Samsung Foundry faced several challenges in its transition to the 3nm process, so much so that it failed to secure major manufacturing contracts. The company now aims to turn things around with its 2nm process, but it might have work to do. Reports indicate that Samsung hasn’t been able to improve its 2nm yield rate for months. Its yield still languishes at around 10%, while TSMC has already achieved a 60% yield in its 2nm pilot production.
Samsung struggles to improve its 2nm yield but is hopeful of a turnaround
While Samsung upgraded to the more advanced GAA architecture for its 3nm chips, TSMC stuck with the previous-gen FinFET architecture. It turned out to be a masterstroke from the Taiwanese firm, as its Korean rival failed to secure a healthy 3nm yield, allowing it to run away with all major contracts. With TSMC also switching to the GAA architecture for its 2nm chips, the playing field was expected to be leveled again.
However, according to the Korean media, Samsung’s 2nm yield is still substantially less than TSMC’s. The company hasn’t lost hope though. To ramp up its 2nm capabilities, Samsung is reportedly converting a 3nm production line in one of its Korean foundry campuses. The company aims to complete this transition in the first quarter and start trial production in Q2. it plans to establish a monthly capacity of around 10,000 wafers.
The ultimate goal is to produce the Exynos 2600 on a 2nm process node. The new chip should enter mass production before the end of this year, so Samsung is racing against time here. Its poor 3nm yield delayed the Exynos 2500 — the chip still isn’t official. Time will tell if the Korean behemoth can turn things around in the foundry segment, and do it quickly. TSMC may not start 2nm mass production this year, so if Samsung makes it happen, it might be able to regain some of the lost ground.
