Skip to content
primarysourced Photonics sector GF
GFS
~8 min read · 1,764 words ·updated 2026-04-29 · confidence 64%

AMF Singapore acquisition (closed 2025-11-17)

Updated: 2026-04-29 Status: ✓ Verified via GF press release and AMF press release. Purchase price, employee count, and AMF customer list not publicly disclosed; flagged ⚠ where inferred. Cross-references: Fotonix process overview · Fotonix PDK customers · competitors · foundry relationships

1. Deal at a glance

FieldValueSource
Announced2025-11-17GF press release ✓
AcquirerGlobalFoundries Inc. (NASDAQ: GFS)GF press release ✓
TargetAdvanced Micro Foundry Pte Ltd, SingaporeGF press release ✓
Target heritageSpin-out of A*STAR / IME (Singapore government research agencies)AMF About page ✓
Purchase priceNot disclosed⚠
StructureNot disclosed (cash, stock, mix)⚠
ClosingConcurrent with announcement (acquisition described in present tense in 2025-11-18 globenewswire release)Globenewswire ✓
Effect on GF positionBecomes the largest pure-play silicon-photonics foundry by revenueGF press release ✓
Employee countNot disclosed⚠
Customer list inheritedNot publicly itemized⚠ — to verify in GF FY2025 10-K and follow-on disclosures

The 8-K filing accompanying the announcement is the primary SEC source; GF’s investor-relations page hosts the deal materials at investors.gf.com. The deal terms required no detailed Hart-Scott-Rodino-class disclosure to be public, suggesting either (a) the price was below the HSR threshold (currently ~$120M) or (b) the closing structure permitted private terms — ⚠ exact rationale not in the public 8-K text.

2. Why AMF — strategic rationale

2.1 Capacity and geographic diversification

AMF runs a 200mm silicon-photonics fab in Singapore, complementary to GF Fotonix’s 300mm Malta NY line. The structural value of the AMF acquisition for GF:

  1. Capacity addition: incremental SiPh wafer-output without building a greenfield 300mm SiPh line (SiPh-tooled 300mm capacity is ~$5-10B capex; AMF is a fraction of that).
  2. Geographic resilience: GF’s Fab 8 (NY) + Dresden + Burlington VT footprint adds an Asia-Pacific node that does not exist elsewhere in the GF asset base.
  3. Customer access: AMF’s existing Asia-Pacific customer base — which has historically included a mix of LiDAR, sensing, telecom, and datacom customers — adds revenue and design-win optionality.

2.2 Process / IP additions

AMF brings:

  • 15 years of silicon-photonics manufacturing expertise — one of the longest-running pure-play SiPh foundries globally
  • Two productized process tiers: AMF GP (general-purpose) and AMF HP (high-performance), with HP being the more advanced node (NLM Photonics SOH PIC release) ✓
  • Patent portfolio — inherited IP that strengthens GF’s photonics claims stack (see Patents overview)
  • Skilled photonics-engineering talent in Singapore that can staff the new Center-of-Excellence

2.3 Center-of-Excellence with A*STAR

GF announced a Silicon Photonics Center of Excellence (CoE) in Singapore that will partner with A*STAR (Agency for Science, Technology and Research) — Singapore’s lead public-sector R&D agency — focused on:

“next-generation materials for ultra-fast data transfer at 400 Gbps speed” (GF press release) ✓

Read pragmatically: the 400 Gb/s per lane target is a clear nod to EO-polymer (LWLG, NLM) and TFLN materials — the class of “next-generation materials” demonstrably capable of the V_π·L and bandwidth needed for 400 G/lane. The CoE creates a Singapore-based EO-polymer / TFLN co-development pipeline that complements the Malta-based Fotonix slot-waveguide-polymer integration already active with LWLG.

2.4 Tim Breen quote

GF CEO Tim Breen on the deal:

“Silicon photonics technology is essential for AI infrastructure.” ✓ (GF press release)

This is the simplest possible CEO framing — the acquisition is justified entirely by AI-infrastructure positioning, not by diversifying into telecom or LiDAR (where AMF’s historical revenue mix was weighted).

AMF CEO Jagadish CV’s reciprocal quote:

“We are proud to join forces with a trusted manufacturer with global reach.” ✓

3. AMF — what GF actually bought

3.1 Heritage

AMF was the first dedicated silicon-photonics foundry (AMF About page) ✓, spun out of:

  • A*STAR — Singapore’s national R&D agency
  • IME (Institute of Microelectronics) — A*STAR’s microelectronics research institute, where the SiPh process IP was originally developed in the late 2000s

The A*STAR / IME lineage is structurally similar to Imec (Belgium) and AIM Photonics (US) — government-seeded photonics R&D foundries that productized into commercial offerings. Among that cohort AMF is the only one to have been acquired by a Tier-1 commercial foundry.

