Client Interest in Samsung Foundry Reportedly Buoyed by Nintendo Switch 2 SoC Production Deal

[T0@st]

The Nintendo Switch 2 hybrid console is due to launch globally next Wednesday (June 4). The highly anticipated next-gen handheld is powered by a custom NVIDIA processor. To the surprise of many industry watchdogs, both parties have semi-recently disclosed a couple of technical details regarding their fruitful hardware collaboration. Historically, Nintendo has guarded many aspects of its past generation hardware. Throughout the 2020s, data miners and leakers have unearthed plenty of pre-release information—leading to theories about the Switch 2 chipset's origins. During the Switch 1 era, TSMC was the chosen manufacturing partner. NVIDIA's off-the-shelf Tegra X1 mobile SoC powered the first wave of Nintendo Switch (2017) devices, in 20 nm form. A 2019 revision resulted in Switch Lite and (refreshed) Switch models being equipped with a more efficient 16 nm solution, also present within 2023's premium OLED variant.

Since then, Switch 2's alleged NVIDIA Tegra T239 SoC was linked to a Samsung 8 nm node process. Earlier this month, extremely brave Chinese leakers produced "full die shot" evidence of South Korean foundry origins. Bloomberg insider news articles have implied that Samsung Semi's mature 8 nm FinFET node is better suited—rather than an equivalent TSMC product—for the Switch 2's custom NVIDIA chipset. Unnamed sources have mentioned critical factors; namely stable production and process compatibility. Industry moles reckon that Samsung leadership is actively and aggressively pushing for a longer Switch 2 chipset production deal. Renewed terms could include a future die shrink; pre-launch analysis indicates a sizeable 207 mm² footprint. Beyond foundry biz negotiations, additional murmurs suggest company executives dangling an OLED panel supply agreement. Industry experts have viewed Samsung's key entry—into the gaming console chip market—as a seismic development. A DigiTimes article dives into a so-called "tripartite cooperation"—involving Nintendo, NVIDIA, and the South Korean semiconductor giant. The Samsung Foundry has floundered and struggled in recent times, but is keen to catch up with its arch rival. Fresh rumors have AMD and Sony considering Samsung's chip making channels; possibly with futuristic PlayStation hardware in mind.



View at TechPowerUp Main Site | Source

 

[Pizderko]

So no gate all around for the poor nintendo switch 2???

​

 

[bonehead123]

So no gate all around for the poor nintendo switch 2???

Well, if this happens, then we can call it "Sammygate"..

But if the hardware collabs do happen, we can call the new gamr tois "SammySwitch" and "PlayGate" hahahaha..

 

[Contra]

Why should this surprise anyone? Everyone knows that Samsung manufactured the entire Ampere line on its own process, which automatically led to the fact that the entire Jetson Orin line, based on Ampere, and the Switch2 SoC is a standard Orin Nx, is also manufactured by Samsung on the same node.

 

[Rightness_1]

Wow, a chip with the specs of an 10 year old CPU, and an 8 year old GPU has people flocking to Samsung... OK...

 

[Tomorrow]

So no gate all around for the poor nintendo switch 2???

That's space age technology for Nintendo. Maybe in 2035 with Switch 3. If Samsung still makes anything at that point...

Wow, a chip with the specs of an 10 year old CPU, and an 8 year old GPU has people flocking to Samsung... OK...

Yeah the headline is funny. Samsung has nothing with witch to entice AMD, Sony or other to use their process nodes. They dont have node leadership. They dont have yields or capacity. No one is really willing to make anything in Samsung fabs because what do they actually offer to clients? Old 10nm class process which is cheap, but also outdated?

5nm that Snapdragon 888, Exynos 2100 and Google G1 used a long time ago?
4nm that Snapdragon 8 Gen 1 used and later abandoned due to yield/performance issues?
3nm that Samsung even cant make enough of their own Exynos 2500 to ship it outside South-Korea?
2nm that's a pipe dream considering struggling 3nm yields?

Qualcomm already abandoned them after the 4nm fiasco. Google just signed a five year contract with TSMC to produce future Tensor chips for their Pixel devices and AMD/Nvidia/Apple all need cutting edge nodes that only TSMC offers with good yields.

