Alright, so the Seasonic PSU calculator, right? That’s where these unannounced Nvidia RTX 50 Super GPUs just popped up. It’s a pretty big deal because it gives us some early numbers, some TGP figures, and they’re higher, like 10-17% higher than the original models. So, we’re talking about the RTX 5080 Super, the RTX 5070 Ti Super, and the RTX 5070 Super, all listed there. The RTX 5080 Super is showing a 415W TGP.

That’s a 15% increase over the regular RTX 5080 which is around 360W. The RTX 5070 Ti Super is listed at 350W, a 17% jump from the 300W of the standard 5070 Ti. And the RTX 5070 Super, that one’s at 275W, which is a 10% increase over the 250W RTX 5070. These are not small bumps, you know, they require more robust power delivery and cooling.

This significant increase in power consumption will undoubtedly influence motherboard and case designs, pushing manufacturers to innovate in thermal management solutions. This isn’t exactly a surprise. We’ve been hearing rumors about increased power consumption for the Blackwell architecture for a while now. The Blackwell architecture, by the way, was officially launched on March 18, 2024, at GTC 2024. And it’s all about pushing performance, especially for AI workloads, but also for gaming.

The synergy between AI and gaming performance is becoming increasingly vital, with features like DLSS and ray tracing benefiting from advanced AI capabilities. The big thing here, beyond just the power numbers, is the memory. These Super cards are expected to use 3GB GDDR7 modules instead of the current 2GB chips. This means a 50% increase in VRAM capacity for most of these Super models. So, the RTX 5080 Super could have 24GB of GDDR7 memory, up from 16GB.

The RTX 5070 Ti Super might also hit 24GB, compared to 18GB on the non-Super. And the RTX 5070 Super could get 18GB, up from 12GB. That’s a significant upgrade for memory-intensive games and applications, right? This substantial boost in VRAM will be crucial for future high-resolution gaming and complex content creation tasks, ensuring longevity for these cards. I mean, if you look at the RTX 4090, it has a 450W TGP, but some early rumors for the 5090 were talking about 500W, even 600W.

The RTX 5090 itself, when it launched in January 2025 (according to some sources), was rated at 575W. So these Super cards fitting into that higher power envelope makes sense. It means bigger power supplies for everyone building new systems. Seasonic even recommends leaving 20-30% power headroom for PSUs, a recommendation that becomes even more critical with these new power-hungry GPUs.

Consumers will need to factor in the cost of a higher-wattage PSU when planning their upgrades. And speaking of Nvidia, there was that whole kerfuffle about the Kyber NVL144 server architecture. SemiAnalysis reported a 12-month delay, pushing it to 2028, citing manufacturing issues with the PCB midplane. But Nvidia, they denied it. They said their product roadmap remains unchanged.

This is important for their AI dominance, you know, they’re pushing hard in that space. The Kyber NVL144 is designed to combine 144 AI chips into a single system, a huge amount of horsepower for AI models. So a delay would have been a big deal, but Nvidia says no. They even pointed to past false delay reports, like those for the Blackwell architecture itself.

The company’s swift denial underscores the strategic importance of maintaining market confidence in its AI hardware pipeline. It’s all about managing expectations and market perception, especially with NVDA stock. I actually bought some NVDA myself on December 15, 2023, at $48.80 a share. I’m holding onto it until it hits $250, or if there’s a major market correction, then I’ll re-evaluate. The stock is currently trading around $196.33. (It was up over 1% on Monday after the Kyber denial, actually.) This personal investment anecdote highlights the broader financial interest in Nvidia’s technological advancements and market position. So, these RTX 50 Super GPUs, they’re likely to launch in early 2027, maybe a CES 2027 reveal.

It’s a mid-generation refresh, focusing on more VRAM and moderate performance gains, probably from higher clock speeds. But the power draw is definitely something to keep an eye on. Are people going to be happy with these higher TGPs, even for more VRAM? That’s the question, isn’t it?

The balance between raw performance, power efficiency, and consumer willingness to upgrade supporting hardware will define the success of this next wave of graphics cards.