Verified Platforms
Quick Links
Where to Stay Secure
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Five times more demand than there is stock on offer does not make a share offering more expensive. On 10 August 2026 Intel asked the market for $15 billion of new stock and the order book came back at roughly $100 billion. The company raised the size of the deal to $20 billion, struck the price at $95, and the shares closed down 4.1% that day. Nothing malfunctioned in that sequence. It is what the sequence is built to do.
A crypto-first reader arrives with a strong prior here, and inside its home market the prior is correct. In a token sale, oversubscription is the whole story: the allocation is fixed in advance, the round closes early, and the surplus shows up as a higher price the moment the thing trades. An equity raise runs the other way, because someone on the issuer's side decides how much supply exists, and they decide it after seeing the demand.
This walkthrough puts Tao and Eunha on opposite sides of that gap. Tao is Kodex's bridge between structure and instinct, closer to student than master, and today he arrives holding an answer he is sure of. Eunha lives between structure and emotion, and she takes an answer apart by asking about it rather than by correcting it.
Tao gets there before the filing is open. "A book five times covered is a bidding war," he says. "The price should have gone up."
Eunha asks him a small question first. Who decides how many shares exist?
Tao starts to answer and stops, because in the market he knows best, nobody does. Supply is written into the contract before anyone bids on it. He has spent years reading order books where the quantity is the one thing that cannot move, so he reads every book as a contest between buyers.
In a follow-on equity offering the quantity is the one thing that can move. A company decides it wants to raise a number, say $15 billion, and hires banks to find out what the market will pay for it. Those banks spend a day or two collecting bids from institutions: how many shares, at what price, from whom. That process is a book-build, and it ends with one number that applies to everyone. There is no ladder of winners paying more than losers. Every buyer in the deal gets the same price. That price is set slightly below where the stock is trading, because a buyer who could simply click the open market needs a reason to take a block instead.
"So the discount is the fee for size," Tao says.
"The discount is what closes the book," Eunha answers. "You are not paying for the shares at the screen price. You are being paid a little to absorb a quantity the screen could not absorb without moving."
One detail decides who ever sees that price. A book like this is built among institutions, and the shares are allocated by the banks running it. A retail account watching the same headline cannot put in an order at $95, which is why the deal arrives on your screen as a completed fact rather than an opportunity you declined.
That much is familiar to anyone who has sized a position against thin depth. Familiarity ends when the bids come in far above the ask. If you know equities mainly through tokenized stocks, this is the moment the two markets stop rhyming.
Put $100 billion of orders against a $15 billion deal and the surplus has to go somewhere. Tao's instinct sends it into the price. There are three doors it can actually go through, and the price is behind none of them.
The first door is a smaller allocation. Everyone who bid gets a fraction of what they asked for, the deal stays the size it was, and the excess simply evaporates as unfilled interest. The second door is a bigger deal: the company looks at the demand, decides it will take more money at this price, and prints more shares. The third door is both at once, which is the common outcome.
A higher price is not one of the doors, because the seller here is not a holder trying to get the best fill. The seller is the company, and the company is manufacturing the supply as it goes. When a seller can create the thing being bid for, competition among buyers stops being a price mechanism and becomes a quantity mechanism.
Tao pushes on it. "Then the company is choosing to leave money on the table."
Eunha does not disagree with the arithmetic, only with the framing. A raise is not an auction the issuer is trying to win. It is a transaction the issuer needs to complete, at a size it has already decided on, with buyers it may need again in two years. Squeezing the last two dollars out of a book is how a company discovers that the next book is thinner. The choice between rationing and upsizing gets made by the issuer, in a phone call, on the day, and it shapes the outcome more than the size of the book ever does.
She adds one caution about the headline number itself. A book five times covered describes how many shares were asked for at or above the price that was struck. It does not describe what anyone would have paid at a higher one. Coverage counts quantity at a level, so a large multiple tells you the deal was easy to fill, not that the stock was cheap.
"That is the part I was reading as a bidding war," Tao says.
"Secondary offering" hides a fork, and the fork decides whether you lose anything at all. Two different transactions travel under the same label. One of them changes the number under your position. The other leaves it exactly where it was.
When a company creates and sells new shares, that is technically a primary follow-on. This is the case where dilution is real. The share count rises, and every existing holder owns a slightly smaller slice of the same business. When existing shareholders sell shares they already own, that is a true secondary offering, and the share count does not move by a single unit. Ownership changes hands. Nothing is created, so nobody is diluted.
Retail search collapses both into "secondary offering," and the collapse matters, because the two transactions have opposite consequences for the number that sits under your position. Intel's deal is the first kind. New shares, new claims, a bigger denominator.
