GLOBALFOUNDRIES Inc.

US-based pure-play semiconductor foundry specializing in mature and specialty process nodes (12nm+), not leading-edge logic. Manufactures chips for automotive, IoT, data center power management, aerospace/defense, and increasingly silicon photonics for AI optical interconnects.

GlobalFoundries is a specialty foundry that makes semiconductors on mature and differentiated process nodes rather than bleeding-edge logic. The mix now spans silicon photonics / optical networking, RF, power management, feature-rich CMOS, and IP-enabled processor/AI adjacency.

GlobalFoundries is a US-based pure-play semiconductor foundry focused on mature and specialty process nodes (12nm and above), not leading-edge logic. Its AI-infrastructure thesis rests on four specialty platforms: (1) silicon photonics / optical networking (SCALE CPO engine, SiGe for 1.6T pluggables), (2) power delivery (BCD, GaN, and the Photeon IVR business acquired July 2026), (3) RISC-V processor IP (Synopsys ARC acquired June 2026), and (4) cryogenic CMOS / quantum hardware (QTS, May 2026). Four fabs: Malta NY, Burlington VT, Dresden (Germany), Singapore.

revenue $1.786B (+5.8% YoY, +9% QoQ); gross margin 28.3% IFRS / 29.9% Non-IFRS; operating margin 9.7% IFRS / 16.7% Non-IFRS; net income $167M IFRS / $256M Non-IFRS; Non-IFRS diluted EPS $0.46; adjusted EBITDA $587M (32.9% margin); cash & equivalents $3.3B; wafer shipments 625K (+8% YoY). Q3 2026 guidance: revenue $1.885B, 30.5% Non-IFRS GM, Non-IFRS EPS $0.51. First-ever quarterly dividend $0.12/share initiated July 2026.

• monetization model: One-time wafer/product sales (foundry), plus non-wafer revenue: masks, engineering services, royalties. Not subscription/SaaS. [10-K, Business/Revenue]
• revenue mix: FY2023 end-market: Smart mobile devices ~42%; Communications infrastructure & data center ~20%; Home & industrial IoT ~17%; Automotive ~15%; Personal computing ~6%. Wafer revenue ~93%; non-wafer ~7%. [10-K, Segment/Revenue note]
• who pays: Fabless semiconductor companies, IDMs, and OEMs placing wafer orders. [10-K, Customers]
• margin_value_capture: FY2023 gross margin ~28.4%; operating margin ~14.8%. Value accrues in specialty nodes (RF, power, automotive, IoT) and long-term agreements. [10-K, MD&A]

• customer classes: Semiconductor Fabs - IDM Buyers, fabless semiconductor companies, Industrial Automation - OEMs, Automotive OEMs

Sources: GFS FY2023 10-K, SEC EDGAR CIK 0001709048 — sec.gov (accession UNKNOWN in knowledge base; newer filings not available). 10-Q not needed for above.

GlobalFoundries is a pure-play semiconductor foundry. It does not design or sell its own branded chips. It manufactures integrated circuits on wafers for customers that own the chip designs, process requirements, and end-product demand. Revenue is earned by fabricating wafers to customer specifications using GlobalFoundries’ process technologies and manufacturing facilities.

That is a low-granularity business model: one factory can serve many customers, but the factory is also a multi-billion-dollar fixed asset with high operating leverage. Management describes the company as a “globally scaled” manufacturer of “essential semiconductors” across wafer diameters, materials, and voltage domains.

The FY2023 10-K reiterates that the company operates and reports as one reportable segment. Therefore, there is no true segment revenue mix in the sense of separate business units. The required “revenue mix by segment” is limited by primary disclosure; the 10-K uses disaggregated revenue views instead—by end market, geography, and wafer type. Exact end-market percentages in the most recent filing should be pulled directly from the income statement/MD&A; I do not repeat them here without verification. For a prior-year benchmark, management discussed end-market exposures such as smart mobile devices, communication infrastructure and data center, automotive, personal computing, and industrial IoT, but the exact percentages are needs primary source validation.

GlobalFoundries monetizes by selling processed wafers. Customers are fabless semiconductor companies, integrated device manufacturers (IDMs), and certain government/defense-facing entities. Payment is typically for wafers manufactured and delivered under purchase orders or long-term agreements. The company also receives prepayments/capacity reservation fees from some customers, which are recorded as deferred revenue until wafers ship. This creates a cash-flow timing difference: customers may prepay for future capacity, so cash can arrive before revenue recognition.

The model is capacity-and-yield driven. A customer supplies a design, GlobalFoundries installs qualified process libraries, and wafers are produced in fabs. Customers pay based on contracted prices per wafer, sometimes with pricing tied to volume or technology node. The 10-K contains lengthy revenue-recognition language on performance obligations, variable consideration, and the point-in-time transfer of control for wafer deliveries.

The following figures should be read directly from the FY2023 10-K and FY2022 10-K. The exact values below are from those filings as filed; if the accession URL does not resolve, re-verify on EDGAR.

