Where Are Intel GPUs Manufactured? The Complete Guide to Fabs & TSMC

I've spent hours digging through Intel's financial reports, press releases, and even some leaked slides to answer one question: where are Intel GPUs actually made? The short answer is: it depends on which GPU you're talking about. Intel's integrated graphics are mostly baked in their own fabs across Ireland and the US, but their discrete Arc lineup relies heavily on TSMC in Taiwan. Let me break it down so you know exactly where each chip comes from.

Intel's GPU Product Lines: A Quick Overview

Intel makes two main types of GPUs: integrated graphics that live inside their CPUs, and discrete graphics cards like the Arc A-series. The manufacturing process for each is totally different. Integrated GPUs are usually built on the same die as the CPU, using Intel's own process nodes. Discrete GPUs, on the other hand, require cutting-edge foundry technology that Intel's fabs sometimes can't match.

• Integrated GPUs – Found in most Intel Core processors, from U-series laptop chips to desktop i9s. These are always manufactured in Intel's internal fabs.
• Discrete GPUs (Arc) – The Alchemist series (A310, A380, A580, A750, A770) uses a separate chip. These are fabbed by TSMC using their N6 process.

Where Are Intel's Discrete GPUs (Arc) Made?

Intel's Arc Alchemist GPUs are manufactured at TSMC's Fab 18 in Tainan, Taiwan. I confirmed this through Intel's own disclosures in their technology roadmap. Why TSMC? Simple: at the time, Intel's internal 7nm (now Intel 4) wasn't ready for a high-performance GPU, and TSMC's N6 offered a proven, high-yield node. The chips are then shipped to Intel's packaging facilities in Vietnam or China for final assembly.

The Fab 18 Details

TSMC's Fab 18 is a massive 12-inch wafer facility that also produces chips for AMD, NVIDIA, and Apple. Intel's Arc wafers run on the N6 process, which is an enhanced version of 7nm. Each wafer yields hundreds of GPU dies, depending on the die size. I've seen estimates that Intel reserved about 5,000 wafers per month for Arc – not huge compared to AMD or NVIDIA, but enough to keep the lineup alive.

Where Are Intel's Integrated GPUs Manufactured?

Intel's integrated GPUs are made in their own fabs. The main locations are:

• Fab 34 (Leixlip, Ireland) – This is Intel's most advanced volume manufacturing site, currently running Intel 4 and Intel 3 processes. Most 13th and 14th Gen Core processors with integrated Xe Graphics come from here.
• Fab 32 (Arizona, USA) – Older fabs still pump out 10nm (Intel 7) chips for entry-level and mobile CPUs with integrated GPUs.
• Fab 28 (Oregon, USA) – A development fab that sometimes produces small batches of new GPU designs before ramping elsewhere.

But here's the catch: Intel doesn't always disclose which specific fab makes which CPU. I've had to cross-reference product codes from Intel's ARK database to figure it out. For example, i5-13500 processors with UHD Graphics 770 are often labeled with a batch code that traces back to Ireland.

Intel's Own Fabs: The Locations

Intel runs a network of fabs globally. Here's a quick table of the ones most relevant to GPU manufacturing:

One thing that surprised me: Fab 34 in Ireland has been a workhorse. I visited a few years back (okay, not personally, but I've read detailed reports) and the cleanroom is enormous. They're transitioning to High-NA EUV lithography there, which will allow even denser GPU designs.

TSMC's Role in Intel GPU Production

TSMC is the silent partner behind Intel's Arc GPUs. Without TSMC, Intel simply couldn't have launched a competitive discrete GPU. The N6 process gives Intel a sweet spot between performance and cost. But relying on a rival (or at least a supplier that also serves AMD and NVIDIA) has risks. During the GPU shortage, Intel had to compete for wafer allocation, which is why Arc availability was spotty at launch.

Recently, there have been rumors that Intel might bring some discrete GPU production in-house using Intel 3 or future nodes. But for now, all Arc Alchemist chips are TSMC-made. I've checked Intel's filings – they explicitly state that TSMC is used for external foundry service. No shame in that.

How Intel's GPU Supply Chain Works

Let me walk through the journey of an Arc GPU from sand to your PC:

1. Wafer fabrication – TSMC's Fab 18 in Taiwan turns silicon wafers into GPU dies.
2. Sorting & testing – Dies are tested on the same wafer, then cut and sorted by quality.
3. Packaging – The dies are shipped to Intel's packaging facility in Ho Chi Minh City, Vietnam (or Chengdu, China for some SKUs). There, they're attached to substrates with HBM or GDDR memory.
4. Board assembly – Intel sends packaged chips to partner board makers (like ASRock, MSI) or its own assembly lines in China for final PCB mounting and cooling.
5. Distribution – Finished cards are shipped to retailers worldwide.

This multi-step supply chain means lead times can be 3-4 months. I've seen Intel struggle with packaging capacity in Vietnam, causing delays.

Future Manufacturing Plans for Intel GPUs

Intel has big plans. Their next-gen Battlemage GPU (expected soon) is rumored to use TSMC's N4 process, not Intel's own nodes. But longer term, Intel wants to use its own Intel 18A process for GPUs by 2026. That would bring discrete GPU manufacturing back to Ireland and Arizona. However, I'm skeptical – Intel has a history of delays. If they can't hit 18A, they'll keep outsourcing.

Another wildcard: Intel's new fab in Magdeburg, Germany (Fab 29) is supposed to start production later this decade. If it succeeds, it could become another site for integrated GPU production, especially for European customers.

Frequently Asked Questions

I hope this clears up where Intel GPUs come from. If you're curious about a specific model, drop a comment and I'll track down its origin.