Phew, so NVIDIA launched the new RTX 50 series of graphics cards at CES 2050 – including the RTX 5090, RTX 5080, RTX 5070 Ti and the RTX 5070. The other GPUs in the series (like the RTX 5060 and maybe RTX 5060 Ti) will come later on in the year.

However these cards have a higher power requirements – with the RTX 5090 having a TDP of 575w, much higher than the 450w of the RTX 4090.

So can you PSU cope with this? Do you need to run out and buy a new PSU? I answer these questions and more, and I also talk in detail about the ‘melting’ 12vhpwr connector, ATX 2.0 vs 3.0 vs 3.1, and whether you need to change your 12vhpwr cable to a 12vh-2×6 ‘cable’ (spoiler alert: you don’t… but you might need a new PSU!).

If you prefer text over video, please read on for the guide/transcript version of this video.

Video Transcript And Guide

Hey everyone, the new RTX 50 series of GPUs is finally here, and I’m really excited about it—apart from the power requirements. That’s because the 40 series was plagued with negative stories about the new 12-volt power adapter melting and taking people’s RTX 490s with it. Also, the new 50 series of GPUs actually use more power.

So, the RTX 570 needs 50 watts more than the RTX 4070, and the RTX 590 actually needs a whopping 125 watts more than the RTX 490, which is a bit scary considering the 490 was melting.

If you’re looking at upgrading to the 50 series of GPUs, do you actually need a new PSU? Well, I’ll answer that in detail in this video so you have all the information you actually need to know.

My Corsair ATX 2.0 and 3.0 PSUs

I actually have two power supply units here. I’ve got a white ATX 2.0, which is the Corsair RM750. So that’s the white one, which is 2.0. And then I’ve got this black one, which is the Corsair RM750 that is actually ATX 3.0. What that means is Corsair shipped the 12-volt power adapter in the box, so I can actually use this for powering any of Nvidia’s 40 series or 50 series of graphics cards—assuming, of course, the TDP doesn’t get exceeded, because this is actually a 750-watt power supply unit.

So, if the actual graphics card says it needs 1,000 watts, like the RTX 5090, I obviously couldn’t use this one. But what’s kind of confusing is, let’s say this was actually a 1,000-watt power supply unit—I probably still wouldn’t use it to power my RTX 5090, even though it actually comes with a 12-volt power cable from Corsair. I probably still wouldn’t use this, and I’ll explain why throughout this video.

It kind of comes down to the quote: “A chain is only as strong as its weakest link.”

Power Method #1: ATX 3.0 And 12vhpwr

That would be one way of actually powering your 50 series of graphics cards: you get an ATX 3.0 or 3.1 PSU. (And I’ll talk about ATX 3.1 in a minute.) You get one of them, plug in the relevant cables into your PSU, and then use the 12-volt power adapter to connect to your GPU. That is a valid way of powering it, and that’s the first way. I’ll talk about the pros and cons of that later. It’s mainly got advantages, not disadvantages, but I’ll still talk about some of the flaws of this approach later on.

Method #2: ATX 2.0 W/ PSU Adaptor

The other way you could do it is you could carry on using your ATX 2.0 power unit. But, of course, you can actually sometimes go out and buy a cable directly from the PSU maker. This is a Type 4 connector type, and as a result, I can go onto the Corsair store and buy a Type 4 cable. That’ll actually give me a 12-volt power adapter.

Then, because this is a 750-watt power supply unit, I could use that to power a few of the different 50 series of graphics cards. So, that is something I could do, and that’s the second way of actually doing things.

Method #3: ATX 2.0 W/ NVIDIA Adaptor

The third thing you could actually do is you could keep your ATX 2.0 power supply unit and then use the adapter that comes in the Nvidia box. For example, with the RTX 5070, it comes with an adapter that allows you to plug into old-style 8-pin PCI Express cables. That gives you a 12-volt power connector that you then plug into your graphics card.

That is quite a good option. It’s why Nvidia included it in the box. But I personally wouldn’t use that either. Again, I’ll talk through where I would and wouldn’t use all of these methods in a minute, but I just wanted to walk through the five different methods you could use for powering your RTX 50 series of graphics cards.

Method #4: ATX 2.0 W/ CableMod Adaptor

The fourth option open to you, again, is to keep your existing power supply unit and go to a fairly reputable company like CableMod and buy one of their adapters. Obviously, not the right-angle adapter that got recalled after catching on fire—you buy one of the other 12-volt power adapters and use that. So, that’s the fourth way you could do it.