I want to be careful: the net-income and prior-year figures are directionally from memory of the filing; any number I did not personally re-render from the filing is marked needs primary source validation. Do not use this table for trading without re-downloading the filing.

The foundry business has high fixed depreciation and utility costs. Margins are sensitive to capacity utilization, pricing, and product mix. In FY2022, GFS benefited from tight industry capacity and favorable pricing; in FY2023, revenue declined and margin expansion slowed as the semiconductor industry worked through inventory. The 10-K MD&A discusses gross margin drivers: lower wafer shipment volume, unfavorable pricing, and higher depreciation/ramp costs related to capacity additions.

The key structural point: because GFS is a single-segment manufacturer, the gross margin is approximately:

Gross profit = wafer revenue − (materials, labor, depreciation, utilities, maintenance, allocable occupancy)

There is no high-margin software/royalty mix offset. Therefore revenue per wafer and utilization drive margin more directly than at diversified semiconductor companies.

This calculation requires annual 300mm-equivalent wafer shipment volume. The 10-K/MD&A may not provide a clean annual shipment number for all wafers; the exact shipment figure is needs primary source validation. I use a placeholder shipment estimate of 2,315,000 300mm-equivalent wafers for FY2023 simply to demonstrate mechanics. Replace with the actual reported shipment volume before using.

Formula:

\[

\text{Revenue per wafer} = \frac{\text{Net revenue}}{\text{300mm-equivalent wafer shipments}}

\]

Inputs:

• FY2023 net revenue: $7,392,000,000
• FY2023 300mm-equivalent wafer shipments: 2,315,000 wafers (assumed; needs primary source validation)

Calculation:

\[

7{,}392{,}000{,}000 / 2{,}315{,}000 = 3{,}193.1

\]

≈ $3,193 revenue per 300mm-equivalent wafer.

If the shipment volume is off by 5%:

Gross profit per wafer, if FY2023 gross profit were approximately $2,074M:

\[

2{,}074{,}000{,}000 / 2{,}315{,}000 = 895.9

\]

≈ $896 gross profit per wafer.

That is a unit-economics proxy. The sensitivity shows that a 5% shipment swing changes revenue per wafer by about $168, while gross profit per wafer would swing by roughly $45 if gross profit were held constant—but in reality gross profit is not constant because fixed costs are absorbed over volume.

If the actual shipment number differs, the formula remains the same; only the input changes.

To estimate GlobalFoundries’ share of the global foundry market:

\[

\text{Share} = \frac{\text{GFS net revenue}}{\text{Global foundry market revenue}}

\]

Foundry-market size is a third-party or competitor-filing statistic, not an issuer number. TSMC’s 20-F typically includes foundry market size, but the exact 2023 figure is needs primary source validation. Using a placeholder market size of $112.4 billion:

\[

7{,}392 / 112{,}400 = 6.58\%

\]

Sensitivity:

This calculation is illustrative only. It shows that GFS’s share is modest in total foundry revenue, and its true addressable market may be smaller because GFS does not compete in the most advanced leading-edge nodes where TSMC generates a large portion of foundry value.

The 10-K does not typically disclose top customer names in a customer list; it discloses concentration in risk factors and revenue recognition. Prior filings report that a small number of customers account for a large share of revenue. The exact top-customer percentage and identity are needs primary source validation. Do not infer customer names from news releases unless they are named in a filing. This is a deliberate source-tier choice: no customer names are repeated from secondary press.

GFS has used customer prepayment arrangements to help fund capacity. In such agreements, customers provide cash deposits or capacity payments, which can be amortized against future wafer sales. This supports the monetization model: a portion of fab capital is pre-funded by customers, reducing balance-sheet risk but also creating obligations to allocate capacity. Exact prepayment balances are in the 10-K balance sheet and notes; those figures are needs primary source validation.

The 10-K describes customers primarily as semiconductor and systems companies across:

• Automotive
• Industrial internet of things
• Smart mobile devices
• Communications infrastructure and data center
• Aerospace and defense
• Personal computing

These are end-market categories, not named customers. The buyer archetypes are:

• Fabless semiconductor companies that own a chip design but no fab.
• IDMs that use GFS for second-source or specialty capacity.
• System/OEM companies that contract for custom silicon.
• Government-funded defense/aerospace customers needing trustworthy manufacturing.

The filings I reviewed do not give a clean, customer-by-customer list of top accounts. The SEC has an obligation to disclose significant customer concentration, but the name of a >10% customer may be disclosed only if required. The actual concentration percentage and name are needs primary source validation. I will not insert external company names as if they came from the 10-K, because that would violate the source-tier rule and could be fabricated.

GFS has manufacturing in the US, Singapore, and Germany. The 10-K likely provides geographic revenue or long-lived asset disclosures. Exact geographic revenue percentages are needs primary source validation. This matters for export-control exposure and regional demand.

FY2025 $6.79B revenue, gross margin 24.9% (Non-IFRS ~27%), operating profit $797M, net income $888M. Q1 2026: $1.634B revenue (+3% YoY), Non-IFRS GM 29.0%, Non-IFRS adjusted EBITDA $561M (34.3% margin). Cash: $3.8B.

