GFS Open Research Questions

This file enumerates the questions about GlobalFoundries that the company has not publicly answered as of 2026-04-29 — questions that materially affect the bull / bear thesis but cannot be resolved from primary-source materials at this date. Each question gets:

The research path: where the answer would surface (forward filings, industry conferences, partner disclosures, etc.)
Current best-guess with explicit confidence flag (◐ partial / ⚠ inferred / ✓ corroborated)
Thesis-impact mapping to bull / bear pillars and risk IDs

The list is prioritized roughly by thesis-significance: highest-impact unanswered questions first. Where the analyst genuinely does not have an answer, the entry says so explicitly — speculation is flagged inline rather than substituted for primary-source verification.

Q1. What was the financial cost of the AMF (Advanced Micro Foundry) acquisition?

Research path: (a) FY 2025 10-K acquisitions footnote (target filing window mid-Feb 2026 — already filed, but accession + acquisition-cost line item needs verification); (b) GF 8-K acquisition disclosure (typically filed within 4 business days of closing); (c) any post-close reconciliation of goodwill + intangibles allocations.

Current best-guess ⚠ inferred (no primary-source dollar amount available): The Nov 17 2025 GF press release (GF press release ✓) stated “the company did not disclose the financial terms of the transaction.” Industry-typical valuation for a private SiPh foundry of AMF’s scale (200mm capacity, ~15 years of operating history, A*STAR spin-out, single-process specialization): analyst estimate $200-500M range, but this is a wide-band inference. The 10-K acquisitions footnote should disclose the gross purchase consideration plus working-capital adjustments + earn-out structure if any.

Thesis impact: bull case Pillar 2 (consolidation thesis) + risks K3, K4. Material for assessing capex-cycle pressure (Pillar 4 bear) and goodwill-impairment risk if AMF underperforms post-close revenue expectations.

Q2. What was the InfiniLink acquisition price?

Research path: Same as Q1 — FY 2025 10-K acquisitions footnote, possibly an 8-K disclosure if the transaction crossed materiality thresholds.

Current best-guess ⚠ inferred (no primary-source dollar amount available): InfiniLink was a Cairo-based silicon-photonics startup founded 2020, focused on optical transceiver chipsets + CPO design IP (Arab Founders ✓). Industry-typical valuation for a 5-year-old fabless photonics-design startup with no disclosed product revenue: analyst estimate $30-100M range based on team-and-IP-acquisition comp set. The deal is structurally a tuck-in, not a strategic-scale acquisition.

Thesis impact: bull case Pillar 2; secondary to AMF in cost terms but material for the design-IP-enrichment pillar argument.

Q3. What are Mubadala’s lock-up terms post the March 2026 secondary offering?

Research path: (a) The S-3 / 424B7 prospectus filings tied to the March 11 2026 secondary offering should disclose lock-up periods; (b) the related 8-K announcement of the offering closing (Globe and Mail 2026 ✓); (c) the underwriter agreement (Cleary Gottlieb is on record as Mubadala’s counsel for the 2026 deal — agreement would be exhibited in the secondary-offering registration statement).

Current best-guess ◐ partial: Industry-typical lock-ups for selling-shareholder secondary offerings of this size run 90 to 180 days. With a March 11 2026 launch + March 13 2026 closing, the implied lock-up window expires somewhere between mid-June 2026 and mid-September 2026. The next selldown event would be possible from mid-September 2026 forward, though the cadence-pattern from May 2024 (the prior selldown) implies the actual timing is more likely in the 12-18 month range = March 2027 - September 2027 rather than the immediate post-lockup window.

Confidence: ◐ partial — primary-source verification (the lock-up term sheet within the S-3 / underwriter agreement) is pending.

Thesis impact: bear case Pillar 1 (Mubadala overhang) + risks K1. Critical for forward-catalyst dating; the next Mubadala selldown event is one of the most material recurring catalysts on the calendar.

Q4. What is GF’s specific Fotonix wafer-volume forecast for 2026, 2027, and 2028?

Research path: (a) Investor Day 2026 (May 7 2026) — most likely venue for unit-volume guidance; (b) FY 2026 10-K segment-revenue disclosures (would imply wafer volumes if combined with disclosed ASP); (c) industry analyst forecast updates (LightCounting, Dell’Oro, 650 Group); (d) supply-chain triangulation via SOI-wafer purchases from Soitec.

