Samsung is replacing FinFET transistors with Gate-All-Around
Samsung expects its 2nm chips to feature a 5% performance hike compared to 3nm
So how much improvement is there between the 3nm Exynos 2500 and 2nm Exynos 2600? Previous estimates called for 2nm SoCs to generate:
- Up to 12 percent increased performance compared to 3nm.
- Up to 25 percent better power efficiency compared to 3nm.
- Up to 5 percent reduced area compared to 3nm.
- Up to 5 percent increased performance compared to 3nm.
- Up to 8 percent better power efficiency compared to 3nm.
- Up to 5 percent reduced area compared to 3nm.
Samsung Foundry’s all-important yield at 2nm has reportedly risen from a low 30% to a range in the 50%-60% area. That is important because it means that Samsung Foundry can now increase wafer production from the previous monthly output of only 15,000 units. As a result, Samsung Foundry should be able to supply enough Exynos 2600 APs for Samsung’s 2026 flagship line.
Are we seeing a sustainable turnaround for Samsung Foundry?
Samsung Foundry has inked deals with MicroBT and Canaan, two of the world’s biggest cryptocurrency mining equipment manufacturers, for 2nm GAA chips. The foundry also has a multibillion-dollar deal with Tesla. This is good news for Samsung Foundry as it had only 7.3% of the foundry market in the second quarter this year compared to TSMC’s leading 70.2% slice of the global foundry pie. Samsung Foundry is expected to turn profitable in 2027 at which time Samsung hopes to have its market share up to 20%.
This would represent a major turnaround for Samsung Foundry considering that it was just a few years ago when Qualcomm pulled the Snapdragon 8 Gen1 AP from Samsung, revised the specs, and had TSMC build the Snapdragon 8+ Gen 1. Since then, all Snapdragon chips have been manufactured by TSMC. Every year there is talk about Qualcomm switching back to Samsung Foundry, but so far this has not taken place despite the annual rumor.
