Earnings Highlights: GlobalFoundries (GFS) | FY2026 Q2 — Optical Comms +62% in One Quarter, Gross Margin Nears 30% a Year Early, SiPho Full-Year Target Doubles

2 days ago

8 min read

1. This is no longer a “mature-node foundry” story. Communications Infrastructure and Data Center (CID) revenue grew 62% YoY in the quarter — the fastest growth since 2022 — and management raised its full-year CID growth target on the spot from the “high 30s” to 50–60%, while upgrading its full-year silicon photonics (SiPh) revenue target from “growth” to “more than double in 2026 versus last year.”

2. Gross margin hit its target a year early. Non-IFRS gross margin was 29.9%, up nearly 470 basis points YoY. Management had originally said it would “reach 30% by the end of 2026”; Q2 is already right there, Q3 is guided straight to 30.5%, and full-year gross margin is now expected at ~30%. That signals a structural improvement in product mix, not a cyclical upswing.

3. Two U.S. government checks + three new growth vectors. Quantum is set to receive an expected $375 million, and silicon photonics signed another $300 million LOI. Add SiGe being “sold out through 2027,” AI data center power (IVR), and MIPS/ARC custom silicon all igniting at once. GF is rewriting itself as a “differentiated process platform for AI infrastructure.”

This article is for technology and industry trend analysis only and does not constitute investment advice.

1. Three Core Conclusions (Executive Takeaway)

First, the growth engine has completely changed hands. Smartphones, which used to carry revenue (about 36% of the mix this quarter), are still down 6% YoY. What truly defined GF's narrative this quarter was CID at only ~16% of revenue: up 62% YoY and 20% QoQ, and its seventh straight quarter of double-digit YoY growth. The fourth-largest end market contributed almost all of the “story premium.”

Second, the quality of gross margin has changed. Non-IFRS gross margin jumped to 29.9%, but the point is not the number itself — it's that the CFO stated plainly that the roughly $100 million of additional revenue versus a year ago fell almost “dollar for dollar” to gross profit. When the marginal gross margin on incremental revenue approaches 100%, it means the additions are high-end processes like silicon photonics and SiGe, and the mix improvement is structural. Last quarter, in Earnings Highlights: GlobalFoundries (GFS) | FY2026 Q1, we flagged this “margin target reached early” trajectory; this quarter simply cemented it.

Third, this was an “upgrade everything” call. Full-year CID growth raised (high 30s → 50–60%), home and industrial IoT raised (mid-single digits → 10–15%), full-year silicon photonics revenue raised (growth → doubling), technology services revenue raised (MIPS contribution of $60–100 million → $100–120 million including ARC), and full-year gross margin raised (touching 30% at year-end → ~30% for the full year). The only two items moving the other way were smartphones (dragged by memory shortages, now expected down low double digits for the year) and CapEx as a share of revenue trending toward the upper end of its range.

2. Revenue and Financials

Key numbers at a glance (non-IFRS, comparable basis).

Revenue of $1.786 billion, up 6% YoY (+5.8%) and 9% QoQ (+9.3%), on shipments of about 625,000 300mm-equivalent wafers (+8% YoY, +8% QoQ); revenue and profit landed at or above the high end of guidance. Non-IFRS gross margin of 29.9%, up 470 basis points YoY — the company's record for a second quarter (IFRS gross margin 28.3%). Operating income of $298 million, operating margin 16.7%, up 140 basis points YoY. Net income of about $256 million, up about $22 million YoY; diluted EPS of $0.46 (non-IFRS; IFRS $0.30), at the high end of guidance, on about 556 million diluted shares.

Cash: operating cash flow was $405 million, CapEx net of government incentives was about $408 million (about 23% of revenue), and adjusted free cash flow was -$3 million (in line with prior guidance). Cash and marketable securities at quarter-end were about $3.3 billion, with total debt of about $1.1 billion. On July 14, GF paid its first-ever quarterly cash dividend of $0.12 per share and declared another dividend of the same amount.

Technology services revenue was about 11% of total, with gross margins structurally above the company average. Boosted by the MIPS and Synopsys ARC IP acquisitions, the full-year technology services contribution was raised from the original $60–100 million estimate to $100–120 million, with the full-year share now trending toward the upper end of the 10–12% range.

By end market (Q2 share / YoY / QoQ): smart mobile devices ~36% / -6% / +15%; automotive ~19% / -10% / -13% (due to customer pull-in timing; still expected to grow low double digits for the year); home and industrial IoT ~19% / +10% / +30% (fastest YoY since 2022); communications infrastructure and data center ~16% / +62% / +20%.

(

3. Technology and Business Highlights: Three New AI Data Center Vectors Ignite at Once

GF split its growth story into three parts this quarter: quantum, optical communications and power, and IP and custom silicon.

Optical communications is the main act. Today, the vast majority of silicon photonics revenue comes from pluggable optical modules. GF already works with four of the top five optical transceiver makers and has upgraded its full-year silicon photonics revenue target to “more than double in 2026.” On the technology roadmap, 200G/lane is in volume production, 400G has been demonstrated, and higher bandwidth is in development. The real highlight is the SCALE platform — the industry's first OCI MSA-compatible near-packaged (NPO) / co-packaged (CPO) optics solution, with seven active engagements, one SCALE-related tape-out completed in Q2 and another expected in Q3. GF stressed that NPO and CPO share the same photonic IC (PIC), so whichever form factor customers adopt, GF benefits. We broke down the industry backdrop of this “line-side spec war” in Meta, Broadcom and AMD Jointly Define the OCI 200G Line-Side Spec — GF's SCALE is a bet on the OCI MSA side. On the packaging side, the “detachable fiber / glass waveguide” hurdle, GF and Corning's results at ECTC 2026 (GlobalFoundries × Corning Detachable Glass Waveguide Connector) are the physical evidence.