Current best-guess ⚠ inferred: GF has not publicly disclosed Fotonix wafer-volume forecasts. Analyst-derived approximation:

FY 2025 SiPh revenue ~$200M
Estimated 300mm SiPh wafer ASP: $5,000-10,000 per wafer (analyst estimate, ⚠ inferred)
Implied FY 2025 Fotonix wafer volume: 20K-40K wafers (analyst estimate, ⚠ inferred)
FY 2028 implied at $1B run-rate + ASP stable: 100K-200K wafers (analyst estimate, ⚠ inferred)

Confidence: ⚠ inferred — primary-source verification not available. The implied wafer-volume range is consistent with industry-survey reports of GF’s SiPh capacity.

Thesis impact: bull case Pillar 3 + Pillar 4. Bridge between the revenue trajectory and the capacity-deployment cycle.

Q5. What is the LWLG commercial PDK ramp curve at GF Fotonix?

Research path: (a) LWLG’s 10-Q + 10-K disclosures for FY 2026-2028; (b) GF customer-side commentary at Investor Day or industry conferences naming LWLG; (c) joint-press-release events (e.g., the Mar 16 2026 PDK live announcement was the first such event); (d) PDK download / customer-tape-out disclosures from LWLG IR.

Current best-guess ◐ partial: LWLG’s commercial PDK at GF has been live since March 16 2026 (per universe.json ✓ and LWLG bull case ✓). Industry-typical commercial-PDK-to-customer-tape-out cycle: 6-12 months. Industry-typical first-customer-product-revenue cycle: 12-18 months from PDK-go-live = March 2027 to September 2027 for first product PR. Volume-revenue ramp would follow 12-24 months later = 2028-2029. LWLG-attributable revenue contribution to GF’s SiPh segment is structurally small in the near term (LWLG’s chromophore is a process-flow input rather than a wafer-replacement) but grows non-linearly with customer-product-revenue ramp.

Confidence: ◐ partial — directional inference, primary-source customer-tape-out disclosures pending.

Thesis impact: bull case Pillar 1 (customer roster) + cross thesis implications (LWLG-GFS coupling).

Q6. What is GFS’s specific SiPh ASP per wafer?

Research path: (a) 10-K + 10-Q segment-data combined with capacity-utilization commentary; (b) industry analyst surveys (LightCounting, Dell’Oro, 650 Group); (c) customer-side commentary on per-wafer cost economics; (d) supply-chain-triangulation via Soitec capacity allocation reports.

Current best-guess ⚠ inferred: GF has not publicly disclosed Fotonix ASP per wafer. Analyst-derived range: $5,000-10,000 per wafer for 300mm SiPh on Fotonix (⚠ inferred from industry-typical 300mm advanced-process wafer pricing + SiPh complexity premium). The wide range reflects:

Lower bound ($5K): consistent with mature 300mm logic + integration premium
Upper bound ($10K): consistent with low-volume specialty-process pricing + SiPh customer-PDK economics

The $1B end-2028 SiPh revenue target implies approximately 100K-200K wafers per year at the analyst-estimated ASP range. Either the wafer-volume math or the ASP math will be disclosed (directly or indirectly) at Investor Day 2026.

Thesis impact: bull case Pillar 3 + Pillar 4; risks M2 (mature-node ASP — note SiPh ASP dynamics are different from mature-node mass-market ASP dynamics).

Q7. Will GF publish a SiPh-only segment-revenue line item separate from “Communications Infrastructure & Datacenter”?

Research path: (a) FY 2025 10-K segment disclosure (Item 8 financial statements + segment footnote); (b) Investor Day 2026 deck — likely venue for any new segment-disclosure decision; (c) analyst-day Q&A commentary.

Current best-guess ◐ partial: Currently, GF reports “Communications Infrastructure & Datacenter” as one of five end-markets in the segment-mix breakdown (Q4 2025 transcript ✓). The ~$200M SiPh revenue is within this segment but not separately broken out. The bull-thesis-supportive disclosure would be to break out SiPh as a separate segment — this would (a) make the $1B trajectory explicit, (b) signal SEC-level financial-disclosure transparency, (c) support photonics-narrative valuation premium. Whether GF will do this is unanswered. Analyst expectation: Investor Day 2026 (May 7) is the most likely venue if it happens.