Method #5: PSU With 12vhpwr Slot

The final method is that you might have seen some power supply units that actually have a native 12-volt power or 12-volt 2×6 power slot directly on the PSU. You could go out and buy one of those PSUs and use that. Actually, that’s the solution I would favor, and I’ll explain why in a minute. Basically, with that, you have fewer points of failure. Again, that comes down to “a chain is only as strong as its weakest link.”

Part of the fundamental problem of the RTX 50 series, in my opinion, is it just needs so much power. Obviously, the 12-volt power adapter isn’t really suited for that. There are loads of stories online about how it’s not really suited. That’s why the 12-volt 2×6 connector came in with ATX 3.1. What that did was actually change the sensing pins and the conductors inside to make things slightly safer. There have been fewer stories of melting and things like that since that’s been brought out.

The 12vhpwr Connector Is VERY Flawed

That is actually one of the first points I needed to talk about. With the actual change to this connector, some people actually get confused and think that you need to run out and buy a new PSU or new cable. That’s not actually the case. This is an ATX 3.0 power supply unit, and it comes with the 12-volt power cable. But I can’t actually change this to a 12-volt 2×6 cable.

You Can’t Buy A 12vh-2×6 Cable

The change for 12-volt 2×6 is on the female connector side. So, this is a male plug that plugs into a graphics card. It’s actually at that point—the connector on the graphics card—that is changed with ATX 3.1 for the 12-volt 2×6 power connector.

In other words, if you have a cable like this, you don’t actually need to run out and change it for a 12-volt 2×6 cable. You just need to make sure that your graphics card does actually use 12-volt 2×6 and not 12-volt power connectors. You might be thinking, “Obviously, that’s going to be happening,” but actually, Nvidia’s 40 Super series used 12-volt 2×6 connectors. Yet, some of the board partners used the older-style 12-volt power connector, which was a bit concerning considering it was melting.

I don’t really know why they did that, but if you are buying an RTX 50 series of graphics cards, just make sure it does have a 12-volt 2×6 connector on the graphics card side and not an older-style one. At the time of filming, it’s actually a bit too early to say whether the board partners are using 12-volt or 12-volt 2×6 connectors. I’m assuming they are using the newer, safer version, but I just wanted to flag that point.

A Chain Is Only As Strong As…

Right, so with all that said, you don’t need to change your PSU in that case. But when do you need to change your power supply? Well, it all comes back to “a chain is only as strong as its weakest link.” I know I’ve said that quite a few times, but essentially, if you imagine a chain with loads of links—I can’t really show that very well; I feel like I’m doing an “Incy Wincy Spider” now or something—but if you imagine a big chain with loads of links, the idea is that if one of the links is loose, then the entire chain will break.

The same is true of powering your RTX 50 series of graphics cards. If you take the RTX 5090, for example, the adapter that’s included in the box requires four 8-pin PCI Express connectors. In other words, you’d get an old-school PCI Express cable with eight pins on it—this one, this is a Corsair one—you’d need to put it together. You’d get four of these, and what you’d need to do is actually plug them into your RTX 5090.

Then, at the other side of that adapter, you have a 12-volt power cable that you plug into your RTX 5090. That sounds great, doesn’t it? You don’t need to change your PSU or even buy any new cables. Well, there is a massive, fundamental flaw with that, in my opinion.

When NOT To Use The NVIDIA Adaptor

Firstly, if you have an 8-pin cable, what you might notice is you’ve got this particular pigtail. I won’t rant about that much more; I have another video where I called this “devil spawn” and talked about how awful it is. But with the RTX 5090, you can’t use the pigtail. Yeah, so you can’t just get two cables to then double up on the pigtail to get your four connections into the RTX 5090—you’re not allowed to do that. This might not deliver enough power.

So, what you’re actually going to do from your power supply unit is have four different PCI Express cables plugged into your PSU. Then, you have four of these going into the RTX 5090’s adapter, and then you’ve got one cable from there going into the RTX 5090.

But have you thought about the problem yet? Well, you might have, but if not, I’ll just explain it to you. Basically, on the PSU side, you’ve got four different things connecting to your PSU. So, that’s four things that could go wrong. Then, on the adapter side, you’ve got four more of these plugging into the adapter, bringing us to eight different things that could go wrong. Then, you have the 12-volt power adapter that goes into your RTX 5090—that is a ninth thing that could go wrong.

