Samsung’s foundry business is finally starting to look less like a money pit and more like a race again. The company’s 2 nm chip yield has pushed above 60%, and that improvement, combined with stronger order intake, could put Samsung Foundry back in the black in the third quarter of 2026 – three months earlier than it had expected.
That matters because the contract chipmaking arm has spent years absorbing heavy losses while Samsung poured money into new fabs and chased advanced nodes against TSMC, which still sets the pace in cutting-edge manufacturing. A higher yield is the difference between a flashy roadmap and something that actually prints money.
Samsung’s 2 nm chip yield is moving out of the danger zone
According to Chosun, Samsung had previously expected profitability only by the end of the year. The shift appears to be coming from better production performance on advanced processes and a growing pipeline of customers lining up for 2 nm capacity.
The current yield figure is still short of the roughly 70% level generally seen as sufficient for efficient mass production, but it is no longer the kind of number that makes executives sweat through their shirts. In chip manufacturing, crossing 60% is often the point where optimism stops sounding silly.
Tesla, HBM4 and Samsung’s new customer list
Samsung’s order book is also getting a lift from a few big names. The company recently landed a Tesla deal worth about $16.5 billion, while Nintendo, Apple and Nvidia are being mentioned as potential customers. Separately, base dies for HBM4 memory should add more revenue as demand rises in AI systems, accelerators and data centers.
- 2 nm yield: above 60%
- Economic mass-production target: around 70%
- Expected profitability: third quarter of 2026
- Tesla contract value: about $16.5 billion
- 2 nm order volume in 2026: up about 130% year on year
Samsung is chasing TSMC with a fuller pipeline
Samsung’s internal forecast calls for 2 nm orders in 2026 to rise by roughly 130% from the previous year, which suggests customers are warming to the process faster than the company first expected. That is the real story here: not just better yields, but enough demand to make the manufacturing curve worth climbing.
If the trend holds, 2026 could become the first year in a long while when Samsung Foundry looks sustainably profitable instead of merely expensive. The open question is whether it can keep yields rising while ramping volume fast enough to hold on to those customers before TSMC, as usual, reminds everyone who still runs the market.