far above it. Multiply the price by the share count and you get the market's verdict on the whole company:

At $500 a share, Maria's shop has a market cap of $50,000 against $30,000 of book value. Hold on to the distinction between those two numbers. When SpaceX priced at $135.00, its market capitalization was 13,075,865,175 shares × $135 ≈ $1.77 trillion — against tangible book value, as we will see, of about $7.85 a share. The gap between what a share costs and what it is backed by on paper is one of the central facts of this subject, and the prospectus is required to disclose it.

Three ways to raise money — and what "non-dilutive" really means

Maria wants $10,000 to buy a second oven. Watch carefully what happens to the share count in each of the three ways she can get it, because this is the whole game.

Door one: sell new shares. An investor offers Maria $10,000 for 20 percent of the business. Here is what that sentence means arithmetically. The investor wants to end up owning 20 percent, so the company manufactures new shares — it does not take Maria's — until the new shares are 20 percent of the post-deal total:

The company issues 25 new shares to the investor. Maria still owns exactly 100 shares — not one was taken from her — but she now owns 100 of 125, which is 80 percent. That is dilution: her share count is unchanged, her percentage fell, because the denominator grew.

Notice where the money went: into the company. The shop now has $10,000 it did not have before. This is called a primary issuance, and it is the defining feature of most IPOs, including SpaceX's. And notice something subtler. The shop's equity is now $30,000 + $10,000 = $40,000, across 125 shares — book value per share rose from $300 to $320, because the new shares were sold for $400 apiece, above the old $300 book value. Selling new shares above book value raises book value per share for everyone. File that away; it is exactly what happened to SpaceX's existing holders at $135, on a much grander scale.

Was Maria harmed? She went from 100 percent of a business with no spare cash to 80 percent of a business with $10,000 in the till. If that $10,000 earns more inside the shop than the 20 percent she surrendered was worth, she wins. If not, she doesn't. That is the entire test, and it applies identically to a pizza oven and to a $75 billion offering. Dilution is not inherently bad. Dilution without a corresponding increase in value is bad. The number to watch is never the share count by itself. It is value per share.

Door one, variant: the founder sells her own shares. Suppose instead the investor buys 20 of Maria's existing 100 shares for $8,000. No new shares exist. The investor owns 20 of 100 — the same 20 percent — but the company received nothing. The $8,000 went into Maria's pocket. This is a secondary sale, and nobody was diluted, because the denominator never moved. On Shark Tank you occasionally see a founder "take money off the table" this way. In an IPO prospectus you see it on the cover page as shares offered by "selling stockholders." This primary-versus-secondary distinction is the single most useful thing in this letter: primary shares fund the company and dilute everyone; secondary shares fund the seller and dilute no one. Every offering is some mix of the two, and the cover page tells you which within about four seconds.

Doors two and three: the "non-dilutive" deals. A reader heard a Shark Tank deal described as non-dilutive and asked how that works, since giving an investor a piece of the business without diluting sounds like a contradiction. It mostly is — and untangling it teaches something important.

Strictly speaking, the only equity that changes hands without dilution is equity that already exists — a secondary sale like Maria's, where the founder pays personally by giving up her own slices. When a shark says "non-dilutive," though, they almost always mean a deal that is not equity at all. The classic is the royalty: "I'll give you $10,000 today, and you pay me fifty cents on every pizza until I've collected $25,000." No shares are created. Maria still owns 100 percent. The cap table — the list of who owns what — is untouched, which is why the structure gets the friendly name.

Now do the arithmetic the television cameras skip. The shop sells 8,000 pizzas a year, so the royalty costs $4,000 a year for a little over six years — $25,000 returned on a $10,000 advance. That is an implied cost of money north of thirty percent a year, taken off the top of every sale, before rent, before flour, before profit exists. A bank loan at 9 percent would have been vastly cheaper. "Non-dilutive" is a statement about the share count, not about the price. The money always costs something; the only question is which claim the investor takes — a slice of the ownership, a senior IOU, or a skim of the revenue — and every one of those claims sits somewhere in line ahead of, or alongside, the owner.

To be fair to the royalty, it has one genuine virtue: its cost is capped. If Maria builds a hundred-store chain, the royalty investor still collects only $25,000, while the 20 percent equity she might have sold instead would eventually be worth a fortune. Equity is the cheapest money if you fail and the most expensive money if you succeed; royalties and debt are the reverse. Neither is free. They are expensive in different futures.

Three footnotes before we scale up, because you will meet all three in the wild. First, grants — government or foundation money with no repayment — are the only genuinely free capital, which is why they are rare and fought over. Second, the convertible note and its cousin the SAFE, the standard instruments of startup seed rounds, are marketed as debt-like but convert into shares at the next financing: they are not non-dilutive, they are deferred dilution, a blank check against the future denominator. Public companies have an exact analogue in convertible bonds. And third, royalty financing is not a television gimmick — entire public companies do nothing else. Pharmaceutical royalty funds, precious-metals streaming companies, and music-catalog owners all run door three at billion-dollar scale, buying revenue skims instead of shares. When you understand Maria's fifty-cents-a-pizza deal, you understand their whole business model.

What a share actually is — and no, you cannot take the cash out

Now we can be precise about what all these slices are slices of, because it is not what most people assume.

A share of common stock is a residual claim. Not a claim on the assets, not a claim on the revenue, not a claim on the cash in the bank. A residual claim: the right to whatever is left over after everyone with a prior claim has been paid. The line runs in a fixed order, and you are at the back of it.

This resolves a question we get constantly, and it is a good one: if a company already has cash in the bank and I buy 10 percent of it, do I own 10 percent of that cash? Can I take it out?

You own 10 percent of the residual claim on that cash — which is not the same thing. Take SpaceX as it stood on March 31, 2026, from the balance sheet in the prospectus. It showed $15.9 billion of cash and cash equivalents. It also showed $29.1 billion of long-term debt and $60.5 billion of total liabilities. Buy 10 percent of the common equity that day and you have not bought $1.59 billion of spendable cash; you have bought 10 percent of what remains after $60.5 billion of obligations are satisfied.

