Nvidia RTX 50-series Blackwell designs and specifications expected to be finalized this month — RTX 5080 rumored to have 400W TGP

Jensen Huang at GTC 2024
(Image credit: Nvidia)

The designs of the RTX 5080/D and RTX 5090/D are expected to be completed in September, reports BenchLife.info, which cites sources within the factories working on these upcoming GPUs.

These GPUs are scheduled to officially launch this month (according to Google’s translation of the article), which apparently moves their launch forward from the initially leaked CES 2025 launch. However, a Tweet by @Zed__Wang asserts that this is a (Google Translate) machine mistranslation. Instead, what BenchLife.info meant is that the designs will be finalized by this month. If that is the case, then it seems to be on track for a late 2024 or early 2025 launch — likely at CES 2025.

In related news, @kopite7kimi confirmed through a private reply to another X (formerly Twitter) user that the RTX 5080 will require 400 watts. The X user said, “So, 600w for 5090 and 400w for 5080 is right?” and the leaker simply said “Yeah,”. If this information is true, then it means that the 5080 will get an 80-watt bump in TGP. It’s also a bigger bump than what was previously expected, as Seasonic’s PSU wattage calculator accidentally leaked lower numbers for Nvidia’s 50-series GPUs.

Swipe to scroll horizontally
Header Cell - Column 0 TGP (W)Next GenTGP (W) – Seasonic ‘leak’TGP (W) - @kopite7kimiTGP (W) – BenchLife.info
RTX 4090450RTX 5090500600550
RTX 4080320RTX 5080350400350
RTX 4070200RTX 5070220220-
RTX 4060115RTX 5060170--
RTX 4050Desktop N/ARTX 5050100--

While Seasonic’s initial information showed a modest TGP jump for Nvidia’s Blackwell GPUs, @kopite7kimi believes that Team Green will likely be more aggressive with its power requirements. However, BenchLife.info’s sources say that the 400- and 600-watt figures are the maximum heat dissipation capacity of the cards in question, not their actual TGP.

Furthermore, it also claimed that while there was a 400-watt cooling module being built for Nvidia earlier, development has since been canceled or suspended but work on the 600-watt heat dissipation module is still ongoing. Its contacts in the module factory for Nvidia have also said that the company is preparing five desktop RTX GPUs, so we’ll likely see an RTX 5050 appear on the horizon with this generation of Nvidia graphics cards.

We’re seeing rumors proliferate regarding Nvidia’s next-generation GPUs, especially as their purported launch date edges toward us. Nevertheless, all of these are just leaks and rumors, and no matter how accurate the sources have been in the past, the only time that we’ll know the exact specifications of Nvidia’s latest graphics cards and how they perform will be when we have them in our hands and are running them at full tilt to see what they can do.

Jowi Morales
Contributing Writer

Jowi Morales is a tech enthusiast with years of experience working in the industry. He’s been writing with several tech publications since 2021, where he’s been interested in tech hardware and consumer electronics.

TOPICS
  • thisisaname
    Whatever happened to do more with less :(
    Reply
  • jlake3
    Its contacts in the module factory for Nvidia have also said that the company is preparing five desktop RTX GPUs, so we’ll likely see an RTX 5050 appear on the horizon with this generation of Nvidia graphics cards.
    I’m reading this the opposite way? If they’re preparing 5 cards and we know there a 5090 and 5080, the only way they make it down to 5050 is if there’s no Ti cards (unlikely), they don’t count the Ti as a separate GPU (most people discuss them as separate products), or when they say “five GPUs” they mean “five desktop dies”.
    Reply
  • helper800
    thisisaname said:
    Whatever happened to do more with less :(
    Power limit 100% ---> 50%
    Reply
  • Flayed
    I suspect no one will mind as long as they are substantially more performant than their predecessors.
    Reply
  • JarredWaltonGPU
    Flayed said:
    I suspect no one will mind as long as they are substantially more performant than their predecessors.
    Yeah, this is the main key. If they're similar speed (or only slightly faster) and use less power, people would be far more unhappy than getting higher speeds with more power.
    Reply
  • valthuer
    @JarredWaltonGPU Sorry to bother you on this one, but, since you 're an expert on the subject, could you please explain to me what does this rumoured power draw translate into? For example, will my 1000W PSU be enough for 5090? Thank you in advance for your time.
    Reply
  • JarredWaltonGPU
    valthuer said:
    @JarredWaltonGPU Sorry to bother you on this one, but, since you 're an expert on the subject, could you please explain to me what does this rumoured power draw translate into? For example, will my 1000W PSU be enough for 5090? Thank you in advance for your time.
    So, my general opinion is that you don't want to run a PSU at more than ~70% load. Sure, it's technically possible to do so and should be fine, but I prefer a larger buffer between rated output and used output. So if the 5090 ends up at 600W, that's 60% of a 1000W PSU right there. Add in another 100~200 watts for the CPU and rest of the PC and you're at 70% to 80% load. If you have a high quality PSU, you should be fine. Anyone using an 80 Plus Bronze PSU, I'd start to be concerned.

    Your sig says you have a Zalman Wattera 1000W, which from what I can tell launched in 2015. PSUs do age over time, so if that's an up to date system components list in your sig and the 5090 does end up with a 600W TGP rating, I'd seriously consider buying a new 1200W or larger 80 Plus Platinum PSU.

    And it perhaps goes without saying, but I'm still very concerned about pulling 600W through a single 16-pin connector, no matter way Nvidia and PCI-SIG say! Those are relatively small wires and connectors and that's a lot of power. I'd be happier with dual 16-pin or even quad 8-pin. If the 5090 has a 550~600 watts TGP, I will be shocked if there aren't more melting connectors down the road.
    Reply
  • Notton
    More with less?
    Nvidia already does that with their RTX 4000 Ada Generation.
    It uses the same AD104 die as the RTX 4070, but at significantly less power. It's only slightly slower, but fits in either a LP or single-slot size.

    I'd expect similar results with an RTX 4000 Blackwell Generation, whenever it arrives.
    Reply
  • valthuer
    JarredWaltonGPU said:
    Your sig says you have a Zalman Wattera 1000W, which from what I can tell launched in 2015. PSUs do age over time, so if that's an up to date system components list in your sig and the 5090 does end up with a 600W TGP rating, I'd seriously consider buying a new 1200W or larger 80 Plus Platinum PSU.

    Thank you for taking the time to explain this to me! That's the exact piece of advice that i was hoping for! Unfortunately, i'm a layman, so i can't figure out what a 600W power draw translates into, in PSU terms. It's always very helpful to know the minimum and recommended suggestions.
    Reply
  • blargh4
    As long as perf/power is moving the right direction, I'm not particularly bothered. Lowering the power limit is easy enough.
    Reply