Skip to content
primarysourced Photonics sector GF
GFS
~4 min read · 879 words ·updated 2026-04-29 · confidence 39%

GF diversified customers (cross-context only)

Updated: 2026-04-29 Status: ◐ This is a secondary-scope file. The KB’s primary thesis is the photonics layer; this file provides 10-K-level context on the rest of GF’s customer concentration. Cross-references: Customers — photonics (primary scope) · overview · overview

1. Scope discipline

This file is intentionally compact. The full diversified-customer narrative — automotive design wins, RF / 5G mmWave program by program, IoT and smart-home customer pipelines, defense and aerospace specialty volume — is appropriate for a generalist semis hub but not for a photonics-research site.

The reason this file exists at all: GF’s photonics business is funded by the diversified customer base. Photonics R&D capex (Fotonix process refinement, AMF integration, A*STAR CoE, slot-waveguide modifications) is paid for by gross profit from automotive, RF, IoT, and smart-mobile wafers. Understanding how robust that funding base is matters for the photonics thesis.

2. End-market segmentation (per GF 10-K)

GF organizes its customer-revenue disclosure around five end markets (per GF 10-K Item 1 Business ✓):

End market% of revenue (rough)Representative customers
Smart mobile devices~30-35%Qualcomm, MediaTek, Apple (modems / RF)
Communications infrastructure & datacenter~25-30%NXP, Infineon, AMD (DPU), Marvell, hyperscaler-direct programs (low share but growing); Photonics customers (LWLG, NLM, Marvell, Ayar Labs, Lightmatter, etc.) sit inside this segment
Automotive~15-20%NXP, Infineon, Renesas, STMicroelectronics, AutoTalks, automotive-grade specialty SoCs
Industrial / IoT~10-15%Cirrus Logic, Murata, NXP, Microchip, multiple
Aerospace / defense / other~5-10%Defense primes, satellite communications, government programs

⚠ Exact percentages vary year over year and across reporting period; see latest GF 10-K for definitive numbers.

3. Top customer concentration (10-K disclosure)

GF’s 10-K typically discloses customers above 10% of revenue. Recent disclosures have included (verify in latest 10-K):

  • AMD — historic relationship from 2009 spin-out; long-term wafer-supply agreement (LTSA) in place; GF still fabs some AMD legacy products
  • Qualcomm — long-term capacity commitment for RF / smart-mobile wafer volume
  • NXP — automotive and industrial wafer commitments

⚠ Specific 10% thresholds and customer naming should be re-validated against the FY2025 10-K (expected filed February 2026 with a fiscal-year-end Dec 31 2025 reporting period — GF reports FY on calendar basis).

4. Process-portfolio mapping to customer types

ProcessEnd-market customers
22FDX (FD-SOI)Smart mobile (RF), automotive radar, IoT — Qualcomm, MediaTek, automotive Tier-1s
12LP / 12LP+ (FinFET)High-performance SoCs — AMD legacy, networking, some AI
45RFSOImmWave RF, 5G front-end modules, satellite — Qualcomm, multiple RF-FEM suppliers
130 BCDPower management, automotive, industrial — multiple
90 SiGe BiCMOSHigh-speed mixed-signal, automotive radar, optical-comm drivers (overlaps with photonics)
45SPCLO (Fotonix)Silicon photonics — see Customers — photonics
9WG (Fotonix variant)Silicon photonics — cost-down variant — Lightmatter, others
AMF GP / HPSilicon photonics — 200mm Singapore — NLM Photonics, inherited AMF customer book

The photonics processes (45SPCLO, 9WG, AMF GP, AMF HP) sit alongside — not separate from — the diversified specialty-process portfolio. Fab 8 in Malta NY runs Fotonix wafers on the same site as 12LP, 22FDX, 45RFSOI, 90 SiGe BiCMOS — sharing tools, infrastructure, and overhead.

5. Long-term wafer-supply agreements (LTSAs)

GF disclosed in its IPO S-1 and subsequent 10-K filings that a substantial portion of revenue is committed under multi-year LTSAs. Examples include:

  • AMD LTSA — original 2019-2024 agreement, extended through 2025 and beyond per public disclosures
  • Qualcomm LTSA — multi-year commitment for RF / smart-mobile capacity
  • Multiple automotive Tier-1 LTSAs — particularly post-2021 chip shortage when automotive customers locked in capacity
  • U.S. DoD / federal program LTSAs — defense and aerospace volume tied to specific qualifications

⚠ Specific dollar value, term length, and customer-by-customer breakdown should be sourced from the latest GF 10-K and 10-Q.

The relevance to the photonics thesis: LTSA-backed revenue is a stable funding source for photonics R&D and capex. Mubadala’s 77.05% ownership (post the March 2026 $840M secondary, down from ~89% pre-IPO) amplifies this — the company can underinvest in short-term margin to fund long-term R&D in a way a pure-public-comparable might not tolerate.

6. Open audit items

  1. ⚠ Latest 10-K customer concentration — re-validate against FY2025 10-K (expected Feb-Mar 2026 filing).
  2. ⚠ End-market revenue mix percentages — refresh quarterly from 10-Q disclosures.
  3. ⚠ AMD LTSA status — historically the largest single customer; current contract terms.
  4. ⚠ Mubadala-related-party transactions — Mubadala’s separate enterprise customers may also be GF customers; check 10-K related-party disclosure.
  5. ⚠ CHIPS Act customer-allocated capacity — GF received CHIPS Act funding tied to specific customer commitments; allocation specifics.

7. Cross-references

Sources