Here is the same equity formula from the pizza shop, applied to a real balance sheet — with one wrinkle that trips up even experienced readers:

Assets minus liabilities gives $41.6 billion, yet SpaceX's balance sheet showed total shareholders' equity of $34.5 billion. The missing $7.0 billion is the redeemable convertible preferred stock, which accounting rules park on its own line between liabilities and equity — the "mezzanine" — because in certain events those holders can demand their money back. It is claim number three in Exhibit B, made visible. When you compute stockholders' equity from a real filing, subtract the mezzanine too; the common shareholder stands behind it.

And no — you cannot take the cash out. This is where intuition trained on small business fails. When you own an LLC outright, the company's checking account is functionally your checking account. When you own 10 percent of a corporation, the cash belongs to the corporation, a separate legal person. You own a claim on that person's leftovers, and cash legally reaches you in exactly three ways: a dividend the board declares, a buyback you choose to sell into, or a liquidation at the end of the company's life.

The debt cuts the same way, and this is the more dangerous half of the question. Buy 10 percent of a company carrying $29 billion of debt and you have not personally borrowed $2.9 billion — your loss is capped at what you paid for the shares, which is the entire point of limited liability. But you have absolutely bought 10 percent of the consequences of that debt: the interest that eats the earnings before they reach you, the covenants that constrain the board, and the fact that in a bad enough outcome the lenders take the company and you receive nothing. SpaceX paid $1,476 million of cash interest in 2025. That money left the building before a single dollar could ever have reached a common shareholder.

Where this lives in a filing: the balance sheet appears twice in an S-1 — summarized up front under "Summary Historical Consolidated Financial and Operating Data," and in full, audited, in the F-pages at the back. The debt detail is in "Management's Discussion and Analysis — Liquidity and Capital Resources." And the single best page for the whole capital structure is the Capitalization table, which lays out cash, every layer of debt, the mezzanine, and each class of equity — before and after the offering — on one page.

The real thing: SpaceX's dilution math

Scale Maria's door one up by nine zeros and you have the SpaceX offering. The company had 12,520,309,620 shares outstanding before the deal. It created 555,555,555 new ones — primary shares, every one of them; there were no selling stockholders on this cover page — and sold them at $135.00. Afterward, 13,075,865,175 shares existed.

Existing holders were diluted by just four and a quarter percent — remarkably little for the largest capital raise ever. And that explains the lopsided table we opened with. Because the company was already so valuable, it could take in $75 billion while surrendering only four percent of itself. The existing investors' $6.48 average price is not a discount anyone handed them; it is the arithmetic residue of having bought in when SpaceX was a startup in a warehouse, and of having been right for twenty-four years. The public paid $135.00 because the risk those early investors took had already been resolved. Whether $135.00 was a fair price for the risk that remained is a different question — and the market has spent the weeks since answering it in both directions.

The prospectus then discloses a second, stranger-looking number, in a mandatory section titled "Dilution," and it is worth understanding exactly what it measures. Recall net tangible book value — the pizza-shop equity formula with the intangibles stripped out, because a brand or a customer list cannot be sold off the way an oven can:

SpaceX's version of the waterfall, per share:

You can reproduce every line yourself. Tangible book before: $28,251 million ÷ 12,520 million shares = $2.25. Add the offering's $74.4 billion of net proceeds: $102,697 million ÷ 13,076 million shares = $7.85. (The roughly $13.3 billion gap between SpaceX's $41.6 billion of pro forma equity and its $28.3 billion of tangible equity is goodwill and intangibles, mostly from folding in X and xAI.) Buyers paid $135.00 for a claim on $7.85 of tangible book value; the other $127.15 was payment for expectations — Starlink's growth, reusable-launch economics, whatever the AI segment becomes.

Now, two misreadings to avoid, in opposite directions. A $127.15 "dilution" figure does not mean you overpaid by $127.15 — nearly every good business trades far above tangible book, and should; Coca-Cola's brand appears on no balance sheet. But the figure does tell you something real: how much of your purchase is backed by things you could touch versus backed by a story. Here the ratio was 94 percent story. That is not automatically wrong. It is simply worth knowing which bet you are making. And notice, in the fourth row, who the mechanical winner is: existing holders watched their tangible book value rise from $2.25 to $7.85 the moment new investors paid above book — Maria's $300-to-$320 move, scaled up. Completely standard, completely legal, and the clearest possible picture of what the pie-cutting accomplishes.

Who decides what happens to the money

A reader asked a question that goes deeper than it first appears: before you take on an investor, you decide what happens to the profits — you can plow them back into the business if you like. Once an investor comes in, do they get to decide what happens to their share of the profits?

The answer is no, and the reason is the fundamental trade of corporate ownership. The investor does not receive 20 percent of the profits to direct as they please. They receive 20 percent of the claim on those profits, and the decision about what happens to the money moves to a body neither owner fully controls: the board of directors. The board chooses among exactly four uses — reinvest in the business, pay down debt, pay a dividend, buy back shares — and it owes its duties to shareholders collectively, not to you specifically. A shareholder who wants a dividend and a board that prefers reinvestment is a disagreement the board wins. Your genuine remedies are two: vote for different directors, or sell.

SpaceX is unusually direct about this. Its Dividend Policy section — a short, mandatory section that always sits right after Use of Proceeds — states that the company does not anticipate paying cash dividends in the foreseeable future, intends to retain earnings to fund growth, and, in a clause worth reading twice, that covenants under its credit agreements restrict its ability to pay dividends at all. That last part is not a preference; it is the lenders — claim number two in Exhibit B — contractually limiting what can flow down to claim number four. The residual claim, restricted by the senior claim, exactly as the diagram promised.

