After repeatedly failing to hit yield targets on earlier nodes, the Korean tech goliath appears to have finally twigged that biting off more than it can chew is not paying off.
The 3nm GAA node was marketed with plenty of fanfare as the first to deploy gate-all-around transistors, but it ended up being a lemon in terms of volume readiness. According to Sedaily, the company has mothballed plans for a 1.4nm production line at Pyeongtaek 2, which had been pegged to begin as early as the second half of 2026.
The delay is hardly shocking. Samsung’s foundry business is floundering, with low order volumes and tanking revenue. The unit reportedly lost around 2 trillion won (€1.34 billion) in Q1 alone and is slashing its annual investment budget from over 10 trillion won (€6.7 billion) to just 5 trillion (€3.35 billion). In short, it is circling the wagons.
That 1.4nm test line was supposed to be the division’s only meaningful play in bleeding-edge nodes this year. Its cancellation underscores just how badly it is struggling to land major customers.
Now the focus has pivoted to making the 2nm process work. Samsung’s foundry team is trying to shore up yields ahead of producing the Exynos 2600 chip for the Galaxy S26 smartphone. Nam Seok-woo, the unit’s CTO, is leading a dedicated task force to sort it out.
Even so, yields are stuck in the 20-30 per cent range, which makes the economics a tough sell. Samsung is angling for orders from Tesla, Qualcomm and other North American giants. The Taylor fab in the US is also being considered for 2nm deployment, but that hinges on getting the process to behave.
Depending on demand, Samsung may flip part of its Hwaseong S3 fab’s 3nm line into 2nm before year-end. Asked for comment, the company gave the usual non-answer, refusing to confirm any foundry investment plans.
Despite pushing 1.4nm out to what now looks like 2028, Samsung’s spin doctors insist this is a “restructuring” to focus on what is “rapidly advancing” at 2nm. Mass production on that node is now scheduled for the end of 2025 in Korea and early 2026 in the US at the Taylor fab if the stars align.
The foundry unit is haemorrhaging money and investment is being throttled accordingly. That makes the talk of optimism on the 2nm front feel more like desperate PR than actual progress.
With TSMC already snapping at its heels and other competitors not exactly sitting still, Samsung needs the 2nm node to deliver something resembling a win.