SiGe is underappreciated. Management said outright that GF's silicon-germanium (SiGe) business is larger than silicon photonics today, and that 2027 capacity is already oversubscribed. The reason: at high bandwidth, the TIAs and drivers that CMOS used to handle must move to high-performance analog solutions like SiGe at 200G/400G per lane — so SiGe growth is the product of two forces, the “shift to optics” and “bandwidth upgrades.” GF is expanding capacity in Vermont (Burlington) and qualifying 300mm SiGe in Singapore.

Power and quantum are two options. In July, GF completed its acquisition of Photeon's IVR (integrated voltage regulator) team, and management summed it up in one line: “IVR is to power what CPO is to optics” — both move functions right next to the chip and integrate them at the wafer level. In quantum, GF formed a dedicated team in May with a “qubit-agnostic” strategy built on FDX's cryogenic CMOS platform. It already works with eight leading quantum players (including PsiQuantum, Quantinuum and Quantum Motion), launched four new customer engagements within three months, and is set to receive an expected $375 million from the U.S. Department of Commerce.

MIPS + Synopsys ARC = a springboard upstream into custom silicon. In June, GF completed the ARC processor IP acquisition, bringing in 150 patents, 300 existing customers and 400 R&D engineers, allowing GF to engage customers' compute architectures earlier in the design cycle and paving a custom silicon path toward physical AI.

4. Management Outlook: Turning “Upgrades” into a Full Package

Q3 2026 guidance: revenue of $1.885 billion ± $25 million (about +5.5% QoQ), non-IFRS gross margin of about 30.5% ± 100 basis points (midpoint up about 450 basis points YoY), non-IFRS operating expenses of $260 million ± $10 million, operating margin of 16.7% ± 170 basis points, and non-IFRS diluted EPS of $0.51 ± $0.05.

The full-year wording deserves a line-by-line read for certainty: full-year CID growth of 50–60% (phrased as “now expect,” i.e., a raised expectation); full-year silicon photonics revenue “more than double” (“now expect to more than double”); home and industrial IoT 10–15% (raised from mid-single digits); full-year gross margin “about 30%” (upgraded from a target of “touching 30% at year-end” to a full-year expectation); CapEx at 15–20% of revenue, trending toward the upper end; full-year free cash flow margin still about 10%.

One signal that's easy to miss but critical for 2027: GF has already agreed price increases with customers across multiple technology corridors this quarter, which will show up in revenue starting in 2027. Management described “very constructive customer conversations” and said it will “continue to evaluate 2027 pricing in the second half of 2026” — layering a pricing lever on top of mix improvement for gross margin, just with timing that lands next year. On the NPO/CPO ramp, GF kept its consistent view: NPO ramps starting in 2027, CPO starting in 2028. We laid out why this timeline is “bullish, not bearish” for the entire optical module supply chain in CPO Volume Pushed to 2028 — but That's Not Bad News.

5. Supply Chain and Customer Clues

Pull out the people and company names mentioned on the call, and the signals get clearer.

Demand validation bypasses direct customers and goes straight to the cloud giants. Management said silicon photonics demand isn't just vouched for by direct customers but has also been cross-checked with large hyperscalers; endorsements for SCALE and silicon photonics span leading XPU vendors, hyperscalers and AI connectivity leaders. Silicon photonics customers now exceed 40, and customers are asking GF to “lock in more capacity for me,” partly to diversify geopolitical risk.

Design wins are blooming across end markets. Automotive: an ADAS radar win for Bosch on the FDX platform and 5V/10V PMICs on the BCD platform. Smartphones: a first BCD power management IC design win with MediaTek, plus a microLED display backplane for a hyperscaler's AR glasses. IoT: three chiplet design wins with Lockheed Martin (FinFET + FDX) and an expanded FinFET collaboration with Microchip. Data center power: the first design win for a smart power stage gate driver on the BCD platform.

The government is a co-investor. Quantum's $375 million plus silicon photonics' $300 million LOI — both checks point to the theme of “advanced manufacturing on U.S. soil,” effectively externalizing part of GF's CapEx risk. On competition, management's view of more rivals entering silicon photonics is that “it shows this path is real.” Read alongside CPO's First Commercial Year Begins — TSMC COUPE Volume Production and the 200G EML Bottleneck, this scale-up battle for capacity and process positioning makes it clearer that GF is positioned to “scale silicon photonics capacity 10x within its existing four walls, without building new fabs.”

6. Conclusion

This quarter GF delivered an earnings call with “ordinary revenue growth (+6%) but extraordinary growth quality.” What to track is not total revenue but three leading indicators:

One, whether quarterly CID revenue can hold above $300 million — based on guidance, CID should run at about $350 million per quarter by year-end, the direct yardstick for whether “50–60% full-year growth” materializes. Two, whether gross margin keeps climbing after settling at 30.5% in Q3 — because 2027 still carries the not-yet-reflected lever of price increases, and only once that stacks on mix improvement does the 40% 2028 exit target become convincing. Three, the number of SCALE tape-outs and the timing of the NPO ramp in 2027 — the watershed for GF upgrading from a “pluggable silicon photonics supplier” to an “NPO/CPO platform supplier.”

In one sentence: this quarter, GlobalFoundries formally rewrote itself from a “mature-node foundry value stock” into a “differentiated process platform for AI optical interconnect and power.” When incremental revenue turns almost entirely into gross profit, and the U.S. government backs its quantum and silicon photonics efforts, what the market needs to reprice is its content — not just its capacity.