So, you don’t need to just think about whether your 12-volt power cable is plugged all the way in—you’ve got to think about all the other things too.

Of course, if you had an RTX 5070 that only needs two actual 8-pin connectors into the adapter, that’s actually not too bad. You’ve then got two bits going into the PSU, two going into the adapter, and then one bit going into your graphics card. But even then, that’s still five different things that could go wrong.

And whilst there’s been no stories of melting RTX 4070s, the RTX 5070 does have more power—50 watts more. Obviously, if it’s overclocked, it would use even more. I personally wouldn’t want to use my ATX 2.0 power supply unit and then use the included adapter for that purpose.

Other Methods I Wouldn’t Use

Actually, the same is true of some of the other methods I’ve mentioned, like going to CableMod and getting one of their adapters and maybe plugging an 8-pin cable into my power supply unit. I personally wouldn’t want to do that. There’s no disrespect to CableMod or anything like that; again, I’m just introducing too many points of failure in that particular power chain.

So that is two of the five methods that I spoke about earlier and why I would or wouldn’t use them. The third one, of course, is using an official cable from Corsair or something like that. So, you know, with your ATX 2.0 power supply unit, you go out and buy an official Type 4 cable, for example, from them. That gives you 12-volt power, and you plug that in.

I would probably do that for something like the RTX 5060 or the RTX 5070 because it’s low-power enough. But I wouldn’t personally do that for an RTX 5090. It needs so much power—575 watts. I personally wouldn’t actually trust plugging two bits of these into my power supply unit and then one into the RTX 5090.

What Type Of PSU I WOULD Buy

And that brings me on to what I would do to power an RTX 5090 or something like that. I would actually go out and buy a power supply unit that has an actual 12-volt power connector or a 12-volt 2×6 connector on the PSU itself. Actually, MSI MAG and Be Quiet, and a few others, do that. At the time of filming, Corsair just announced a new RM 2025 updated power supply unit that also has a 12-volt 2×6 slot on the actual PSU itself.

What that means is you have one cable that plugs into the PSU and then just one cable that connects into your GPU—for example, the RTX 5090 in this case. That would actually, in my opinion, be the best thing you could possibly do because then your power chain just consists of two things: one thing going into the PSU and one thing going into the graphics card.

Safely Plugging In The 12vhpwr Cable

Then, it comes to actually plugging in. The general rule is: make sure you plug in the cable all the way, and there are no gaps around the connector. Then leave it. Don’t start wiggling the cable or regularly pulling it out to see if it’s in properly or not, because you could then actually not plug it in all the way. Obviously, that’s not what you want.

So, plug one bit into the PSU, one bit into the GPU, ram it all the way in—no jokes, please—and then just leave it.

Okay, so that is what I would personally recommend. Before I wrap up, though, there are two points I really needed to talk about.

ATX 3.0 vs 3.1 PSU?

Firstly, if you’re looking at an ATX 3.0 or 3.1 PSU, which should you actually buy? The problem is, it’s not really a simple decision. The ATX 3.1 power supply units are slightly more efficient, but in my opinion, they’re not necessarily safer than the ATX 3.0 ones when it comes to powering the actual graphics card.

You know, you are going to get slightly marginal gains with a 3.1 one, but it’s not super substantial. That means you don’t need to go out and chuck your ATX 3.0 power supply unit or anything like that. That will work as long as it’s at the rated power and everything else.

A Warning About GPU TDP

That brings me to the second point. So, this is a 750-watt power supply unit. Obviously, there are some cards in the series, like the RTX 5070 Ti, that say 750 watts is required. But, of course, that is the minimum, and that’s also the specs from Nvidia itself. So, it’s saying that you need that at the minimum for that particular card.

But when it comes to the board partners, they might actually overclock a 5070 Ti and increase the power usage even more. It might be that they then say you need an 800-watt power supply unit as its minimum level and not a 750-watt one.

In other words, unless you’re buying a Founders Edition directly from Nvidia—which might not be possible due to scalpers and annoying things like that—you should always check the board partners’ own power requirements and TDP when deciding what actual power supply unit to buy.

That really covers everything you need to know. If you’ve got any questions or concerns, though, please leave me a comment. If you found this video useful, please click the thumbs-up button, and please subscribe to see more videos like this. Thanks for watching!