Thesis impact: bull case Pillar 3 (testability of $1B target). Without segment break-out, photonics-narrative tracking is harder.

Q8. Are recent Q1 2026 Form-4 insider sales (13+ filings 2026-03-20 to 2026-04-27) 10b5-1 plans or discretionary?

Research path: (a) Each Form 4 XML filing has a footnote field that can flag 10b5-1 plan execution; (b) corresponding DEF 14A or 10-Q disclosures may reference plan-adoption windows; (c) executive’s separately-filed Form 144 (notice of proposed sale).

Current best-guess ⚠ inferred: The high cadence (13+ Form 4 filings in ~6 weeks per edgar_recent.json ✓) is most consistent with mechanical 10b5-1 plan execution following an open-window after Q4 2025 earnings (Feb 11 2026). Discretionary insider selling at this cadence would be a meaningful negative signal; 10b5-1 plan execution is mechanical and not informative. This requires primary-source verification by reading the underlying Form 4 XMLs.

Confidence: ⚠ inferred — direct Form 4 XML inspection pending.

Thesis impact: risks K5. Material for sentiment / signal interpretation but limited fundamental-thesis impact regardless.

Q9. What is the actual revenue contribution from AMF in GF’s FY 2025 financials?

Research path: (a) FY 2025 10-K acquisition-related revenue footnote (typically discloses ~6 weeks of acquired-company revenue if material); (b) Q4 2025 earnings call commentary on AMF post-close contribution; (c) FY 2026 10-K full-year run-rate disclosure.

Current best-guess ⚠ inferred: AMF closed Nov 17 2025, giving ~6 weeks of FY 2025 contribution. AMF’s stand-alone FY 2024 revenue is not publicly disclosed (private company). Analyst estimate of AMF stand-alone FY 2024 revenue: $50-100M (⚠ inferred from industry-typical 200mm SiPh foundry economics with ~15 years of operating history). Implied AMF Q4 2025 contribution to GF: $5-15M (⚠ inferred).

Thesis impact: bull case Pillar 2 (post-acquisition trajectory) + risks K4 (potential goodwill-impairment risk).

Q10. What customers are migrating from AMF to Fotonix vs staying on AMF 200mm?

Research path: (a) Customer-side press releases naming Fotonix or AMF specifically; (b) Investor Day disclosure of customer-pipeline-by-fab; (c) AMF customer-base annual reports (e.g., satellite-comms customers, LiDAR customers, sensor-customers); (d) eFPGA / III-V / hybrid-integration customer commentary.

Current best-guess ⚠ inferred: AMF’s customer profile (long-haul optical, computing, LiDAR, sensing per GF press release ✓) is largely complementary to Fotonix’s hyperscaler-AI-data-center customer base. Analyst expectation: most AMF customers stay on AMF 200mm through the upgrade window (3-5 years), with selective migration to Fotonix for high-volume customers as the Singapore 300mm line comes online.

Thesis impact: bull case Pillar 2 + risks C5 (AMF customer attrition).

Q11. What is GF’s 2026-2028 capex envelope by fab + by purpose?

Research path: (a) Investor Day 2026 (May 7) — most likely venue; (b) FY 2025 + FY 2026 10-K capex disclosures + multi-year capex plans; (c) CHIPS Act milestone-tracking documents; (d) NIST + Department of Commerce annual reports.

Current best-guess ◐ partial: Industry estimate aggregates to $9-13B for 2025-2028 across:

Malta NY (CHIPS Act-supported expansion + new-fab construction): ~$4-6B
Vermont (GaN modernization): ~$1-2B
Dresden (EU expansion): ~$2-4B
Singapore (AMF 300mm upgrade + integration): ~$1-2B

The breakdown is partial — GF has disclosed $13B+ “over 10 years” for US CHIPS Act-supported projects (NIST 2024-11 ✓) but not the full multi-geo capex envelope.

Thesis impact: bear case Pillar 4 (capex peak vs FCF) + risks K3.

Q12. What is GF’s pricing structure with hyperscaler-tier SiPh customers (Marvell, Broadcom, NVIDIA, Cisco) — multi-year supply commitment, MFN, take-or-pay?

