The Key Distinction
Intel has always manufactured chips.
That does not mean Intel has always operated like a modern external foundry.
A captive manufacturer builds primarily for its own product teams.
A foundry builds for customers that may compete with one another — and sometimes with the foundry owner's own products.
That changes the business requirements.
The Foundry Customer Contract
An external chip designer needs confidence that the foundry will provide:
stable process design kit
reliable IP
EDA tool support
capacity
yield ramp
confidentiality
schedule predictability
commercial neutrality
advanced packagingFoundry credibility is therefore partly technical and partly institutional.
What Intel Foundry Sells
Process technology
Customers can design chips for Intel manufacturing nodes.
Wafer fabrication
Intel manufactures customer designs in its fabs.
Advanced packaging
Intel can integrate multiple dies using technologies such as EMIB and Foveros.
Assembly and test
Intel offers back-end manufacturing capabilities.
Design enablement
Customers need EDA flows, IP, libraries and design rules to successfully tape out.
Why the Accounting Is Confusing
Intel reports Intel Foundry as an operating segment.
The segment charges Intel Products for manufacturing.
Those transactions are eliminated at the consolidated level.
2025:
Intel Foundry segment revenue $17.826B
external third-party foundry / A&T $0.307BThe difference is enormous.
So the headline segment revenue should never be described as customer foundry sales.
Internal Volume as a Foundry Advantage
Intel's own products can provide base manufacturing volume.
That can help a new process ramp before external customer volume becomes large.
Core / Xeon volume
↓
factory utilization
↓
yield learning
↓
process maturity
↓
external customer confidenceThis is a potential structural advantage over a foundry with no anchor products.
Internal Volume as a Foundry Conflict
The same model creates concerns.
External customers may ask:
- Will Intel Products receive priority?
- Can Intel protect confidential design information?
- Will foundry economics be transparent?
- Can Intel operate with customer-first neutrality?
These are business-model questions, not only process-technology questions.
Intel 18A
Intel 18A is the current leading-edge foundation of the foundry strategy.
It entered production in 2025. [S16]
18A-P entered risk production in 2026. [S16]
The family is intended to create a longer-lived platform rather than a one-time node.
14A
Intel 14A is the planned successor family.
Under Lip-Bu Tan, Intel has explicitly tied investment in 14A to confirmed customer commitments. [S6]
This is an important change in capital discipline.
The company is effectively saying:
technology ambition
must be matched by
customer economicsSystems Foundry
Intel uses the phrase “systems foundry.”
The idea is that advanced AI products require more than one wafer process.
A customer may need:
logic die
HBM integration
chiplets
packaging
assembly
power delivery
IP
EDAIntel wants to sell the full system-manufacturing stack rather than only wafer fabrication. [S15]
Advanced Packaging Advantage
Packaging is especially important because it can create customer value even if some dies are manufactured elsewhere.
That broadens Intel Foundry's addressable role.
A customer could theoretically use:
external logic die
+
Intel packaging
+
Intel assembly/testwithout moving every wafer to Intel.
Foundry Customers
Intel has disclosed customer activity across multiple nodes and packaging services.
Microsoft publicly said in 2024 that it planned to produce a chip design on Intel 18A. [S34]
Government programs have also used Intel's manufacturing ecosystem for trusted domestic semiconductor development. [S36]
The strategic question is whether those engagements convert into large, durable external production volumes.
Why Foundry Is Hard
Foundry economics require simultaneously:
- leading technology;
- high yields;
- high utilization;
- large customers;
- predictable roadmaps;
- huge capital spending.
Failure in one area can weaken the others.
What Success Would Look Like
Intel Foundry succeeds if the company can move from:
mostly internal manufacturerto:
internal manufacturer
+
meaningful external foundry
+
advanced packaging platformwithout damaging Intel Products.
Common Misconceptions
“Intel Foundry is already a TSMC-sized external business.”
No.
External revenue remains small compared with internal segment activity. [S1]
“Foundry success depends only on having the best node.”
No.
Customers also require design enablement, capacity, confidentiality, yield, ecosystem support and schedule reliability.
“Intel must manufacture every chip in a customer package.”
No.
Advanced packaging can integrate dies manufactured on different nodes and potentially by different foundries.