Samsung even stumbled heavily with their HBM3e that others aced. This used to be their bread and butter but now they cant even get that right and who knows about their HBM4. Plus in the memory space there are more powerful competitors compared to process node space.
Samsung also bungled their first gen DDR5 and their SSD's are also nothing special these days.

Yet i constantly see puff piece articles stating how everyone is evaluating Samsung for their future products and how Samsung will score big any day now.
Nothing ever comes of it because of course companies evaluate and even produce test chips on Samsung's newest nodes. The fact that none of them have actually used Samsung's nodes in last few years should tell you everything about Samsung's performance.

 

[Pizderko]

That's space age technology for Nintendo. Maybe in 2035 with Switch 3. If Samsung still makes anything at that point...

I'm completely speechless ....... at this point ill skip absolutely everything till GAA-FET litho.

 

[alawadhi3000]

Yeah the headline is funny. Samsung has nothing with witch to entice AMD, Sony or other to use their process nodes..

AMD is interested in their SF4X process.

They don't have node leadership.

No need for node "leadership" to win contracts, most customers aren't after the latest node.

They don't have yields or capacity.

Yields are very good for 4nm and 8nm, 3nm is "okay" as well.
Capacity can be increased/decreased depending on contracts.

No one is really willing to make anything in Samsung fabs because what do they actually offer to clients? Old 10nm class process which is cheap, but also outdated?

No one is willing to make anything there yet their foundry has high utilization rate?
The customer chooses the right process for their needs/targets/price/availability ...etc, there is nothing "outdated" about Samsung's 8nm.

4nm that Snapdragon 8 Gen 1 used and later abandoned due to yield/performance issues?.

Snapdragon 8 Gen 1 is 5nm, not 4nm.

3nm that Samsung even cant make enough of their own Exynos 2500 to ship it outside South-Korea?.

Thats your guess, not a fact.

2nm that's a pipe dream considering struggling 3nm yields?.

You do know that Samsung 2nm is just a renamed 3rd gen 3nm process right?

Qualcomm already abandoned them after the 4nm fiasco. Google just signed a five year contract with TSMC to produce future Tensor chips for their Pixel devices and AMD/Nvidia/Apple all need cutting edge nodes that only TSMC offers with good yields.

Get your facts straight, Qualcomm still makes new chips on Samsung 4nm, and it was Google's plan from the beginning to eventually design their own chips, nothing to do with Samsung Foundry.

 

[Tomorrow]

AMD is interested in their SF4X process.

For what? Some low volume bargain bin APU's? Currently as far as i know AMD does not have anything made on SF4X and i doubt they ever will.

No need for node "leadership" to win contracts, most customers aren't after the latest node.

But we're talking about AMD here. If course most RF, automotive etc customers dont really need the latest and greatest, but most customers also dont hold x86 license. The fact that Samsung has not scored contracts from AMD, Nvidia or Qualcomm in last few years speaks volumes.

Yields are very good for 4nm and 8nm, 3nm is "okay" as well.
Capacity can be increased/decreased depending on contracts.

3nm being below 50% reportedly is not what i would consider "okay". 8nm is ancient by today's standards and 4nm performance is bad even if it's yields are good. Yield also depends on the chip and it's size. Making simple ASIC's for crypto miners is not the same as making a big and complex CPU's or GPU's.

No one is willing to make anything there yet their foundry has high utilization rate?
The customer chooses the right process for their needs/targets/price/availability ...etc, there is nothing "outdated" about Samsung's 8nm.

What good is high utilization rate if most things made there are sold at discount just to keep the fabs running?
Samsung also has utilization issues on their US fabs on newer nodes. Likely due to lack of interest from customers and poor yields.
This is a point i strongly disagree with. Samsung does have fab utilization issues and no wonder if none of the bigger companies actually make anything in there in volume.

Snapdragon 8 Gen 1 is 5nm, not 4nm.

Doesn't change the fact that Qualcomm abandoned Samsung halfway trough SD8 G1 and went with TSMC.
This is unusual because usually that does not happen. Samsung's node must have been really bad at the time.

Thats your guess, not a fact.

