Samsung Reclaims Its Mojo

For much of the past decade, Samsung Electronics (OTCM: SSNLF) has struggled with a credibility problem across two of its most important semiconductor businesses: foundry and advanced memory.

The issue was not capital investment, ambition, or technology roadmaps. It was execution. At critical points, Samsung failed to deliver consistent yields and power efficiency at scale, and customers responded by shifting volume elsewhere.

Today, Samsung is re-establishing itself as a credible supplier at the leading edge.

Foundry: execution -> market share

Samsung entered the foundry business (i.e. manufacturing chips designed and branded by other firms) with meaningful scale in the early 2000s. Then, as process complexity increased, shortcomings in yield and power-efficiency caused customers to shift volume production to suppliers that could deliver consistent results. Samsung’s share briefly expanded to a robust 17% circa 2020 when 5nm chips appeared. But the firm’s share then declined sharply as execution problems surfaced, while the Taiwan Semiconductor (NYSE: TSM) foundry continued to gain share by delivering consistent quality.

The current leading-edge process across the industry is “2nm-class gate-all-around.” Samsung’s version has now reached production readiness. Yield and power-efficiency issues have been materially addressed. That should build customer confidence. Foundry credibility is binary: customers either trust the process, or they do not.

Memory: a single-generation setback

Until recently, Samsung maintained its momentum in high-bandwidth memory, (HBM) holding a stable second-place position through HBM2, HBM2E, and HBM3. The disruption occurred during the transition to HBM3E, when poor power efficiency and thermal behavior harmed credibility with key customers. As Samsung worked through those challenges, competitors were able to steal share.

Samsung’s HBM share declined sharply after 2023, reflecting the HBM3E qualification delay. It appears to be recovering into 2026 as next-generation products enter production. In the meantime, Micron has become a more powerful presence in the HBM market.

Capacity enables share

Market share is not solely a function of technology leadership. Production capacity matters, especially when demand exceeds supply. As AI keeps pumping up HBM demand, Samsung’s ability to regain HBM share is supported by its planned wafer capacity expansion. By 2026, Samsung’s HBM capacity (190 thousand wafers per month) will be more than double its 2024 level and almost double Micron’s 2026 level. At 210,000 wafers per month, SK hynix (KRX: 000660) will maintain its modest capacity lead over Samsung.

Tesla testifies

Tesla’s AI chip roadmap illustrates how foundry execution directly influences customer behavior. Tesla moved production away from Samsung when it lost credibility on yield. Tesla is now returning to Samsung for its next-generation platform as Samsung addressed performance, yield, and long-term supply readiness.

Investor Takeaway

Samsung’s setbacks over the past several years were real. In both foundry and advanced memory, execution failures allowed competitors to gain ground. But the evidence across market share, capacity, and customer behavior suggests these failures were temporary.

Samsung does not need to dominate either foundry or HBM to change its earnings trajectory. It needs to re-establish itself as a credible, large-scale supplier at the leading edge. That, Samsung appears to be doing.