Intel’s ambitions to catch up with TSMC and reclaim its position in chip manufacturing are a constant topic today, but in fact it is Samsung who is currently Taiwan’s closest competitor in advanced technologies. However, even Samsung has struggled to keep pace lately, and their 1.4nm process was recently rumore to be scrapped entirely. But new reports suggest Samsung intends to stay in the race – and is working on a 1nm process.
Scrapping 1.4nm…
In mid-March, reports surfaced that Samsung, facing persistent issues in the development, yield, and cost-efficiency of its 3nm node, had decided to abandon plans for its most cutting-edge process yet. While the 2nm technology was said to remain on the roadmap, the 1.4nm process – slated to follow – was allegedly cancelled. Skipping an entire node is no trivial decision; if the company truly opted for that path, it would likely signal a broader retreat from pursuing the most advanced process technologies, leaving the 2nm node as its final effort.
Such a move would place Samsung among the companies that have already dropped out of the race: GlobalFoundries, IBM, UMC, and STMicroelectronics. Only Intel and TSMC would remain – potentially joined by China’s SMIC, though that company is still grappling with 5nm production due to its lack of access to Western EUV equipment. Still, with the backing of the Chinese government, SMIC has a unique shot at joining the top tier, despite its ongoing technological lag.
If Samsung were to bow out of competition with TSMC, the consequences for the tech market and consumers would likely be negative – less competition typically translates to higher prices and slower innovation. However, roughly a month later after said reports, new developments are emerging that may reignite some hope.
…or stepping up to the “dream process”?
According to Korean outlet Seoul Economic Daily, Samsung has in fact begun preparations for a 1nm (or 1.0nm) process, the next phase following the 1.4nm node. If the reports are accurate, it may also imply that the 1.4nm process has not been entirely shelved after all. The company could therefore be planning to remain in the race for at least two more generations.
Samsung has reportedly established a dedicated team to work on the 1nm process. However, the Seoul Economic Daily report does not clarify the size of this team or whether it is conducting early-stage feasibility studies – which might never lead to a marketable product – or actively developing a commercial manufacturing node meant to actually ship. Process development is a multi-year endeavor, and research relevant to a given node often begins several generations ahead. But if the 1.0nm process is to debut in a few years, current work would have to already be beyond the initial stages.
Official roadmaps state that the preceding 1.4nm process (referred to externally as SF1.4) is expected to be completed or launched in 2027. The 1.0nm process, meanwhile, is reportedly targeted for 2029. This timeline is likely provisional, and delays would not be surprising. Each new generation of chipmaking technology raises complexity and costs steeply, and even TSMC is now beginning to experience schedule slippages.
Achieving this level of miniaturization will probably require new approaches. The process is expected to use High-NA scanners for exposure – machines featuring so-called high numerical aperture optics. We previously covered what High-NA entails in this article.
Samsung reportedly refers to the 1.0nm process internally as the semiconductor “dream process.” It seems the company is eyeing a kind of “moonshot” with this node – a bold leap aimed at enhancing its competitiveness with a groundbreaking technology. But this ambition also brings the risk of significant disappointment. Let’s hope Samsung avoids the pitfall of setting such lofty goals that the project collapses entirely – something Intel notoriously did with its ill-fated 10nm node, which triggered a deep and lasting crisis of the company with its cascading effects. The company doesn’t necessarily needs to outdo TSMC, the 1.0nm process just needs to be functional and usable (even if it ends up falling short of TSMC’s in some performance metrics) and that itself will count as a win compared to earlier reports suggesting the company was dropping all post-2nm development.
Sources: techPowerUp (1, 2), Seoul Economic Daily
English translation and edit by Jozef Dudáš
⠀