What you actually get: votes, classes, and the fine print

So what does a share buy you in governance terms? Ordinarily: one vote per share, a say on directors and major transactions, dividends if declared, and a pro-rata slice in liquidation. Ordinarily. SpaceX went public with two classes of common stock, and the arithmetic deserves to be seen in full.

Class A — every share the public bought — carries one vote. Class B, held by insiders, carries ten. Run the numbers from "The Offering" section:

All of the public's shares, together, control 11.5 percent of the vote. Musk alone controls approximately 82.4 percent. And the raw percentage actually understates the position, because the charter adds structure on top: Class B holders, voting separately as a class, elect 51 percent of the board for as long as a single Class B share exists; Musk can be removed from the board, or from his CEO and Chairman roles, only by a vote of the Class B holders — Class A has no mechanism at all; and the company qualifies as a "controlled company" under Nasdaq's rules, letting it opt out of certain governance requirements, which it has said it intends to do. There is even a ratchet built in: Class B converts to Class A when transferred, so as other insiders sell over the years, the voting power of whoever keeps their Class B mechanically concentrates.

We are not rendering a verdict on founder control; reasonable investors disagree, and the structure has produced both legendary compounding and spectacular ruin, occasionally at the same company. Insulation from quarterly pressure has real force when the mission is rockets to Mars. The removal of the shareholders' primary check has real force in the other direction. What we insist on is only that you know which instrument you are buying: SpaceX Class A is an economic interest with essentially no governance interest attached. That may be a perfectly good trade. It is not the trade most people picture when they hear the word "stock."

One more clause, easy to miss and permanently relevant: the prospectus states that holders of common stock have no preemptive rights. A preemptive right would let you buy into future share issuances to maintain your percentage. Without one — and nearly every U.S. public company is without one — every future issuance simply dilutes you: acquisitions paid in stock, follow-on offerings, and above all the employee equity we take up at the end of this letter. The dilution question is never settled at the IPO. It is a standing feature of holding the stock.

Where this lives in a filing: the "Description of Capital Stock" section, near the back — dry, fifteen to twenty pages, and where the actual rules are. Read the subsections on Voting Rights, Election and Removal of Directors, Conversion, and No Preemptive or Other Rights. If there is more than one share class, the risk factor titled something like "our dual class structure concentrates voting control" restates the consequences in plainer language.

How to read an S-1 in an afternoon

The SpaceX prospectus runs 313 pages, and almost nobody should read it cover to cover. The document has a rigid, legally mandated structure, which means you can navigate it like a reference book. Here is the order we work in, and what each stop is for.

First, the cover page — ninety seconds. How many shares, and offered by whom: the company, selling stockholders, or both? At what price, on which exchange, under what ticker? How large is the over-allotment option? How many share classes? SpaceX's cover answered everything: 555,555,555 shares, all offered by the company, expected at $135.00, Nasdaq under SPCX, an 83,333,333-share over-allotment, two classes, and a founder holding 82.4 percent of the vote.

Second, "The Offering" summary box — the cover page in expanded form: shares outstanding after the deal, voting power by class, use of proceeds, dividend policy, all on two pages.

Third, the footnotes underneath "The Offering." Do not skip these. They reconcile the share count and disclose the overhang — every option, restricted stock unit, and reserved share that is not in the headline number but will become stock eventually. We do the arithmetic on SpaceX's overhang at the end of this letter, and it is startling.

Fourth, "Use of Proceeds." What the company says the money is for, and how much discretion it admits to. SpaceX listed AI compute, launch infrastructure, and satellite capacity — then stated plainly that management will have significant flexibility in applying the proceeds. Honest, standard, and a reminder that stated uses are intentions, not commitments.

Fifth, "Capitalization," and sixth, "Dilution" — the two pages we have already mined. Cash, debt, mezzanine, and equity in three columns; then the tangible-book waterfall and the who-paid-what table. If you read only two pages of any prospectus, read these two.

Seventh, "Risk Factors" — selectively. It is long and drafted to be unfalsifiable. Read the Summary of Risk Factors up front, then jump to the subsection on risks related to the corporate structure and the offering itself, where the structural traps specific to buying this stock — as opposed to the risks of the business — are concentrated.

Eighth, MD&A — management's own narrative of the numbers. The valuable parts are Key Business Metrics, the segment discussion, and Liquidity and Capital Resources. It is where you learn SpaceX reports three segments, and that in 2025 Connectivity earned $4.4 billion of segment operating income while AI lost $6.4 billion.

Ninth, "Underwriting," and tenth, "Shares Eligible for Future Sale" — the fee machinery and the lock-up calendar, each about to get its own section of this letter.

Eleventh, the F-pages — the audited statements and, more importantly, the notes, where share-based compensation, debt terms, and the preferred stock's rights actually live.

All of it is free. The SEC runs two search tools, and knowing which is which saves real time:

The two EDGAR front doors. sec.gov/search-filings is the one to bookmark: type a company name or ticker, get its complete filing history, filter by form type — S-1 for the IPO prospectus, then 10-K, 10-Q, and 8-K once the company is public. Take the most recent amendment; "Amendment No. 2" supersedes the original. sec.gov/edgar/search/ is a different tool: full-text search inside filings from 2001 on — the one you want when you are hunting a phrase ("lock-up," "preemptive," a subsidiary's name) rather than a company. Same database, two doors: one finds documents, the other finds words.

A word on where a paid platform earns its keep, since EDGAR is free and authoritative and we would never suggest replacing it. EDGAR gives you documents; it does not draw you a fifteen-year chart of share-based compensation against revenue, or lay out a company's full bond stack with coupons and maturities on one screen, or plot diluted share count across a decade so a quiet pattern becomes an obvious one. That assembly work is what we use GuruFocus for (affiliate link — subscribing through it costs you nothing extra and may earn us a commission; we use it either way), and we will point out the specific screens below where it saves an hour at a time. The pattern is always the same: find it fast on the platform, confirm it in the filing before acting.