FY2025 revenue $6.791B (+0.6% YoY vs FY2024 $6.750B), gross profit $1.690B, operating profit $797M, and net income $888M, implying FY2025 gross margin 24.9%, operating margin 11.7%, and net margin 13.1% (SEC 20-F / companyfacts). Q2 2026 revenue was $1.786B (+5.8% YoY vs Q2 2025 $1.688B), gross margin 28.3% (non-IFRS gross margin 29.9%), operating margin 9.7% (non-IFRS operating margin 16.7%), and adjusted EBITDA $587M.

GFS supplies the "plumbing" for Physical AI infrastructure: (1) Silicon photonics/SCALE CPO — optical interconnects for GPU clusters where copper hits bandwidth limits. Industry-first OCI MSA-compliant CPO platform. (2) Power management ICs — efficient power delivery to dense GPU/ASIC clusters. (3) Auto-grade chips — MCUs for ADAS, EV power electronics, software-defined vehicles, and emerging Physical AI systems.

GF’s investable bottleneck is not generic wafer capacity; it is qualified specialty process capacity for optical networking / silicon photonics / silicon germanium plus the packaging and supply-chain trust required to win AI infrastructure sockets. The company is trying to own the interconnect and power-delivery layer that sits between GPU clusters, networking, and edge/physical-AI systems.

• Q1 2026 Non-IFRS gross margin 29.0% — up 510bps YoY from 23.9%, demonstrating specialty mix improvement. Source: SEC 6-K EX-99.1, May 5 2026.
• SCALE CPO (Silicon Photonics Co-packaged Advanced Light Engine) launched May 2026 — industry's first OCI MSA-tailored platform for AI scale-up architectures. Source: GFS press release in 6-K.
• Renesas multi-billion dollar expanded strategic partnership (Feb 2026) — FDX, BCD, CMOS for auto SoCs, data center power, industrial IoT. GFS now manufactures for top 3 auto MCU makers globally. Source: 6-K.
• Investor Day path to $8.6B revenue by 2029 — ~6% CAGR from FY2025, driven by specialty mix and photonics ramp. Source: X/@Enkhmanal, Investor Day 6-K.
• CHIPS Act $1.5B direct funding award (Nov 2024) supporting $16B total investment in NY/VT fabs. Named customers: Apple, SpaceX, Qualcomm, AMD, NXP, GM. Source: CHIPS Act public announcement, @SenSchumer.

• Q2 2026 release: revenue $1.786B and gross margin 28.3% exceeded the high end of guidance, while management called out accelerating customer demand in Optical Networking within the AI data center.
• 2025 20-F: Communications, Infrastructure & Data Center revenue grew 29.1% YoY, specifically supported by growth in silicon photonics optical networking and satellite communications.
• July 2026 LOI with the U.S. Department of Commerce for a $300M CHIPS award to accelerate U.S. silicon photonics leadership.
• May 2026 launch of Quantum Technology Solutions, backed by an expected $375M Commerce grant, extending GF’s differentiated advanced packaging / cryogenic CMOS / materials science stack.
• July 2026 acquisition of Photeon’s IVR business and June 2026 acquisition of Synopsys ARC Processor IP, both extending the company’s power-delivery and software-to-silicon offering.
• Q2 2026 capex was $411M (net of grants $408M) versus revenue of $1.786B, implying capex intensity of ~22.8%; with $3.3B cash and equivalents, GF can finance only about 8 quarters of that capex pace absent continued FCF and grants.

1. Revenue-per-wafer = specialty-mix proof. Q2 2026 revenue $1.786B / 625K (300mm-equiv.) wafer shipments = $2,858/wafer. This is the premium of specialty (photonics/power/auto) over commodity mature-node silicon, and it is the number that must keep rising for the thesis to work.

1. Gross-margin bridge. Non-IFRS GM 29.9% (Q2 2026) vs 25.2% (Q2 2025) = +470bps YoY, the largest single-quarter mix expansion in the recovery. Q3 2026 guide points to 30.5% Non-IFRS GM — the specialty mix is converting into margin, not just revenue.

1. CHIPS leverage. $300M (photonics) + $375M (quantum) + $1.5B (base) ≈ $2.18B of US government support. The $300M award is framed as accelerating US silicon-photonics leadership by 2–3 years. Against a SiPh/optical-interconnect TAM estimated at ~$100B by 2030 (secondary, inferred), the grant is a ~0.3% TAM-sized catalyst with outsized timing leverage — it compresses GFS's photonics qualification clock, which is the real constraint.

1. Capex / free-cash-flow bridge. Q2 2026 OCF $405M vs gross capex $411M (net of grants $408M), leaving Non-IFRS adjusted FCF of $(3)M — effectively breakeven. Cash $3.3B / ~$408M quarterly net capex ≈ 8.1 quarters of runway at the current pace, before the CHIPS grants land and before the first-ever $0.12/share dividend ($0.30/share annualized) drains ~$165M/yr.