Intel announced a $15 billion underwritten public offering of common stock before the bell Monday. Shares fell about 5% to roughly $97 in early trading — and almost nothing else in the chip complex moved. AMD slipped 1%. Nvidia and Broadcom were essentially flat. The iShares Semiconductor ETF didn’t budge. The market read this for exactly what it is: a capital-structure event at one company, not a verdict on AI demand.
The mechanics are plain. Underwriters get a 30-day option on up to $2.25 billion of additional shares. J.P. Morgan, Goldman Sachs, Morgan Stanley and Citigroup are joint book-running managers. Intel says the proceeds go to “general corporate purposes,” then spells out what that means: capital expenditures and working capital. At around $100 a share, $15 billion is roughly 150 million new shares — call it 3% dilution against a share count that was already growing.
So the market charged Intel 5% for 3% of dilution. Slightly punitive, entirely ordinary. The number that actually matters is the one nobody put in the headline: Intel is up about 175% year to date through Friday’s close. Twelve months ago this company was raising cash by selling stakes in its own subsidiaries. Today it is selling equity at a price the market handed it after two straight quarters of proof.
What the money is actually for
Intel lifted its 2026 capital-expenditure outlook to $20 billion from $18 billion in July, committed to high-volume 14A production in 2028, and signed Tesla as a 14A foundry customer. Cleanrooms, EUV lithography tools and advanced packaging capacity are lumpy, multi-year and cash-hungry in a way that quarterly operating cash flow does not smoothly cover. You either pre-fund the buildout or you arrive late to it.
The demand story underwriting the roadshow is real and recent. Second-quarter revenue came in at $16.13 billion, up 25.4% year over year — Intel’s fastest growth since 2011 — with Data Center and AI up 59%. Third-quarter guidance of $15.8 billion to $16.8 billion sits above where the Street had been. That is a chart a banker can sell.
There is a debt angle too, and Intel said the quiet part in its own release: the offering is intended to fund growth “while maintaining a strong balance sheet and its commitment to an investment-grade rating.” Equity dilutes. It also carries no coupon and no covenants. For a company whose credit profile was the loudest bear argument two years ago, that is the trade it wants to make — and the same trade CoreWeave has been forced to make in a harder market.
Our take: Last quarter Intel booked an $11 billion GAAP loss that was mostly a mark-to-market charge on the government’s escrowed stake — a liability that swells every time the stock rises. Today that same rally bought $15 billion of fab capacity in cash. One rally, two headlines that look like opposites. Only one of them clears. Selling stock when it’s expensive is what a turnaround is supposed to do; the companies you worry about are the ones issuing equity at the lows because nobody will lend to them.
What to watch
- Pricing terms, likely within 24 hours. Deals this size price fast. The discount to Monday’s open is the cleanest read on how deep the book really was.
- The greenshoe. $15 billion plus $2.25 billion is the ceiling. Whether underwriters exercise that option is a live measure of institutional appetite, not a press release.
- More 14A wins. Tesla is one customer. A 2028 node funded in 2026 needs a longer list before the capex looks committed rather than speculative.
- Capex creep. $20 billion is this year’s number, revised up once already. Watch whether the January guide starts with a 2 — and whether Intel funds the next increment with equity again.
The bet Intel is making is that AI compute demand stays strong enough for long enough that 150 million extra shares look cheap in 2029. The bet shareholders made this morning, by selling, is that they’d rather not underwrite it at $100. Both sides have been wrong about this stock before.