The underwriters: who gets paid, and for what

Twenty-three investment banks underwrote the SpaceX offering, with Goldman Sachs, Morgan Stanley, BofA, Citigroup, and J.P. Morgan as the lead names. What "underwriting" means is more interesting than the org chart.

In a firm-commitment underwriting — the standard structure, used here — the banks are not brokers matching buyers and sellers. They buy the entire offering from the company at a discount to the public price, and resell it. The prospectus uses exactly those words: the underwriters "severally agreed to purchase" the shares. Once the agreement is signed, the company's money is locked in; if the banks cannot resell at the offering price, that is their problem. The discount they keep — the gross spread — is the fee for taking that risk, running the roadshow, building the order book, and supporting the stock afterward.

The preliminary prospectus leaves the fee table blank, as they all do before pricing, but you can back the number out from the disclosures around it:

Context makes that number the lesson. The textbook gross spread for a U.S. IPO is seven percent, and for small and mid-sized deals it has been stuck near seven percent for decades. SpaceX paid roughly a tenth of that rate — for the same reason Facebook paid about 1.1 percent and Alibaba about 1.2: the work of underwriting does not scale with deal size, and giant issuers have the leverage to price accordingly. Turn that around and it is a genuinely underappreciated fact about small companies: a $50 million biotech hands $3.5 million to its bankers before it sees a dollar, a recurring tax on small-company capital formation and one of the honest arguments for the alternative routes we cover later. One SpaceX-specific wrinkle worth noticing: the banks agreed to take no discount on shares sold through the over-allotment option — not typical, and it changed the economics of exercising it.

Next, the over-allotment option, nicknamed the greenshoe after the 1963 Green Shoe Manufacturing offering that invented it. The banks held a 30-day right to buy up to 83,333,333 additional shares — exactly 15 percent of the base deal, the market convention — from the company at the offering price. The mechanism is elegant: the underwriters deliberately sell investors more shares than they bought, about 115 percent of the deal, leaving themselves short 15 percent. If the stock trades up, they exercise the option and cover their short at the offering price — the company sells more stock and collects more money. If the stock trades down, they cover by buying in the open market instead — real demand hitting the tape precisely when the stock is weak, which supports the price, and the option expires unused. On the covered portion the banks cannot lose; that is the design, and it is what lets them stabilize the price without taking directional risk. The prospectus discloses the rest of the toolkit too: a possible naked short beyond the option, and Morgan Stanley as named stabilization agent authorized to bid for shares to steady the price. All of it is legal, regulated price support, disclosed in advance — and a good reason to treat the first few weeks of any IPO's trading as something other than a clean market. In SpaceX's case the option was exercised in full, lifting total proceeds to roughly $85.7 billion.

Now the incentive question a careful reader should already be asking. The company wants the highest possible price. The banks' fee is a percentage, so nominally they want that too — but they also have to place the stock with the same institutional buyers they will call again next month, and those buyers remember how the last deal traded. The result is a persistent, well-documented lean toward pricing a little below what the market will bear. SpaceX priced at $135.00, opened at $150.00, and closed its first day at $160.95:

Fourteen billion dollars of value transferred from the company's treasury to first-day buyers — disproportionately the institutions the banks most need to keep happy. We will be fair: some underpricing is genuinely necessary, since nobody buys an untested security priced to perfection, and a broken deal does lasting damage; the banks price under real uncertainty. But when the gap runs to twenty-nine times the fee, it is at minimum worth asking whose interests the price served. And then the market complicated the story on its own: the stock touched $225 within days and, as we write, trades near $115 — below the offer. The verdict on whether $135 was "too low" has aged in an unexpected direction.

Lock-ups: the supply that is still coming

Only the shares sold in the offering trade freely on day one. Everything else — founder, funds, employees — is frozen by contract with the underwriters. That freeze is the lock-up. It is not law; it is an agreement the underwriters can waive, which is itself worth knowing. The convention is 180 days, and the popular image is a cliff: one dreaded date when the insiders' shares flood out. Modern lock-ups are nothing like that. SpaceX's is a staircase with three separate handrails, and every step of it is disclosed.

Three groups. Most holders: 180 days, with early-release tranches. A group of large shareholders: locked until two trading days after the second-quarter 2027 results — more than a year out, with staged early releases of its own that begin only in 2027. And Musk: 366 days, with the prospectus stating flatly that his shares have no early-release provisions of any kind. Together, Musk plus the extended group hold roughly 7.8 billion shares — more than 63 percent of everything outstanding before the deal.

The staircase for the 180-day group, from the "Shares Eligible for Future Sale" table: up to 911.5 million shares release two trading days after the first earnings report; another 455.8 million release at the same moment but only if the stock closed at least 30 percent above the offer — $175.50 — on five of the ten trading days into that report; roughly 319 million more at each of days 70 and 90; 59 million affiliate shares at day 91; 328 million at day 105; further seven-percent tranches at days 120 and 135; another 28 percent after the third-quarter report; and everything remaining at day 180.

Stop on that dashed step, because it is the strangest incentive in the whole document: a price-contingent release, in which good stock performance triggers additional supply. A holder who wants out spends ten specific trading days rooting for strength. These provisions have become common and are almost never discussed. And one more disclosure hiding in plain sight: SpaceX reserved up to five percent of the offering — roughly 27.8 million shares — for a directed share program, sold at the offer price to employees and people chosen by executives, and the prospectus states those shares are not subject to any lock-up at all. Two sentences, two places in the document, absent from every piece of coverage we saw.

If you own a recently public stock, the practical move is unglamorous: open "Shares Eligible for Future Sale," and put the dates in your calendar. Supply arrives on a published schedule.

How to actually buy an IPO — and the trap on the way out

For most of modern history, retail investors simply could not buy at the offering price; allocations went to institutions and favored clients, and the public bought in the aftermarket — which, recall, is how first-day buyers came to capture that $14.4 billion. That has genuinely changed, and SpaceX is the clearest case yet: the prospectus itself names Charles Schwab, Fidelity, Robinhood, SoFi, and E*TRADE as retail channels, at the same price and the same time as the institutions, with roughly 30 percent of the deal reportedly set aside for retail.

