Samsung hofft auf großen Erfolg mit AMDs Chipbestellung und fordert TSMCs Marktführerschaft heraus

It’s better than Intel’s Foundry that’s getting nothing, but it’s hard to see it as a win because TSMC’s 3nm production lines are 100% booked by NVIDIA, Apple, Intel, Qualcomm etc.

However, I don’t agree with the media’s recent narrative that Samsung has completely failed because

Samsung’s upcoming Exynos 2500 + HBM4 +Mach-1 chips will be make-or-break moments for them.

Samsung has also repeatedly mentioned since 2018 or 19 that their 2nd generation 3nm GAA process will be the real deal, not the first generation, and they’ve acknowledged they will lag technologically and logistically behind TSMC before they make it into production.

They need to put their money where their mouth is by showing that Exynos 2500 (which are to built on the 2nd gen 3nm GAA process, now marketed as 2nm) is superior in both computing power and efficiency. If that’s proven, the tides could potentially change, but these little ‘wins’ mean little.

Meanwhile, their collab on Mach-1 with Naver could potentially make HBMs less favourable in the market, and buy some time to play catch up to Hynix’s HBM prowess, since it will be only 1/8th the price of Hynix’s HBM and consume 1/10th the power of an HBM. This would theoretically allow them to completely dominate the AI accelerator + memory chip market if what they’re saying is true, but there’s nothing to say their claim is plausible without a measurable product.