"So a lockup expiry is the second kind," Tao says.
"Close," Eunha says. "A lockup expiry does not even sell anything. It only makes existing shares eligible to be sold."
That distinction runs through the IPO lockup mechanism as well: float changes, share count does not. A follow-on is the case where the count itself moves, which is why it is the only one of the three that can dilute you.
That count moved in a filing, and the filing rewards close reading, because the precision is where the mechanism becomes visible.
Intel priced 210,526,315 shares at $95.00 on 10 August 2026, upsizing the deal from the announced $15 billion to $20 billion, for net proceeds of about $19.7 billion. Underwriters hold a 30-day option on a further 31,578,947 shares at the same price, and the offering closes on 12 August. It is Intel's first public sale of common stock since the company listed in 1971.
The $100 billion order book is Bloomberg's reporting, carried onward by outlets including CNBC. Intel's own release says nothing about the size of the book. It also says nothing about why the money is being raised. The stated use of proceeds is "general corporate purposes, which may include, but are not limited to, capital expenditures and working capital." That is the language a company uses when it is not committing to anything. Any AI or foundry story attached to this raise was not supplied by the issuer. It was imported by the reader.
So the sequence Tao found impossible is on the record: demand five times the deal, and the issuer answered it by selling a third more stock at a price below Monday's close.
Tao tries to reconcile the figures the intuitive way. The fall should be the discount plus a signal about what management thinks the shares are worth. Three numbers, added together. Watch what happens when you actually add them.
| The number | What it measures | Value |
|---|---|---|
| The discount | $95 against the prior close | 2.6% |
| The dilution | 210.5M new shares against ~4.3B outstanding | ~4.9% |
| The fall | INTC's close on 10 August | 4.1% |
The fall is larger than the discount and smaller than the dilution. It is not the sum of anything. Two of those three numbers cannot be added to reach the third, and anyone doing the addition ends up hunting for a missing 3% that was never there.
Read the two that do line up and the result is clean: the share count rose about 4.9%, and the market marked the equity down about 4.1%. That is not a verdict on strategy, but a proportion. The new claims got priced almost directly, with no panic premium attached. If the underwriters exercise their option, the dilution goes to about 5.6%.
Eunha makes Tao say the implication out loud, because it is the part that transfers. A move of that size, on that news, is the market doing arithmetic in public. It is not the market rejecting anything. That distinction is the same one running under a stock falling on good earnings, where a number that looks like disappointment is a number about something else entirely.
There is a second mechanism in the fall, and it works on the screen rather than on the share count.
Once $95 is public, it is available. Any institution that wants Intel stock can take it in the deal at $95 instead of lifting offers in the open market. Demand that would otherwise have hit the exchange gets absorbed by the offering, and the exchange price drifts toward the level where size is actually clearing. The offer does not get dragged up to the market. The market comes down to the offer.
That reversal is the part Tao had backwards. He had been reading the offer price as an output of the tape, when in the days around a deal this size it works as an input.
This is a different kind of price gravity from the one that governs the spread on a low-priced ETF. There, what you pay is set by the wrapper's own share price, not by any single reference level. Both are cases where the number on your screen is produced by structure the chart does not show.
One question is still sitting there, and Tao asks it as a matter of fairness. Existing holders had no chance to buy at $95. Should they not get first refusal on shares that dilute them?
In some markets they would. Rights issues in Europe and much of Asia hand existing holders a pre-emptive right, a claim on their pro-rata share of the new stock. A US follow-on of this type carries no such right. The shares go to whoever the banks allocated them to, and every holder who was not in the book is diluted without being consulted.
That shows up most clearly in the largest holder in the register. In August 2025 the US government took a 9.9% stake in Intel, buying 433.3 million shares at $20.47, a passive holding with no board seat and no governance rights. Those 433.3 million shares are still 433.3 million shares. The fraction of Intel they represent is smaller today than it was on Friday, because the denominator moved and the numerator did not.
No announcement was made about that. No filing marks it. Dilution is not something done to you in a transaction, but something that happens to a fraction while you hold it, through arithmetic performed elsewhere.
Tao gets to the version he can carry. The book being five times covered was never evidence about price. It was evidence about how much stock Intel could sell, which is a different question than what a share is worth, and the company answered the question it was actually asking.
The habit that survives this deal is small and mechanical. When a stock falls on a corporate action, find the share count before you reach for a reason. New shares over old shares gives you a number you can check in a filing, and it will usually account for more of the move than the story running above it. What is left over after the arithmetic is the part worth arguing about.
Arithmetic is easy on a stock you do not own. Open the simulator, hold a tokenized stock through its next headline drop, and see how much of the fall you could have accounted for without a narrative. Tao's mistake costs nothing there.