Here is the process at Schwab, since that is where many of our readers sit. Check eligibility weeks ahead — Schwab requires stated investment objectives, sufficient knowledge and experience, an eligibility questionnaire, and a minimum liquid net worth; people routinely discover they don't qualify on the morning the book closes. Find the deal under the Trade tab on the IPO page's Calendar of Offerings. Read the preliminary prospectus — you must attest that you have, and having read this letter you now know which ten pages matter. Submit a Conditional Offer to Purchase, an indication of interest specifying shares and the top price you will accept, not yet an order. Reconfirm after pricing if the final price lands outside your stated range, or your interest lapses. And expect less than you asked for: SpaceX's book was reported around $150 billion against a $75 billion deal, so allocations were cut hard. One hard stop: FINRA Rules 5130 and 5131 bar securities-industry personnel, their immediate families, and executives positioned to steer investment-banking business from buying IPOs at the offer, at any broker.

Then comes the part that costs people: the flipping rules. Selling allocated shares shortly after the IPO angers underwriters, who punish brokers whose clients do it with smaller future allocations — so the brokers police their own customers, each with rules of its own invention:

Each of those is the broker's own published policy as of this writing, not regulation — which is exactly why they differ this much, and why you should read yours before submitting the order rather than after. Note the asymmetry while you are at it: you are penalized for selling too early, but no one is obligated to protect you if the stock falls while you dutifully hold. The lock-up restrains insiders; the flipping policy restrains you; only one of the two ends in a lifetime ban. A last mechanical detail: when a brokerage participates in distributing an offering, even secondary-market purchases of that stock are typically not marginable for the first 30 days.

What a mega-IPO does to your index fund

You may own SpaceX today. You may even have chosen to. Here is the version where you didn't.

Under the Nasdaq-100's long-standing rules, a newly public company waited at least three months for index eligibility and had to keep at least ten percent of its shares in public hands. In May 2026 — weeks before this IPO — Nasdaq adopted a fast-entry rule: any newcomer whose market cap ranks inside the index's top 40 becomes eligible after just 15 trading days, float requirement waived, with as little as five days' notice. SpaceX joined the Nasdaq-100 on July 7, 2026, fifteen trading days after its debut — the fastest inclusion of a major U.S. benchmark on record, under a rule written on the eve of its arrival.

The mechanics were automatic from there. More than $800 billion sits in products that track the Nasdaq-100 outright, with all index-linked assets estimated near $1.4 trillion, and every tracking dollar became a forced buyer. JPMorgan put QQQ's buying alone near $4.3 billion, with total flows across Nasdaq-100- and Russell-linked products estimated at $22 to $27 billion, executed around the July 6 close and July 7 open with the stock in the $157–161 range. No constituent was removed — the index simply ran over 100 names — so the purchase was funded by shaving every other holding. If you owned QQQ in a 401(k), you sold slivers of Apple, Microsoft, and Nvidia and bought SpaceX at roughly $160, without being asked. As we write, the stock trades near $115. And the thin float made it stranger still: with only a few percent of the company actually trading, a $4 billion mechanical bid landed on a market that had little inventory to absorb it — and each future lock-up release that widens the float triggers further index rebalancing on a schedule everyone can read in advance.

The half of the story that deserves equal emphasis: S&P Dow Jones declined to follow. On June 4 it rejected its own proposal to fast-track mega-cap IPOs into the S&P 500, keeping the twelve-month seasoning requirement and the GAAP profitability test — positive as-reported earnings in the latest quarter and over the trailing four. SpaceX lost $4.9 billion in 2025 and another $4.3 billion in the first quarter of 2026; it fails the test today and cannot clear seasoning until at least mid-2027. Holders of S&P 500 funds — the default in a great many retirement plans — were untouched entirely.

We draw no conclusion about which committee was right. The lesson is narrower and more useful: index funds are not neutral. They are rules-based; the rules are written by committees; committees can rewrite them; and when they do, tens of billions of dollars move mechanically in response. "Passive" describes your behavior, not the fund's. We think most people should own index funds — and should also know which rulebook they have signed.

Four doors into the same building

An IPO is one route to a public listing, not the only one. The four routes end in the same place — a share you can trade — but they cut the pie in very different ways, and the differences are exactly the concepts we have already built.

The direct listing is the secondary sale writ large: existing shares simply begin trading, priced by an opening auction rather than a negotiation, typically with no underwritten book, no new shares, no dilution — and often no lock-up at all. Spotify pioneered it in 2018; Coinbase, Slack, and Palantir followed. The appeal is everything we just documented in reverse: no underpricing gap, no fourteen billion left on a table, no expiry dates to dread. The catch is equally direct: the company raises nothing. (The rules now technically permit raising capital in a direct listing, but it remains uncommon.) It suits a business that is already well capitalized and merely wants liquidity for its holders. It cannot fund a Mars program.

The SPAC deserves a slower walk, because its pie-cutting is baroque. A special purpose acquisition company is a shell with no operations that IPOs at $10.00 a share, parks the cash in trust, and has a deadline — typically 18 to 24 months — to merge with a private company, which becomes public through the merger (the "de-SPAC"). Two features drive the economics, and both come out of your slice. The first is the sponsor promote: historically the sponsor received about 20 percent of the shell's shares for a nominal sum — pure dilution of everyone else, and an incentive to close some deal, any deal, since the promote is worth millions if a merger closes and nothing if it doesn't. Terms have improved — promotes have compressed toward 10–15 percent and increasingly vest on stock performance rather than mere completion — but the conflict is structural. The second is the redemption right: every SPAC shareholder may take back their ~$10 plus interest instead of holding through the merger, and in recent years most have — redemption rates ran above 90 percent through 2023 and much of 2024, easing to roughly 79 and then 68 percent in late 2025. Many SPAC IPO buyers never intend to own the target at all; they hold a cash box with a free option attached. The consequence for you: a SPAC that raised $300 million can arrive at closing with $30 million, while the target was valued — and the promote sized — as if it all showed up. The record reflects the structure: academic studies of the 2018–2022 vintages put average one-year post-merger returns around negative 30 to 60 percent. The market has genuinely matured — de-SPAC fees have settled near 3–4 percent and contingent on closing, disclosure now resembles a real prospectus, litigation rates have fallen, and roughly 251 SPACs were hunting targets with about $47 billion in trust as of June — but read any de-SPAC proxy with one question fixed in mind: how much of this pie was cut for the sponsor before I arrived?

