Samsung Could Start 1nm Mass Production Sooner Than Expected

[Nomad76]

Samsung Foundry business is set to announce its technology roadmap and plans to strengthen the foundry ecosystem at the Foundry and SAFE Forum in Silicon Valley from June 12 to 13. Notably, Samsung is expected to advance its 1 nm process mass production plan, originally scheduled for 2027, to 2026. This move could look like a surprise since recent rumors (denied by Samsung) emerged about HBM3 and HBM3E chips running too hot and failing to be validated by NVIDIA.

Previously, Samsung successfully mass-produced the world's first 3 nm wafer foundry in June 2022. The company plans to start mass production of its second-generation 3 nm process in 2024 and 2 nm process in 2025. Speculations suggest Samsung may integrate these nodes and potentially begin mass-producing 2 nm chips as early as the second half of 2024. In comparison, rival TSMC aims to reach the A16 node (1.6 nm) in 2027 and start mass production of its 1.4 nm process around 2027-2028.



View at TechPowerUp Main Site | Source

 

[AnarchoPrimitiv]

Samsung being able to seriously compete on the newest nodes would be a huge benefit to all consumers, plus it doesn't hurt to have manufacturing facilities that are not located in Taiwan

 

[Chrispy_]

How big are 1nm gates, really?
The next node is presumably 0nm, and if anyone believes that's a real node measurement, they're going to absolutely lose their sh1t at -1nm!
Time for a better naming scheme this year, we've hit the end of the road for the "nanometers" system.

 

[watzupken]

I feel the biggest problem for Samsung fab is the poor yields. Being able to produce whatever nanometer with poor yields is not going to win them any clients.

 

[Tomorrow]

I feel the biggest problem for Samsung fab is the poor yields. Being able to produce whatever nanometer with poor yields is not going to win them any clients.

Exactly. Glad people have caught up to Samsung's BS. They announce their nodes WAY in advance but the products on those nodes dont materialize for years, or ever.

The "3nm" as they call it is a perfect example as first produced in 2022. We're in 2024 now and not even Samsung's own Exynos 2400 chips that are small and used in their own products use 3nm (neither the first or second gen). Their next Exynos 2500 is supposed to finally use the second gen 3nm but only starting from 2025. Even their own flagship S24 Ultra uses Qualcomm's 4nm chip. That about sums up Samsung's process nodes...

Couple that with recent news about their HBM3e failing qualification and you can surmise that any announcement from Samsung has to be taken with a boatload of salt.

 

[Prima.Vera]

Samsung Foundry business is set to announce its technology roadmap and plans to strengthen the foundry ecosystem at the Foundry and SAFE Forum in Silicon Valley from June 12 to 13. Notably, Samsung is expected to advance its 1 nm process mass production plan, originally scheduled for 2027, to 2026.

You mean Samsung's Marketing team decided to step-up and said it "fruck it" . Let's be first to name the same 5 years old process, 1nm, because it sounds better on paper with investors

 

[ARF]

The "3nm" as they call it is a perfect example as first produced in 2022. We're in 2024 now and not even Samsung's own Exynos 2400 chips that are small and used in their own products use 3nm (neither the first or second gen). Their next Exynos 2500 is supposed to finally use the second gen 3nm but only starting from 2025. Even their own flagship S24 Ultra uses Qualcomm's 4nm chip. That about sums up Samsung's process nodes...

What's the benefit of using Qualcomm's chip, instead of their own, even if produced on a slightly older node? For example, Samsung's 8nm still served quite well even for nvidia's high-end GPUs.
Any process is good enough for a phone.

 

[Tomorrow]

What's the benefit of using Qualcomm's chip, instead of their own, even if produced on a slightly older node? For example, Samsung's 8nm still served quite well even for nvidia's high-end GPUs.
Any process is good enough for a phone.

Qualcomm for all it's anti-competitive and monopolistic behavior and higher prices on it's SoC does have the best modem and the best GPU (i guess Apple would argue this point). It's also more power efficient inherently by design (not just because it's 4nm). Exynos was supposed to eclipse (Xclipse) Qualcomm's Adreno GPU by pairing AMD's RDNA with Exynos but thus far it has not done so. Ironically Adreno itself is a distant progeny of ATI from back in the day if im not mistaken.

I would not say that any process is good for a phone. It needs to have good power efficiency, good density and the modem is the hardest part to get right - just ask Intel who tried and failed and Apple for all it's engineering resources has been unable to produce their in-house design for years and likely years to come - forcing them to buy their modems from Qualcomm.

But my point was that if Samsung is not using 3nm for even their own Exynos chips and are using essentially TSMC's 4nm then how can they win contracts for it?
Plus Exynos has a really poor reputation -. just look at comment section when discussing mobile SoC's and you'll see much hate for it.
Obviously it's not THAT bad but it's no Snapdragon either. In the past the only benefit Exynos had against Snapdragon was bootloader unlocking but since most people no longer do that it's likely not a big argument (assuming Samsung has not locked this down too).

 

[Minus Infinity]

Exactly. Glad people have caught up to Samsung's BS. They announce their nodes WAY in advance but the products on those nodes dont materialize for years, or ever.

The "3nm" as they call it is a perfect example as first produced in 2022. We're in 2024 now and not even Samsung's own Exynos 2400 chips that are small and used in their own products use 3nm (neither the first or second gen). Their next Exynos 2500 is supposed to finally use the second gen 3nm but only starting from 2025. Even their own flagship S24 Ultra uses Qualcomm's 4nm chip. That about sums up Samsung's process nodes...

Couple that with recent news about their HBM3e failing qualification and you can surmise that any announcement from Samsung has to be taken with a boatload of salt.

You better tell Lisa Su, she's just done a deal with Scamsung for their new 3nm node with GAAFET. AMD will use TSMC and Scamsung going forward.

 

[DemonicRyzen666]

How big are 1nm gates, really?
The next node is presumably 0nm, and if anyone believes that's a real node measurement, they're going to absolutely lose their sh1t at -1nm!
Time for a better naming scheme this year, we've hit the end of the road for the "nanometers" system.

That 1nm is the smallest line that can measure of the the cubic volume of the gates. Since things like FINFET & trigate exist it's not possible to count it as single transistors but if you count it as the physical voume by the size it takes up by cubic it's around 50-45nm cubed currently.

 

[Tomorrow]

You better tell Lisa Su, she's just done a deal with Scamsung for their new 3nm node with GAAFET. AMD will use TSMC and Scamsung going forward.

Likely some lower end chips. TSMC will remain AMD's main manufacturing partner.

 

[Minus Infinity]

Likely some lower end chips. TSMC will remain AMD's main manufacturing partner.

What do you mean by lower end chips?