Monday, November 2nd 2026
GPU Makers Focus on RTX 3060/3050 as RTX 50 Series Inventory Dries Up
GPU makers across Asia, particularly in China, are reportedly redirecting their sellers to older graphics cards as an alternative due to the low supply of NVIDIA's latest GeForce RTX 50-series "Blackwell" graphics cards. They are focusing on the recently resurrected GeForce RTX 3060 and RTX 3050, which are entry-level graphics cards from two generations ago, reminiscent of the "Ampere" generation. A ChannelGate report notes that some add-in card partners are already running out of GeForce RTX 50-series inventory due to cryptocurrency mining and e-commerce promotions in China, which are significantly draining the supply of these GPUs across retailers.
In June, NVIDIA reintroduced its GeForce RTX 3060 12 GB edition. This SKU features 3,584 CUDA cores and 12 GB of GDDR6 memory. For this, NVIDIA has once again used Samsung's 8 nm DUV node, as it did in the past. The entire NVIDIA "Ampere" architecture lineup was produced on the 8 nm DUV node. Reportedly, retailers are forcing sales of RTX 3060 and RTX 3050, in addition to some AMD Radeon SKUs, to offload demand for GeForce RTX 50-series designs. Below image is EVGA's older GeForce RTX 3060, not in production anymore.
Source:
ChannelGate
In June, NVIDIA reintroduced its GeForce RTX 3060 12 GB edition. This SKU features 3,584 CUDA cores and 12 GB of GDDR6 memory. For this, NVIDIA has once again used Samsung's 8 nm DUV node, as it did in the past. The entire NVIDIA "Ampere" architecture lineup was produced on the 8 nm DUV node. Reportedly, retailers are forcing sales of RTX 3060 and RTX 3050, in addition to some AMD Radeon SKUs, to offload demand for GeForce RTX 50-series designs. Below image is EVGA's older GeForce RTX 3060, not in production anymore.
80 Comments on GPU Makers Focus on RTX 3060/3050 as RTX 50 Series Inventory Dries Up
I was really hoping that production would at least pivot a little bit and we could get some special looking low profile 3050 and 3060 cards that can still hang.
At this point I'm hunting around the used enterprise lot and antique 2-4GB Pascal cards that sell around $35-45 mark are getting hammered like crazy.
It's also real bad with 4-8GB Turing cards for some reason. Guys, we're being priced out of retired enterprise/workstation cards. It's getting hella rough.
Naturally, they need the space to bring in new, mass-produced models to sell in the stores.
But there are precious few *good* games out there today that can't be played on a 3060 at reasonable quality settings. That's the silver lining for the gamer crowd; we've had years of performance stagnation on the hardware side, and years of putrid AAA development on the software side, so you're not really losing much of anything if you don't have a great GPU.
Far more concerning from my perspective are the hikes to RAM and storage. Storage is one of those things on which you really can't cut corners when shopping for replacements.
Ofc I'd pick a 5060 instead if the 3060 was $200 and the 5060 was $300, but those aren't the prices anymore.
When the 3060 rerelease was announced I think pricing was like $20 below the 5060 MSRP, which is of course ridiculous, because even a 5050 is a better buy and that's one of the worst options this generation, topped only by the RX 90"50". But I suppose the context is that nothing is priced at MSRP anymore.
Despite having the option of a £1000 5080, £750 5070 Ti and £550 5070 when I bought my TUF, somehow I don't regret not spending a bit more now as much as I did a while ago. Sure. The 5080 would have been the play in hindsight, but I'm happy with the 5070 and it's the same/better than my 3080 Ti.
"Well, you're not living in a time when RTX 40x0 and 50x0 are available. It doesn't matter what was available before, only what's available right now, this second. Memory of a gold fish is a recipe for long and happy life!"
We heard that even from reviewers, for instance here at TechPoweUP when they tried to hype the $1200 RTX 4080 as a great deal - but you must forget about RTX 3080's $800 MSRP, think of the cost at the height of cryptomining madness!
All sold within the last month - and there are more pages under.
And then the usual selling points on features because everyone became a gamer and creator since the coof.
I don't need the B-Frames on encoders that I'm not ready to use (platform lockout) but enough video editors want it.
Also when you're coming from something like a RX 580, 1060 or anything just as old, 7th gen NVENC and up is good.
Something something RT, tensor, AI, CUDA, mindshare, whatever. I just care about high gcore clock and memory.
The benchmarks are also the most extreme scenario you're going to have while gaming and streaming. It's bearable.
If fitment isn't an issue, there's still the consideration that a 1080 goes for ~$100 now (and it still pulls way ahead).
288€ - rtx 3050 6GB
334€ - rtx 3050 8G
480€ - rtx 3060 12GB
359€ - rtx 5050 8G
474€ - rtx 5060 8GNot likelyImpossible.
GTX 680 -> 770
HD 7850 -> R7 265
HD 7870 -> R9 270(X)
HD 79x0 -> R9 280(X)
R9 290(X) -> R9 390(X) (they got double the VRAM)
Most RX 400 -> RX 500
Have there been other mid to higher end rebrands in the last 15 years?
Can't really be bothered to dig out examples, but they're there. Off the top of my head, RX 480 and RX 580 were identical silicon, RX 590 used a better node IIRC.
A 3050/3060 on the updated Samsung 8 nm would have been more interesting. If it's an identical node to the original (revised 10 nm marketed as 8 nm), bit of a shame.
Kinda agree that as they bring back old models, why not have them shrinked. Like AMD did with their Ryzen AF CPUs, practically a 2000 series Zen+ CPU with a 1000 series naming.