3.2 Process portfolio

Process tierDescriptionPublic knowledge
AMF GPGeneral-purpose 200mm SiPh; lower-risk, broader photonic-device libraryConfirmed via NLM Photonics 2026-03-16 release ✓
AMF HPHigh-performance 200mm SiPh; advanced node where higher-bandwidth modulators and sub-wavelength gratings are validatedConfirmed via NLM 2026-03-16 release; HP PDK 6.0 in development ✓

3.3 200mm vs Fotonix 300mm

CapabilityAMF (200mm)GF Fotonix (300mm 45SPCLO)
Wafer size200 mm300 mm
Wafer-output economicsLower volume per wafer; more amenable to short runsHigher volume per wafer; better suited to high-volume merchant SiPh
CMOS integrationPhotonics-only (no monolithic CMOS); driver/TIA on separate dieMonolithic CMOS+photonics on same die
Process nodePhotonics-process-only (no transistor node)45 nm SOI CMOS + photonics co-fabricated
Customer fitSmaller-volume / specialty customers (LiDAR, sensing, niche datacom, R&D pilots)High-volume datacom and CPO
Future directionPlan to scale to 300mm “as market needs grow” per GF press release ✓Already 300mm

The two are complementary, not redundant. AMF lets GF serve customers who don’t need (or can’t afford) the monolithic 300mm CMOS+photonics path — and lets GF scoop up Asia-Pacific photonics customers without re-routing them through Malta NY.

3.4 300mm migration plan

GF stated AMF plans to “scale to 300mm as market needs grow” (GF press release) ✓. This is forward-looking and unspecified-timing — there is no announced 300mm AMF tool-in date. Given typical SiPh 300mm fab buildout (24-36 months from groundbreaking to first wafer) and GF’s existing 300mm Fab 8 capacity, the practical interpretation is:

  • 200mm AMF stays at 200mm for the foreseeable future
  • The “scale to 300mm” framing keeps the option live without committing capex
  • If a major customer (Marvell, NVIDIA, Microsoft, Meta) commits AMF-Singapore-specific 300mm volume, the upgrade trigger would justify the investment

⚠ No primary-source 300mm AMF capex disclosure exists as of this audit date.

4. Cross-company implications

4.1 POET Technologies (POET) — collapsed AMF triangulation

The POET KB (foundry relationships) explicitly states POET selected SilTerra Malaysia over GlobalFoundries Fotonix and Advanced Micro Foundry as their silicon foundry. The reasoning included process-customization access, capacity availability, and “China Plus One” geographic positioning.

The November 2025 GF-AMF acquisition collapses one branch of POET’s prior optionality:

  • POET cannot now treat AMF as an independent alternative SiPh foundry — it’s now a GF subsidiary
  • Any process-IP that POET historically considered for AMF is now under GF control
  • The “China Plus One” geographic posture POET maintained via SilTerra is unaffected, but the second non-China SiPh foundry option (AMF) is no longer independent

This is a structural shift, not a thesis-breaker for POET — POET’s primary fab relationship is with SilTerra, which is unaffected. But it removes a hedge / second-source option and makes GF a more dominant single-counterparty risk for the broader Asia-Pacific SiPh fab landscape.

4.2 LWLG — capacity and PDK alignment

Per the LWLG KB (platform overview §Foundry-PDK status), LWLG’s prior AMF agreement (May 13 2024) demonstrated sub-1V drive at 200 Gbps PAM4. The November 2025 GF-AMF acquisition consolidates that integration into GF’s portfolio:

  • The LWLG-AMF technical proof remains valid
  • The commercial relationship now flows through GF, simplifying procurement and harmonizing PDK templates
  • LWLG’s GF Fotonix (Malta) PDK live (Mar 2026) and the AMF process (Singapore) are now part of the same foundry’s portfolio — strategically positive for LWLG, as it consolidates two independent integrations into one supplier relationship

4.3 NLM Photonics — first commercial deployment on AMF HP

The fact that NLM Photonics’ first sampling product (announced 2026-03-16) is on the AMF HP process rather than on Fotonix 45SPCLO is itself a tell:

  • NLM was an AMF customer before the GF acquisition
  • Post-acquisition, NLM is now de facto a GF customer
  • The HP PDK 6.0 development under GF management is the path to NLM’s second-generation product

NLM benefits from the consolidation because it gets access to GF’s broader EDA / packaging / customer-engineering resources without changing its already-validated AMF process recipe.

4.4 Tower Semiconductor — competitive pressure

Tower’s response to the GF-AMF acquisition was within four weeks: Tower announced CPO foundry technology on its 300mm SiPh + EIC platforms on 2025-11-12 (Tower press release) ✓ — a defensive move signalling Tower will not cede merchant SiPh foundry leadership. The GF-AMF combination + Tower 300mm port + Xscape Photonics on-chip laser (Tower 2025-08-25) ✓ now represents the live competitive set in 200mm-and-300mm SiPh foundry.

5. Open audit items

  1. ⚠ Purchase price — not disclosed; verify in GF FY2025 10-K (expected filed Feb-Mar 2026).
  2. ⚠ Employee count transferred — likely 100-300 based on AMF’s historical scale; verify against GF 10-K Item 1 Business or 8-K supplementary materials.
  3. ⚠ Inherited customer list — AMF’s pre-acquisition customer book has not been itemized.
  4. ⚠ 300mm AMF buildout — no announced capex / timeline.
  5. ⚠ Integration timeline for AMF HP PDK 6.0 — NLM Photonics has confirmed the PDK is in development but no GA date has been disclosed.
  6. ⚠ A*STAR partnership financial structure — research-funding split not public; likely involves Singapore EDB grants but specifics not disclosed.

6. Cross-references

Sources