Samsung's Foundry Collapse and the Diversification Lie
Samsung's foundry share has stalled at 7% while TSMC dominates at 73%. As Washington and Brussels bet billions on Samsung as a TSMC alternative, the data suggests Samsung may be a liability, not a lifeline.
The 7% wall
The latest quarterly data from Counterpoint Research tells a story that few Western policymakers want to hear. Samsung’s foundry business, once billed as the world’s most credible alternative to TSMC, remains trapped at 7% of the global pure-play foundry market. That number is not a fluctuation — it is a floor. Samsung posted 7% in Q2, the same as Q1. It has held that level since the third quarter of last year, after falling from 9% a year ago and 8% two years prior. Meanwhile, TSMC sits at 73% and has not blinked.
The context makes the stagnation sting. The global pure-play foundry market grew nearly 30% in the second quarter compared to the same period last year, driven by a surge in AI-related orders. Every major foundry player rode that wave. Samsung did not. While TSMC expanded 3nm volume and moved 2nm into mass production, Samsung is still working on yield improvement for its own 2nm process, internally designated SF2.
Who wins, who loses
TSMC wins, obviously. It now holds three-quarters of the market and shows no sign of loosening its grip. Its 8-inch and 12-inch mature-process capacity is tight, advanced packaging supply is constrained, and demand continues to outstrip its ability to deliver. That is not luck — it is the compounding result of two decades of focused investment and customer loyalty.
Samsung’s largest potential customers lose. Qualcomm, NVIDIA, Apple, AMD — the list reads like a roll call of the AI boom. These are companies that have placed enormous bets on Samsung as a secondary source, partly for supply-chain resilience and partly to give themselves leverage over TSMC pricing. None of those bets have paid off. Samsung’s foundry yields on advanced nodes remain unreliable, and customers are learning that a second source with poor yields is worse than no second source at all. It creates the illusion of diversification while concentrating risk.
China’s SMIC, at 5% share, wins in a different way. Counterpoint notes that SMIC posted record revenues in Q2, crushed by domestic demand and overseas orders it could barely absorb. The Chinese state has built a domestic ecosystem around SMIC that functions efficiently within its constraints — slow by the most advanced standards, but adequate for a huge segment of chip demand that would otherwise flow to Taiwan or South Korea. Samsung, for all its geopolitical subsidies, cannot touch that market. Geopolitics has closed the door.
The diversification gamble
Here is the uncomfortable truth that Washington, Brussels, and Tokyo are struggling to reconcile with their policy frameworks. The entire Western strategy for semiconductor diversification assumes that Samsung is a viable backstop — a company that can absorb orders when TSMC is geopolitically exposed or physically stretched thin. The CHIPS Act incentives flowing to Samsung’s Texas fab, the European interest in Samsung as a potential partner, the Japanese government’s quiet calculations — all of them rest on the premise that Samsung’s foundry business is on an upward trajectory.
The data contradicts that premise.
Samsung’s 7% share is not stable because it is growing. It is stable because it is not growing at all. While the pie has expanded 30% this quarter, Samsung’s slice has not. That means Samsung’s absolute revenue growth in foundry is roughly keeping pace with inflation and wafer price increases, not with demand. In real terms, Samsung is losing ground even as it stands still.
Counterpoint projects that foundry conditions will remain tight into the second half of the year, with wafer average selling prices continuing to climb. That sounds positive for Samsung on paper — higher prices should boost revenue. But it is TSMC that sets those prices and captures most of the margin. Samsung’s customers are the ones paying more for access to a supplier that cannot reliably deliver.
Why the Korean press is sounding alarms
The Korean outlets covering this story are using language like “ultra-emergency” — a term reserved for national crises. That is not hyperbole when you consider what is at stake. Samsung’s foundry division accounts for a meaningful portion of the company’s overall profit base, and its strategic importance to South Korea extends far beyond revenue. The foundry business is the anchor around which Samsung hopes to retain relevance in the most valuable segment of the semiconductor industry. If that anchor fails, the entire company narrative shifts.
More importantly, South Korea’s position in the global chip order depends on Samsung’s foundry success. The country’s trade surplus, its technological prestige, its bargaining power with both the US and China — all of it is tied to a single company’s ability to compete with TSMC. Samsung losing that race is not just a corporate setback. It is a national recalibration.
What happens next
Samsung has two realistic paths forward, and neither is reassuring. The first is to pour even more capital into SF2 yield improvement and hope for a breakthrough that attracts major customers. This is the path Samsung is currently on, and it has already consumed billions with no market-share return to show. The second is to pivot toward mature-node specialization, competing on cost rather than leading-edge performance. That path cedes the AI boom entirely to TSMC and leaves Samsung as a commodity player in a market where margins are thinning.
The US and EU may respond by throwing more money at the problem — additional subsidies, regulatory favours, forced customer allocations. But subsidies cannot manufacture yield. They can build fabs, but they cannot rebuild trust with customers who have been burned by delayed deliveries and inconsistent output.
The diversification strategy is not dead. TSMC’s dominance is a real vulnerability, and having alternatives matters. But Samsung is not the alternative the West needs it to be. Until Samsung’s foundry business demonstrates that it can move above single-digit market share with reliable execution, it remains a geopolitical prop rather than a genuine option. And props do not survive contact with reality.
The 7% wall is not just a number. It is a signal — that the competitive gap between Samsung and TSMC is structural, not cyclical, and that the diversification bet may have been placed on the wrong horse from the start.