Analysis
Intel priced an upsized public stock offering, confirmed in Intel's own press release and reported by CNBC:
- Price and size -- $95 a share, 210,526,315 shares sold, $20 billion in gross proceeds, up from the $15 billion deal floated just days earlier.
- Net proceeds -- roughly $19.7 billion once the sale closes on August 12, 2026.
- Greenshoe option -- underwriters hold a 30-day option to buy up to 31.6 million more shares, which could push the total higher still.
- Demand -- order books reportedly topped $100 billion, more than five times what Intel ultimately sold, which is what pushed bankers to upsize mid-process rather than leave demand on the table.
Intel's own filing says proceeds go toward general corporate purposes, which may include capital expenditures and working capital -- corporate-speak for funding the fab buildout at the center of its turnaround. Pulse has covered Intel's AI-era turnaround repeatedly; see the full Intel coverage archive for prior stories on its CHIPS Act stake, foundry backlog and earnings.
“- Price and size -- $95 a share, 210,526,315 shares sold, $20 billion in gross proceeds, up from the $15 billion deal floated just days earlier.”
How Intel got here
This raise doesn't happen in a vacuum. Intel beat earnings estimates earlier this year on AI data-center demand, only to see the stock crater on capex worries -- the market has repeatedly punished Intel for spending big on foundry capacity before it has proven it can win leading-edge customers at scale. The US government also took an equity stake of roughly 10% in Intel in 2025, converting CHIPS Act support into direct ownership, which means Washington now has skin in the game every time Intel goes back to capital markets. Intel Foundry's backlog has grown to roughly $15 billion, but independent assessments still place Intel and Samsung two to three years behind TSMC in production volume and yield for AI accelerator workloads.
The competitive landscape
The foundry market Intel is fighting for is growing fast and getting more concentrated at the top:
- TSMC -- record Q2 2026 revenue of $40.2 billion, up 36% year over year; raised its full-year AI chip growth outlook above 40%; now controls roughly 60% of the global foundry market and more than 90% of leading-edge production.
- Samsung -- planning more than $73 billion in 2026 capex and R&D as it works to recover foundry share, the other credible leading-edge alternative to TSMC.
- Global foundry market -- projected to grow nearly 25% in 2026 to about $218.8 billion, a pie that's expanding but one where Intel is still the smallest of the three credible leading-edge players.
- AMD and Nvidia -- remain the AI chip customers everyone wants to win, and neither has committed its most advanced designs to Intel's process nodes at any meaningful volume yet.
Numbers in context
$20 billion is a big number for Intel in isolation, but it needs a scale check against the rest of the capital stack around it:
- TSMC capex -- will spend more in roughly six months than Intel just raised in this entire offering.
- Samsung capex/R&D plan -- more than 3.5x the size of this raise in a single year.
- Net proceeds vs. original plan -- $19.7 billion in net proceeds, larger than the $15 billion Intel had originally planned to raise.
- Dilution -- 210.5 million new shares, a mid-single-digit hit for existing holders, real but not crushing against the capital raised.
That proceeds figure also arrives on top of the government's direct equity position, CHIPS Act grants, and Intel's existing foundry backlog.
What it means for founders, GPs and LPs
For anyone underwriting AI infrastructure exposure, this is a reminder that public equity markets are still willing to fund capex-heavy, pre-profitability chip bets when the AI narrative is attached -- the same dynamic that has kept CoreWeave, Oracle and Nvidia's financing partners flush with capital this year. It's also a signal that turnaround stories in hard tech can raise growth capital even with a volatile stock, as long as the AI capex thesis holds. LPs watching semiconductor and AI-infra allocations should treat this as one more data point that the capital is there -- the harder question is execution.
The bear case and what the headline misses
The headline number is $20 billion raised -- what it doesn't tell you is how much of that actually goes to leading-edge foundry capacity versus general corporate needs, or whether Intel can convert its $15 billion backlog into volume production that competes with TSMC on yield. Announced capital is not deployed capital, and Intel has a track record of capex commitments outpacing execution timelines. The stock also dipped in premarket trading on news of the upsize, a reminder that dilution has a real, immediate cost even when the capital-raising story is framed as strength. If AI capex growth decelerates in 2027, a foundry business still years behind TSMC on the metrics that matter is exposed.
What to watch: whether Intel names new external foundry customers at scale before year-end, how the government's equity stake behaves through this dilution event, and whether Q3 earnings show the capex-to-revenue ratio improving or still widening.