Research path: (a) Material-contracts disclosure footnote in FY 2025 10-K (only required for individually-material contracts); (b) customer-side disclosures (Marvell 10-K, Broadcom 10-K, NVIDIA 10-K) about long-term supply agreements; (c) any litigation-or-dispute disclosures.

Current best-guess ⚠ inferred: Foundry-customer relationships at the hyperscaler tier typically include multi-year supply commitments with capacity-allocation guarantees and prepayment structures. Specific commercial terms (MFN clauses, take-or-pay, ASP-step-down schedules) are not publicly disclosed. Industry-typical structure: 3-year capacity reservation + customer prepayment covering ~25-50% of NRE + per-wafer pricing with annual ASP step-downs of 5-10%.

Thesis impact: bull case Pillar 1 (customer-relationship moat) + bear case Pillar 5 (lateral-competition pricing pressure).

Q13. Is GF qualifying any next-generation modulator architectures (TFLN, BTO, EO-polymer beyond LWLG/NLM, metasurface) on Fotonix?

Research path: (a) Industry-conference disclosure venues (OFC 2026, OFC 2027, ECOC 2026); (b) joint-press-release announcements with material-IP startups; (c) Investor Day technology-roadmap deck.

Current best-guess ◐ partial: Currently, two material-IP layer customers are publicly known on Fotonix:

Lightwave Logic (commercial PDK live Mar 16 2026)
NLM Photonics (parallel tapeout per universe.json ✓)

Other material-IP architectures (TFLN, BTO, metasurface) may be in qualification pipelines but are not publicly disclosed. The bull-case argument is that Fotonix is modulator-architecture-agnostic and can integrate any of these architectures via heterogeneous integration; the bear-case argument is that competitor foundries (Tower, Intel, TSMC) may achieve faster qualification cycles for the next-generation architectures.

Thesis impact: bull case Pillar 1 + risks T3.

Q14. What is GF’s strategic posture toward Marvell post-Polariton (April 22 2026 close)?

Research path: (a) GF or Marvell joint-press-release (post-April 22 2026 — none yet observed); (b) Marvell Q1-Q2 CY 2026 calls; (c) Marvell process-technology roadmap commentary referencing Fotonix.

Current best-guess ⚠ inferred: Marvell’s pre-Polariton Fotonix relationship is multi-year. Polariton’s POH modulator IP technically integrates well with Fotonix’s monolithic SOI process (chip-on-Fotonix heterogeneous integration is a known path). The probability that Marvell’s post-Polariton roadmap continues to use Fotonix as the foundry of record is high (estimated >70% probability ⚠ analyst inference). Counterfactual: Marvell could qualify Polariton at TSMC SiPh or Tower PH18 if those processes offer compelling economics — but the switching cost is meaningful, and Marvell’s existing Fotonix engagements are deep.

Thesis impact: bull case Pillar 1 + cross thesis implications.

Q15. What is the explicit Mubadala posture beyond 50% control threshold?

Research path: (a) Future SC 13G/A amendments; (b) Mubadala IR disclosures about strategic-investment-portfolio rotation; (c) any Mubadala-Government-of-Abu-Dhabi-strategy-document references.

Current best-guess ⚠ inferred: At current cadence (~$1B selldown every 12-18 months), Mubadala reaches 50% sometime in 2027-2028. Mubadala has not publicly committed to a target stake post-50%. Industry-typical sovereign-wealth-fund posture: maintain 30-40% strategic stake long-term in flagship industrial-policy assets; allow free-float to reach 50%+ for institutional investor accommodation.

Thesis impact: bull case Pillar 5 (capital structure) + bear case Pillar 1 (overhang) + risks K1.

What this list deliberately does not include

The research-prompt direction explicitly noted that the open-questions file should be honest where the analyst doesn’t know — not fabricate answers. The following types of questions have been deliberately not added because the analyst genuinely has no informed inference yet:

Specific named hyperscaler-customer revenue concentration percentages
Detailed cost structure of the AMF + InfiniLink integration
Multi-year forward gross margin guidance by segment
Specific GF position on EU Chips Act renegotiation outcomes

These are open-questions categorically, but listing them with speculative best-guesses would violate the source-validation rule.