The spin-off is the route with no transaction in it at all. A parent distributes shares of a subsidiary directly to its own holders — no sale, no proceeds, and if the deal qualifies under Section 355 of the tax code, no tax at either level. You wake up owning two companies instead of one; your economic interest hasn't changed, only divided into separately traded pieces. The last year supplied a parade of examples: Honeywell distributed Solstice Advanced Materials in October 2025 — one Solstice share per four Honeywell shares — then separated Honeywell Aerospace in June 2026; DuPont spun out Qnity Electronics; Comcast separated Versant; Western Digital spun off Sandisk.

Spin-offs deserve particular attention from patient investors, for a structural rather than fundamental reason. When a small subsidiary lands in the accounts of a large parent's shareholders, many of them are not allowed to keep it: index funds tracking the parent's index must sell what isn't in the index, mandates forbid off-benchmark positions, and no analyst covers the orphan yet. The result is often weeks of selling driven entirely by who is permitted to own the shares rather than by what they are worth — forced selling with no informational content, one of the few reliably inefficient corners of a large-cap market. It guarantees nothing; plenty of subsidiaries are spun precisely because the parent wanted rid of them. But it earns a look every single time. Longtime readers will recognize that this is not academic for us: several Capital Appreciation positions — GE Vernova out of General Electric, Sandisk out of Western Digital, Howmet out of the old Alcoa/Arconic line — arrived in the world exactly this way.

ADRs: owning a foreign company from a U.S. account

If you have owned Toyota, Nestlé, or ASML through a U.S. brokerage, you probably did not own their shares. You owned an American Depositary Receipt — and the machinery behind that acronym answers a whole cluster of reader questions about fees, votes, and currency, so we will take it apart properly.

The mechanics. A U.S. depositary bank — BNY, Citi, JPMorgan, or Deutsche Bank, mostly — buys shares on the company's home exchange and holds them with a local custodian. Against those shares it issues receipts that trade in dollars, clear through U.S. systems, and sit in your account like any stock. You are therefore one step removed: you hold a claim on a bank that holds the shares. The ratio need not be one-to-one — one ADR might represent five ordinary shares or a fifth of one — which is why an ADR's price rarely matches the home-market quote directly.

Sponsored versus unsponsored — and who gets your vote. A sponsored ADR is created under an agreement with the company; an unsponsored ADR is created by a bank on its own initiative, without the company's participation. For governance the distinction is decisive. With a sponsored program the depositary typically forwards proxy materials and votes the underlying shares as you instruct — an attenuated voice, but a voice. With an unsponsored program the depositary bank keeps the right to vote the shares it holds in trust: you get the economics, the bank gets the vote. Even sponsored voting is imperfect — earlier deadlines, later materials, and some deposit agreements let the depositary vote uninstructed shares as management recommends — so if governance is part of your thesis, read the deposit agreement before assuming you have a say. Sponsored programs come in three levels: Level 1 trades over the counter with minimal SEC registration; Levels 2 and 3 list on the NYSE or Nasdaq and file annual reports on Form 20-F; only Level 3 can raise capital here. Do not assume Level 1 means small — Nestlé, Roche, and Heineken trade that way.

Who makes the money. The depositary bank, in ways that are disclosed but easy to miss. Depositary service fees are charged straight to holders — a typical schedule runs up to $5.00 per 100 ADRs for issuance or cancellation, a cent or two per ADR skimmed from each dividend, sometimes an annual maintenance fee that appears on your statement as an undecipherable debit. Since a rule change in the mid-2000s, the investor bears these costs in sponsored and unsponsored programs alike. The bank also earns on foreign exchange, converting every dividend from local currency to dollars at a spread, often through its own affiliates — on a large program paying quarterly dividends, a substantial and nearly invisible business. For unsponsored programs, banks even pay rebates to brokers who route issuance their way, which the broker may or may not share with you.

The risks worth naming. Currency, first and always: your return is the stock's return in its home currency plus that currency's move against the dollar, and over any given year the currency leg can dominate the equity leg — a European stock up 10 percent while the euro falls 10 percent leaves you roughly flat. Withholding tax, second: most countries withhold 15 to 30 percent of dividends to foreigners before you see them. In a taxable account you can usually reclaim it via the foreign tax credit; in an IRA you generally cannot, because there is no U.S. tax to credit it against — the withholding is simply gone, which is one of the few genuinely asset-location-dependent facts in ordinary investing and an argument for holding foreign dividend payers in taxable accounts where practical. Add program termination (a depositary can wind up an ADR program on roughly 30 days' notice and sell the underlying, realizing gains on your behalf at a moment you didn't choose), thinner disclosure (a 20-F annually, no quarterly 10-Qs required), and, for OTC programs, thin volume and wide spreads. None of this makes ADRs a bad instrument — for most people they are the only practical way to own great businesses domiciled elsewhere, and the do-it-yourself alternative of foreign accounts and tax reclaims is worse for nearly everyone. But the costs are real, the vote is often not yours, and the currency exposure is not optional.

After the IPO: sell stock, borrow, or issue preferred?

Going public does not end the financing question; it changes the menu. A public company has three main ways to raise more capital, and its choices among them are one of the clearest windows into how management thinks. SpaceX gave us a live demonstration eleven days after listing.

