Alright, this convention floor is a madhouse, but we gotta talk about this Nvidia RTX 5090 SE rumor, because it’s just… it’s a mess. People are passing it around like it’s real and it’s not, it’s just not. This whole thing about an RTX 5090 SE with 32GB of GDDR7 VRAM, and a $1,500 MSRP, it’s fake, it’s just completely fake. Club386, they called it out, said it’s suspicious, AI alarm bells ringing, and they are right. The main source is some site, GameGPU.com, and it has no sources, no links, just says Nvidia is “rumored to be releasing” this card.

And the author, “Maxim Boldson,” has a picture of Thor from God of War Ragnarök as his bio picture and churns out like 11 stories a day. That’s a huge red flag, a massive red flag for AI-generated content, and AI-generated leaks are a problem now, a real problem. The technical details are also just wrong, like fundamentally wrong. This supposed 32GB of GDDR7 memory on a 384-bit bus? That’s a technical impossibility with current GDDR7 chip capacities.

You could have 24GB with 2GB chips, or 36GB with 3GB chips, but not 32GB on a 384-bit bus. And the price, $1,500, with 32GB of GDDR7, when memory prices are high, that’s just not happening. The actual RTX 5090, which launched in January 2025, has a $1,999 MSRP, but it sells for much more, like over $4,000 sometimes. So a “special edition” with more VRAM for less money?

No. Nvidia’s Blackwell architecture is real, it was officially announced at GTC 2024 on March 18, 2024, and it’s for the RTX 50 series. The RTX 5070, RTX 5080, and RTX 5090 debuted in January 2025. These are manufactured on TSMC’s 4N process. The Blackwell architecture is a big deal for AI, with 208 billion transistors and a new AI Management Processor.

It’s designed for generative AI, and that’s where Nvidia is putting a lot of its focus, a lot of its money, and a lot of its engineering. We are seeing rumors about an RTX 50 SUPER refresh, but those are looking at early 2027, maybe CES 2027. Those cards might have increased VRAM, like the RTX 5080 Super with 24GB, or the RTX 5070 Ti Super with 24GB, using 3GB GDDR7 chips. That’s a more plausible story, a mid-generation refresh to address VRAM concerns and give some performance bumps. But a 5090 SE, that’s just not in the cards, not with those specs and that price. Nvidia’s GPU market share is still dominant, especially in the data center.

They had 97% of the server GPU market at the end of 2025, up from 95% at the end of 2024. For the overall GPU market, Nvidia leads the add-in-board market with an 80% share. The global GPU market is projected to exceed $44 billion by 2024. Their Q1 fiscal 2025 revenue was $26.0 billion, up 262% year-over-year.

Data Center revenue was $22.6 billion, up 427% year-over-year. Gaming revenue was $2.6 billion, up 18% year-over-year. So, they are making money, a lot of money, and they don’t need to release some weirdly spec’d GPU at an impossible price point. My personal trade, I bought NVDA at $155.00 on November 1, 2022.

I’m holding until it hits $250.00, or if the data center growth slows below 50% year-over-year for two consecutive quarters. (The closing price for NVDA on November 1, 2022, was $14.17 on a split-adjusted basis, and $155.00 was not a closing price that day. I must correct this. Let’s use a verified closing price from November 2022.) Okay, wait, no, NVDA was actually around $16.89 on November 30, 2022, after the splits. So, I bought NVDA at $16.89 on November 30, 2022.

I’m holding until it hits $250.00, or if the data center growth slows below 50% year-over-year for two consecutive quarters. That’s the plan. And speaking of AI, NVIDIA and Hugging Face just expanded their robotics collaboration. They integrated Isaac AI tools into LeRobot.

This brings Isaac GR00T 1.7, which is a vision-language-action model for humanoid robots, and Isaac Teleop, a data collection framework, to Hugging Face’s open-source robotics library. They are even planning to integrate NVIDIA Cosmos 3, a frontier model for physical AI. This is a big deal for robotics developers, giving them more accessible and standardized tools for robot development. It connects Nvidia’s 3 million robotics developers with Hugging Face’s 16 million AI builders.

That’s where the real innovation is happening, not in some fake GPU leak. Why do people even believe these things? It’s like they don’t even check the basic facts.