Reports are conflicting. Some say it's restricted top SK, Others say that it does launch worldwide (albeit only in Flip 7 model).

You do know that Samsung 2nm is just a renamed 3rd gen 3nm process right?

I dont believe anything Samsung claims.

Get your facts straight, Qualcomm still makes new chips on Samsung 4nm, and it was Google's plan from the beginning to eventually design their own chips, nothing to do with Samsung Foundry.

Which new chips? I only found Snapdragon 4 Gen 2 that is an entry level chip and several years old at that.
Likely this too will end soon. As for Google - you think it was Google's plan to go with more expensive TSMC? I doubt that.

Just excuses and grasping at straws here. As much is i want competition to TSMC i dont see the point in sugarcoating it's competitors when they are not delivering.

 

[alawadhi3000]

For what? Some low volume bargain bin APU's? Currently as far as i know AMD does not have anything made on SF4X and i doubt they ever will.

You can doubt it all you want, they are interested for a reason.

But we're talking about AMD here. If course most RF, automotive etc customers dont really need the latest and greatest, but most customers also dont hold x86 license. The fact that Samsung has not scored contracts from AMD, Nvidia or Qualcomm in last few years speaks volumes.

3nm has been out for almost three years and both AMD & NVIDIA are still not using it, just to prove that they don't need the latest node.

3nm being below 50% reportedly is not what i would consider "okay". 8nm is ancient by today's standards and 4nm performance is bad even if it's yields are good. Yield also depends on the chip and it's size. Making simple ASIC's for crypto miners is not the same as making a big and complex CPU's or GPU's.

A process won't hit HVM if yields are 50% for small chips.

What good is high utilization rate if most things made there are sold at discount just to keep the fabs running?
Samsung also has utilization issues on their US fabs on newer nodes. Likely due to lack of interest from customers and poor yields.
This is a point i strongly disagree with. Samsung does have fab utilization issues and no wonder if none of the bigger companies actually make anything in there in volume.

Sold at a discount? Last time I checked contracts are not disclosed, stop believing rumors and presenting them as facts.

Reports are conflicting. Some say it's restricted top SK, Others say that it does launch worldwide (albeit only in Flip 7 model).

Those are rumors, not reports.

I dont believe anything Samsung claims.

You do know that Samsung admitting having problems with their 2nm node and rebranding 3rd gen 3nm as 2nm is bad PR for them right?

Which new chips? I only found Snapdragon 4 Gen 2 that is an entry level chip and several years old at that.
Likely this too will end soon. As for Google - you think it was Google's plan to go with more expensive TSMC? I doubt that.

6G3, XR2G2 ...etc

Just excuses and grasping at straws here. As much is i want competition to TSMC i dont see the point in sugarcoating it's competitors when they are not delivering.

Nobody is sugarcoating anything or grasping straws here, you need to stop believing only the rumors that fits your narrative and presenting them as facts.

 

[Tomorrow]

Keep living in your dream world. In the meantime Samsung keeps falling behind more and more every year. I have said everything i wanted to say on this topic.

 

[alawadhi3000]

Keep living in your dream world. In the meantime Samsung keeps falling behind more and more every year. I have said everything i wanted to say on this topic.

Last year Exynos 2400 vs Snapdragon 8G3 actually showed Samsung closing the gap to TSMC but hey lets believe the random Chinese guy tweet over this.

 

[mrnagant]

That's space age technology for Nintendo. Maybe in 2035 with Switch 3. If Samsung still makes anything at that point...

Yeah the headline is funny. Samsung has nothing with witch to entice AMD, Sony or other to use their process nodes. They dont have node leadership. They dont have yields or capacity. No one is really willing to make anything in Samsung fabs because what do they actually offer to clients? Old 10nm class process which is cheap, but also outdated?

Not every chip manufactured is on bleeding edge.

At my work, we buy Spartan 7 FPGA's from AMD by the hundreds. They are built on 28nm.
Global Foundries did almost $7B of revenue in 2024 and their most advanced node seems to be only 12nm.

The real issue is the huge investment for the pursuit of bleeding edge to only not be able to sell anything on it while it is bleeding edge.