Option one: sell more stock. A follow-on offering runs Maria's door one again — new shares, money in, dilution up. Equity's virtues are that it never matures, pays no interest, and carries no covenants; in a true crisis, the all-equity company simply survives. Its costs are permanent dilution and a well-documented signaling problem: managements tend to sell stock when they privately believe it is expensive, investors know this, and announcements of big equity deals routinely knock the price down. The company is selling; management knows more than you; draw the inference. Watch, too, for the shelf registration — Form S-3 — which lets a seasoned company pre-register securities and fire at will later. A freshly filed multi-billion-dollar shelf is not a prediction. It is a loaded option.

Option two: borrow. On June 23, 2026, SpaceX priced its first bond deal — five tranches of senior unsecured notes, $25 billion in all:

The order book reportedly reached $89 billion — three and a half times the deal — and the proceeds retired the $20 billion bridge loan from the xAI acquisition. Why borrow instead of selling more stock at $160? Because debt does not dilute: not one share was created, every holder's percentage untouched. Interest is tax-deductible where dividends are not, and every dollar the business earns above that 5.35-to-6.65 percent hurdle accrues entirely to the shareholders. The price of those virtues is rigidity — interest owed in every environment, covenants constraining the board (recall the dividend restriction we met earlier), and a permanent claim standing ahead of yours in Exhibit B. Notice one mark of competent treasury work in that table: the maturities ladder from 2031 to 2056, so no single year presents a refinancing wall. A company with all its debt due at once is one bad credit market away from a crisis.

Option three: issue preferred stock — the instrument that lives between the other two: a fixed dividend, priority over common in liquidation, usually no vote, no maturity. Issuers like it because it doesn't dilute common voting power and rating agencies treat it as partly equity; buyers should understand it as an income instrument with capped upside. It is the native financing of banks, insurers, utilities, and REITs, and rare among growth companies — which is what makes SpaceX's history with it so instructive.

Before the IPO, SpaceX had raised billions through redeemable convertible preferred stock, the standard instrument of venture investing: liquidation preferences ahead of common, plus the right to convert into common. At the IPO, all of it vanished — by design. The charter provided for automatic conversion upon a "Qualified IPO," defined as a registered offering with at least a $6 billion pre-offering market cap raising at least $250 million; SpaceX cleared both by orders of magnitude. In that instant, 134,451,267 preferred shares converted into 6,722,563,350 shares of common stock — about 3.45 billion Class A and 3.27 billion Class B (fifty-to-one, reflecting the original conversion terms compounded by a five-for-one split that May) — and $7.0 billion moved off the mezzanine line into permanent equity.

Sit with the size of that. Nearly 6.7 billion of the 12.5 billion "pre-IPO shares" in our opening table did not exist as common stock until the moment the deal closed. This is the great hidden denominator of IPO investing: a company can look reasonably capitalized until you notice that a preferred stack converts into more than half the post-IPO share count. It is always disclosed — in the Capitalization footnotes and the preferred note in the F-pages — and it is never in the headline. Read the pro forma column. Always.

How do you check whether any company has bonds or preferred outstanding? Four places, in order of speed. The Capitalization table in an S-1, or the balance sheet in a 10-K — preferred shows up in equity or, if redeemable, on its own mezzanine line; debt splits into current and long-term. The debt note in the financial statements — every instrument, rate, maturity, covenant. EDGAR full-text search for the ticker plus "424B," since bond and preferred offerings file public prospectus supplements. And, least glamorous but fastest, a data platform: pulling five bond tranches out of a 300-page filing takes an evening, while GuruFocus (affiliate link) lays out a company's full debt stack — coupons, maturities, and any preferred with its rate and call date — on one screen. We use the platform to find it in seconds and the filing to confirm it before acting.

Stock-based compensation: the dilution that never stops

Everything so far has concerned an event. This last mechanism is a process, and over a long holding period the process usually matters more.

When a company pays employees in shares — options, restricted stock units, restricted awards — it is buying labor with ownership. Accounting rules require the grants to be expensed at fair value, so the cost does appear on the income statement. The trouble is where else it appears: because no cash leaves the building, stock-based compensation is added back on the cash flow statement, and added back again in nearly every "adjusted" profit metric a company promotes. That framing invites a specific mistake. SBC is not a cash expense; it is an ownership expense. The company did not spend money. It spent you — and you have no preemptive rights with which to object. The honest test: had those employees been paid in cash, the cash would be gone; to keep the share count flat instead, the company must buy the shares back — with cash. The expense is real. It is merely denominated in your percentage rather than in dollars.

SpaceX's numbers, from the cash flow statement and the segment reconciliation:

Now the line that matters most. SpaceX reported 2025 Adjusted EBITDA of $6,584 million — a figure reached by adding the $1,947 million of SBC back. Roughly 30 percent of the headline profit metric was not earned in cash. It was paid in your slices.

Why is too much of this bad? Three distinct harms. It transfers ownership — mechanically, continuously, without consent. It obscures profitability — a company can post handsome adjusted numbers while the share count compounds upward, so per-share results, the only kind you can own, grow far slower than the headline; this is why we chart diluted shares outstanding as a time series next to earnings, and why a rising count against flat earnings is a genuine warning. And it is procyclical in the worst way: when the stock falls, old grants lose their retention power, and companies respond by granting more — dilution accelerating precisely when the stock is weakest. Buybacks are the standard defense, and here is the trap inside the defense: a company can spend billions "returning capital" while its share count never falls, because the buyback is merely mopping up the employee issuance. That is not capital returned; it is compensation paid in cash on a delay, routed through the market. The test is never dollars spent on buybacks. It is whether the diluted share count actually falls.

Can you forecast future SBC? Yes — better than most people expect. A reader asked us this directly while we were drafting, and the answer is one of the most practical things in this letter, because three disclosures turn next year's dilution from a guess into an estimate with a floor, a ceiling, and a blind spot you can name.

