Real Men Have Global Fabs

Author: Claus Aasholm (Semiconductor Business Intelligence / Siliconomy) URL: https://clausaasholm.substack.com/p/real-men-have-global-fabs One-line: A market-data-heavy analysis of GlobalFoundries’ strategy, finances, and EUR 1.1B Dresden expansion — arguing GF is uniquely positioned as the leading NCNT (Non-China, Non-Taiwan) foundry as TSMC’s AI compute focus crowds out everything else.

[PARTIAL] This is a paywalled Substack post. The free portion covers GF’s history, business metrics through Q2 2026, foundry market share, and the Dresden investment rationale. The paid section (“The Strategic Response of GlobalFoundries…”) is behind a paywall and not captured here.

Key claims

• TSMC commands 70%+ foundry market share in Q2 2026 with less than 43% of the top-6 foundry capacity — its dominance is pricing power, not volume. Average price per processed 12-inch wafer: ~$3,300, nearly 30% above competitors.
• GlobalFoundries is the leading foundry with no home-market advantage — its NCNT (Non-China, Non-Taiwan) strategy, baked in from its 2009 founding as an AMD/Mubadala joint venture, is now a geopolitical asset rather than a financial constraint.
• GF’s gross margin target is 40%, currently running at ~28%, with a 2028 deadline and a new long-term aspiration of 45%. The gap closes mainly through utilization improvement and strategic repositioning, not bleeding-edge nodes.
• GF is at 88% utilization in Q2 2026 — near full, making the EUR 1.1B Dresden capacity expansion well-timed: as TSMC’s AI focus crowds out mature-node capacity, GF can absorb that displaced demand.
• TSMC’s 2026 12-inch equivalent wafer capacity growth is 0% — every new 2nm wafer added displaces an older node. This is the structural opening for GF: customers needing non-AI mature-node chips are increasingly crowded out of TSMC’s production schedule.
• European foundry revenue has grown since the last peak, but Europe’s share of global foundry revenue has fallen from 8.6% to 5.1%. GF holds 18% of European foundry revenue — its Dresden expansion is explicitly designed to serve European clients in a market where TSMC is the dominant force but increasingly capacity-constrained on AI compute.
• The non-AI, non-memory semiconductor COGS market is the one GF competes in — and it has grown more in line with GF’s actual revenue trajectory. The AI compute narrative overstates GF’s competitive distance from the market.
• GF recently acquired MIPS (RISC IP), signaling a strategic shift away from pure-play manufacturing toward a tech-company identity — hence the title’s update to “Real Men Have Global Fabs” (plural), not just one commodity fab.

Notable quotes

“Real Men Have Fabs!” — Jerry Sanders III, AMD, early 1990s

“If you don’t control manufacturing, you don’t control your destiny.” — Jerry Sanders III

“You have many explanations but no excuses. I want to know what you are going to do about it!” — Aasholm’s former boss, quoted to frame GF’s margin underperformance

“Despite much fanfare about TSMC’s overseas ventures, the company has both feet firmly planted on the silicon island. More than 80% of the company’s non-current assets are in Taiwan.”

“Because GlobalFoundries deliberately chose not to pursue bleeding-edge technologies, and, with that, the central compute sockets, it was excluded from the hyper growth of the AI compute era.”

“Up until now, the AI revolution has predominantly benefitted TSMC, which has responded by obsoleting and limiting mature wafer runs, creating a significant opportunity for GlobalFoundries moving forward.”

“Only 2.9% of Q2-2026 tool sales went to Europe, so GlobalFoundries’ investment of more than a billion euros is meaningful and needs to be viewed from a European perspective.”

Charts and figures

[image: GlobalFoundries revenue and gross margin, 2020–Q2 2026 — held in the private wiki] GF revenue grew during the COVID upcycle but has flatlined since; gross margins peaked near 34% and sit at ~28% in Q2 2026.

[image: GlobalFoundries gross margin trajectory — held in the private wiki] The 40% long-term GM target has been raised to 45% at the recent investor day; current trajectory needs more than utilization fill.

[image: Top-6 foundry market share, Q2 2026 — held in the private wiki] TSMC holds 70%+ revenue share on <43% of top-6 wafer capacity. GF is second in the non-TSMC tier.

[image: Foundry market share excluding TSMC — held in the private wiki] GF’s non-TSMC share has declined from 14.5% to 12.2%; SMIC has taken the most ground.

[image: Foundry capacity utilization rates, Q2 2026 — held in the private wiki] GF at 88% utilization — near-full, validating the Dresden expansion decision.

[image: Foundry wafer prices comparison — held in the private wiki] TSMC’s average 12-inch wafer price ~$3,300; GF is second at just under $3,200 — the premium is real and sticky.

How it connects

• GlobalFoundries — the subject; GF’s NCNT strategy, Dresden expansion, and margin trajectory.
• TSMC — the structural foil; TSMC’s AI compute dominance is the displacement force that creates GF’s mature-node opening.
• AMD — GF’s founding parent; Jerry Sanders III’s “Real Men Have Fabs” is the intellectual origin of the piece.
• Semiconductors — the broader industry context; fabless/foundry bifurcation, AI compute dominance.
• fabless — the model GF is the alternative to; the foundry side of the fabless/foundry split.
• Chip War — Kyle’s read on the semiconductor geopolitics that make NCNT strategy strategically meaningful.
• Inside TSMC’s Arizona Expansion — companion read on TSMC’s overseas limitations; GF’s geographic distribution is the answer to what TSMC can’t do.