The floor: unrecognized compensation cost. Every company that grants equity must disclose, in the share-based compensation note, the expense for awards already granted but not yet vested, and the period over which it will hit the income statement. SpaceX's Note 15: total remaining expense for unvested options, RSUs, and RSAs of $4,842 million, to be recognized over a weighted-average 3.2 years.

Those awards exist; the employees hold them; the expense arrives whether or not another share is ever granted. Against 2025's actual $1,947 million, the disclosure tells you the next three years start at roughly a billion and a half annually, and everything new stacks on top. One line in one note, available for every company you own.

The ceiling: the reserved pool. The footnotes under "The Offering" disclose 299,256,055 shares reserved for future grants under the equity plan, plus 24,026,920 under the employee stock purchase plan — roughly 323 million shares management is pre-authorized to hand out, about $43.6 billion of compensation capacity at the offer price, no further approval needed. The pool doesn't give you timing; it gives you the outer bound. Two things to watch around it: when a company's reserve runs low, a proxy proposal to enlarge it follows, a reliable early signal of accelerating dilution; and many young public companies carry an evergreen provision that refills the pool automatically every year by a fixed percentage of shares outstanding — dilution on a schedule, forever, disclosed in the equity-plan description and worth checking for by name.

The blind spot: awards not being expensed at all. The accounting policy in the same note states that for awards with performance conditions, no expense is recognized until vesting becomes probable. Now apply that to the most spectacular grant in the document: Musk's 2026 award of 1,302,072,285 restricted Class B shares. One billion of them vest only upon hitting market-cap milestones across fifteen tranches and establishing a permanent, million-person human colony on Mars; the rest require the milestones and off-Earth data centers delivering 100 terawatts of compute a year. Until those become probable, essentially none of the expense touches the income statement — yet the shares exist, sit in the Class B count, and vote today. At $135 the grant is worth roughly $176 billion notionally, about ten percent of the company, at an accounting cost, for now, of approximately nothing. Every element is disclosed and the treatment is correct under the rules. The lesson is simply that "check the SBC line" is not enough; the note tells you what the line is not yet carrying.

A fourth, softer signal: grant-date fair values, also in the note. SpaceX's weighted-average RSU grant value went $15.60 → $17.68 → $54.84 across 2023–2025. Hold the number of units roughly constant while the share price triples and the expense triples with it; run it the other way and a falling stock forces the company to issue far more units per dollar of intended pay — more dilution exactly when it hurts. The relationship is worth modeling in both directions.

Finally, add up the overhang from those footnotes we told you not to skip:

The entire IPO was 555,555,555 shares. The employee equity overhang is roughly 1.7 times the size of the offering — and that excludes Musk's 1.3 billion performance shares entirely. None of it is in the headline count. All of it is disclosed. Almost nobody adds it up.

For tracking this over time, this is the other screen we lean on a charting platform for: SBC as a percentage of revenue across ten or fifteen years, plotted beside diluted shares outstanding, turns a disclosure buried on page 214 of an annual report into a single picture — and a company whose count rises every year despite billions of "buybacks" becomes obvious at a glance. That chart is precisely the view we pull up on GuruFocus (affiliate link) before we open the filing itself.

The takeaway

We have covered a great deal of ground. If you keep five things, keep these.

Check the cover page for primary versus secondary. New shares fund the company and dilute everyone; a selling stockholder's shares fund the seller and dilute no one. Four seconds, and you know what kind of transaction you are looking at — whether it is a pizza oven or a Mars program.

Read the Dilution section and the Capitalization footnotes together. Between them you learn what your money buys in tangible terms, what everyone before you paid, and what converts into common stock at closing. SpaceX's preferred quietly became 6.7 billion shares. The pro forma column is where the truth lives.

Know what your vote is worth before you buy. A share carrying one-tenth of the voting weight of the insiders' shares, at a company where the founder can be removed only by a class he controls, is a different instrument from one-share-one-vote — sometimes an acceptable one, never an identical one. The Description of Capital Stock is where the rules are.

Put the lock-up dates in your calendar, and remember the money always costs something. Supply arrives on a published schedule — here, releases worth 2.5 IPOs at the first earnings report alone, with 63 percent of the company still queued behind them. And whether capital arrives through new slices, a senior IOU, or a skim of every sale, the investor gets a claim; "non-dilutive" describes the share count, never the price.

Track the share count, not just the earnings. Use unrecognized cost for the floor of future stock compensation, the reserved pool for the ceiling, the performance-condition policy for what is not being expensed yet — and judge buybacks by whether the diluted count actually falls.

We will close where we started. On a single morning, new investors put in 48 percent of every dollar ever contributed to SpaceX and received 4.2 percent of the company. Nothing about it was hidden. It was printed in a table on page 71 of a free document, beneath a share-count footnote almost nobody adds up, beside a lock-up calendar almost nobody diaries. The information asymmetry in public markets is not, mostly, that insiders know things you cannot know. It is that they read the filings and you do not. That is a much easier problem to fix — and fixing it is what these letters are for.

See it yourself: every SpaceX filing is on EDGAR under CIK 0001181412 — take Amendment No. 2 of the S-1, dated June 3, 2026. The Dilution table is page 71; the lock-up schedule is under "Shares Eligible for Future Sale"; the SBC note is Note 15 in the F-pages. The bond pricing is in the company's June 23 8-K exhibit. For everything else: sec.gov/search-filings to find a company's documents, sec.gov/edgar/search/ to search inside them. The SEC's investor bulletin on ADRs and FINRA Rule 5130 cover those corners in the regulators' own words; Schwab's IPO process is documented on its own pages.

Tools we use — see our full toolkit (some links, including GuruFocus above, earn us a commission at no cost to you). Nothing here is investment advice or a recommendation to buy or sell any security; we are a publisher, not an investment adviser. Figures are from public filings as cited; market prices are as of late July 2026 and will have changed.